Writers of Pro Football Prospectus 2008

Most Recent FO Features

ManningPey98.jpg

» Game Preview: DEN-CIN

Maybe the Bengals are a paper tiger, but are they really that bad in prime time games? Is Peyton Manning struggling in Denver's new offense? We detail the Monday night clash.

07 Jul 2006

FO Mailbag

by Aaron Schatz

Today we go back, back, back into some very old questions in the Football Outsiders mailbag. We get a lot of e-mail, and there are a lot of comments on the discussion threads. It all simply means too many good questions that require well thought out answers. There gets to be a huge buildup at the bottom of my e-mail. With the book finished, I've been going through, trying to answer some of these old questions. A good number of them provide excellent examples of why we value certain teams and players higher than others.

The backlog is also another reason to remind folks of two things. First, the best way to get your question answered at this point is to use the contact form, not to ask a question in a discussion thread. Second, if it is a question not related to the DVOA stats, it is more likely to be answered if you send it to one of the other writers, not me.

Be aware that we reference plenty of our innovative FO stats here, not to mention their unfamiliar terminology, so if you are a recent addition to the readership you might want to read this first.

Let me start by answering some frequently asked questions about Pro Football Prospectus 2006 and the upcoming season. You can read more about the book on this page, and it will be out in late July. We wanted it to be earlier than last year, but that proved difficult, in particular because the NFL moved the draft back a week. The book is available from Amazon right now as a pre-order, and it will be in your local bookstore as well. I've received a lot of questions about overseas shipping, but that's not really our department -- it all depends on who you buy the book from.

Yes, there will again be a downloadable KUBIAK fantasy projections spreadsheet this year. We're taking some time to really improve it for 2006. Unlike last year, it will be customizable with your league's rules, and it includes team defense. We're also planning things about a little better, so you won't send a donation and manually have us send you the sheet -- instead, there will be a more structured system so that after you pay a fee, you can download as many updates as you want as depth charts change throughout the preseason. We know some people are already asking for the projections, so we're working on putting this out there as soon as possible. The book also includes the projections, of course, ranked for eight different league setups including the much-requested point-per-reception leagues.

On to the questions...

Alan: Judging by conventional statistics, LaMont Jordan's primary strength in 2005 was his receiving. While he posted a rather paltry 3.8 yards per carry and barely topped 1000 yards, he caught 70 passes for over 500 yards, which are high numbers for a halfback. However, in rushing DVOA, Jordan is ranked 17th while his receiving DVOA is 27th. What is the reason for this apparent conflict in statistics?

Aaron: Jordan's ranks aside, he basically came out as average in rushing (1.1% DVOA) and receiving (-0.7% DVOA). So the question is, why isn't his receiving DVOA higher?

First of all, Jordan caught 68% of intended passes. The average for RB is 72%. Remember, we're talking about screens and dumpoffs and a lot of open stuff here.

More importantly, look at the breakdown of Jordan's passes by down.

1st down: 27 catches in 37 passes, 8.6 yards per catch, 2.4% DVOA. Fine.
2nd down: 33 catches in 43 passes, 7.2 yards per catch, 15.2% DVOA. Good.
3rd/4th down: 10 catches in 23 passes, 9.3 yards per catch. -44.3% DVOA. Yikes.

The Raiders threw to Jordan 23 times on 3rd/4th down and he had a grand total of 3 first downs. Enjoy such passes as:

20 yards on third-and-21
15 yards on third-and-18
11 yards on third-and-12
10 yards on fourth-and-14
5 yards on third-and-6

Wow.

I actually wrote about something similar in Pro Football Prospectus 2006 regarding Chris Perry of Cincinnati, who is thought of as a great receiver but has a horrible receiving DVOA. He has this same issue with pointless third-down catches, as well as a high number of catches for lost yardage.

Andrew Jones: I am trying to find out the amount of dropped catches by the Oakland Raiders and how it compares to the rest of the NFL. I know there were quite a few, but were do I get the exact figures?

Aaron: I'll turn our second Raiders-related question over to Bill Moore, coordinator of the Football Outsiders game charting project.

Bill Moore: Ah, the dropped pass, the bane of every NFL fan. Is nothing more frustrating and possibly infuriating? Drops are a fairly subjective statistic. What Andrew might label a drop, I might consider an unreachable overthrow or an excellent defensive move by the cornerback. The NFL doesn't officially track dropped passes. The statistics behemoth STATS, Inc. does track dropped passes, and while they release very little of their proprietary database of football data, one stat they do make available on a player-by-player basis is drops for wide receivers. To the best of my knowledge, they do not provide the information in sortable tables. To get league-wide averages, one would have to look up each player individually and do the math.

Fortunately for FO readers, we began tracking a number of non-official football statistics including drops in the inaugural year of our charting project. A wide array of volunteers recorded each play of the first 16 weeks of the 2005 season. Each volunteer's interpretation of a drop is likely to be different, and therefore is unlikely to exactly match that of STATS; however, as you seen in the table below we are close.

Oakland Raiders Dropped Passes, 2005
Target FO STATS
18-R.Moss 3 2
20-J.Fargas 2 N/A
32-Z.Crockett 1 N/A
34-L.Jordan 6 N/A
49-J.Foschi 2 N/A
83-C.Anderson 1 N/A
84-J.Porter 6 7
85-D.Gabriel 3 2
86-Ran.Williams 3 3
87-A.Whitted 1 1
88-Z.Flemister 1 N/A
Grand Total 29

League wide, we estimate the drop rate to be about 5.7% of all passes, or 14% of all incompletes. The Raiders' receivers dropped 29 passes in the first 16 weeks. Among all pass plays, that is almost exactly on league average. Here are the teams with the most and fewest drops as a percentage of charted pass attempts. We've removed passes marked "Thrown Away," "Tipped at Line," or "Hit in Motion," and our count of drops includes passes that are juggled by a receiver and then intercepted.

Most Drops Fewest Drops
Team Drops Pct Team Drops Pct
PHI 49 8.8% CIN 16 3.3%
NO 39 8.0% SEA 17 3.9%
WAS 30 7.1% IND 19 4.0%
CHI 26 7.0% SF 14 4.1%
MIA 34 6.8% BUF 17 4.2%

Yes, that really does say "Seattle." Look for more drop data in the upcoming Pro Football Prospectus 2006.

David Brude: I'm just curious to understand how Tatum Bell can average over a yard more per carry and yet have a DVOA that is only half of Mike Anderson's? This is a general question I always have about teams that use two running backs. What kinds of situations allow the back with seemingly better stats to be rated lower? I mean, I can see where backs that run out the clock a lot will have a low YPA, but I don't think that is the case here ... or is it?

Are TD's worth that much vs. yards per carry to make that big of a difference? What kinds of situations is Tatum Bell in that keep his DVOA low with such a high yards per rush? Or is it just that most of his runs go for little yardage but he breaks off very long runs that keep his YPA total very high? When he's averaging a yard more than his closest competitor, something must be going on.

Aaron: First of all, here's a look at the stats of the three Denver running backs for 2005:

Player DPAR Rank DVOA Rank Runs Yards YPA TD FUM Suc% Rank
Mike Anderson 37.0 8 20.3% 6 239 1014 4.2 12 1 55% 2
Tatum Bell 16.4 13 7.6% 12 173 920 5.3 8 3 43% 29
Ron Dayne 9.0 -- 25.5% -- 53 270 5.1 1 0 53% --

Why is Anderson's DVOA so much higher than Bell's? The correct answer is definitely "most of his runs go for little yardage but he breaks off very long runs that keep his YPA total very high," as you can see from this graph of how often each back ran for a given amount of yardage:

You can see that Bell has more losses and more runs for no gain, but Anderson has more runs in every category from 3-4 yards to 9-10 yards. Bell had seven runs over 25 yards, and that explains his high yards per attempt. Anderson only had one.

Dave had a second question that he asked months and months ago. You might not think the 2004 Arizona Cardinals are that interesting, but this does a good job of showing why DVOA and the official NFL ratings can be so different.

Dave Brude: I happened to be looking at the team defense ratings for 2004 and was wondering how Arizona's non-schedule adjusted yards per rush allowed could rank dead last yet their non-adjusted DVOA for rush defense is slightly above average. What kinds of things are going on that allows them to jump that much? Did they face a lot of running QB's?

Aaron: Reasons why the Arizona Cardinals rush defense in 2004 scores so much higher in DVOA than in actual yards per carry allowed:

1) The Cardinals caused 12 fumbles on running plays, more than twice the league average. (7 recovered by Arizona, 5 by the offense)

2) The Cardinals run defense that year stiffened in the red zone. They allowed just 2.49 yards per carry in the red zone, 10th in the league.

3) The Cardinals run defense that year stiffened on third down. Only 45% of running plays against Arizona on third or fourth down converted for a new first down, sixth-best in the league.

Now, interestingly, in 2005, the Cardinals allowed just 4.2 yards per carry, middle of the league, but their DVOA hardly budged. Why?

1) They caused only 4 rushing fumbles.

2) They allowed 2.88 yards per carry in the red zone, 21st in the league.

3) They were still very good on third down, however, allowing success on 43% of plays.

Richie Wohlers: When you give the strength of schedule ratings for a team for 2005, you just average out all opponents' DVOA. Have you put much thought into this? Wouldn't it be possible that a team's schedule could look tough because they happened to play a hard division foe twice, and that really weights down their schedule?

In an extreme example, a team could play 5 opponents with DVOA's of: 76%, -5%, -3%, -1%, -4%. These average out to +12.6%, seemingly a tough schedule. When in reality, they are playing 4 opponents that are below average, and one really tough opponent. They would likely go 4-1 over that 5-game stretch. "You can only lose each game one time." (The same thing would go for calculating strength of schedule using W-L of opponents.) Or, in reality, does it just not matter, and there isn't that much of a difference between the best and worst teams?

Aaron: To answer this question, I went and looked at average opponent done two ways: mean, which is how we do things right now, and median, which will lessen the effect of extremes on both ends.

For all but three teams in 2005, the mean opponent DVOA and median opponent DVOA were within 5%. (I mean 5% DVOA, not five percent as an actual percent.) Here are the three exceptions and one team very close to 5%:

TEAM AVG MED AVG RK MED RK
JAC -4.3% -14.7% 25 32
CIN -1.3% -6.2% 18 28
PHI 4.9% 11.0% 5 4
DAL 4.8% 17.3% 6 1

Everyone pretty much knows about Jacksonville's season -- they played the Colts twice, plus the Seahawks, Steelers, Bengals, and Broncos. Baltimore had a DVOA of -5.2% (new upgraded DVOA) and every other opponent was below -10%, and that's why the median is so low.

The NFC East played the AFC East (good) and the NFC West (bad). Add Washington and New York, and Philly and Dallas end up with much higher median opponent DVOA than mean opponent DVOA.

I suppose I could run both numbers on the site. But since DVOA includes every single play of the season, the mean opponent DVOA version of schedule strength is a much better indicator of teams which will have a big difference between VOA and DVOA. Jacksonville is a good example. The Jags' DVOA (17.3%) isn't much lower than their VOA (20.0%) because in some games they get a big bonus for a super-hard opponent, and in some games they get a big penalty for a super-easy opponent, and it mostly cancels out.

(By the way, these strength of schedule numbers are slightly different from those on the 2005 team efficiency page because they are calculated from the new, improved DVOA v5.0 introduced in the book. We'll have those numbers all up on the site in the next few weeks.)

John Fessenden: Loyal FO reader here to ask you about defensive DVOA ratings. First I'll say that I'm a huge fan of this site and its brand of analysis, but I have one quibble with how defensive ratings, as I understand them, are done. Do you think its possible to interpret interceptions as “lucky� events for defenses, in so far as they are reliant on an the opposing quarterback putting the ball in an area to be intercepted?

I'll use a baseball analogy: a defense is to a pitcher in baseball, and a pitcher can have all the right peripherals (K:BB, GB/FB, etc) and still be undone by being unlucky in his BABIP (batting average on balls in play) simply because he has no control over where the batter hits the ball. Similarly a defense can also do all the right things as far as stuffing the running game, low completion %, good pass rush, etc, but in order to generate interceptions it is reliant on the opposing QB throwing the ball into an area that could possibly be intercepted. I certainly think that a defense can be more prone to getting interception with having a good pass rush, good coverage, etc, but should these areas be the ones measured instead of interceptions, because these are the areas which a defense can control? An interception requires luck simply because if the QB throws an errant ball, the defense is in no position to control whether it is interceptable or not.

I also noted that as DVOA stands now, defensive rankings are much more unpredictable from year to year, and I wonder if that would still be true if you treated an interception simply as another incomplete? As a fan of the Ravens my eyes told me that their defense last year was about as efficient as it was the year before, however they simply weren't able to intercept as many balls, hence the lower ranking. I'm not sure if skill or luck generated this lower ranking, but my intuition tells me that their “peripherals� would be pretty close the past two years. Sorry for the rambling query, but this has been on my mind for a while, and I was hoping you could clear this up.

Aaron: Interceptions do in fact correlate from year to year, from a defensive standpoint. Not as well as other things, but much better than, say, fumbles. The newer version of DVOA does control the defensive unpredictability by lowering the penalty for interceptions and treating different interceptions differently -- i.e., picking off a pass at the line is better than picking one off 40 yards downfield. But the interceptions have to be in there to make the system more accurate.

Having the same thought as you, I once tried to stick interceptions into the team projection system, thinking that teams with lots of interceptions would tend to regress in defense the next year, and teams with very few interceptions would improve. It doesn't work. (This is a part of the new team fantasy defense projections in PFP 2006, as teams will regress towards the mean in interceptions, but it doesn't help in projecting actual team defensive quality as judged by DVOA rather than fantasy points.)

People interested in these issues should read this article from two summers ago.

Posted by: Aaron Schatz on 07 Jul 2006

85 comments, Last at 25 May 2007, 12:38am by Quinton

Comments

1
by David (not verified) :: Fri, 07/07/2006 - 2:46pm

TATUM BELL IS FAR BETTER THAN MIKE ANDERSON ITS CALL YPC

2
by Pat (not verified) :: Fri, 07/07/2006 - 2:47pm

To answer this question, I went and looked at average opponent done two ways: mean, which is how we do things right now, and median, which will lessen the effect of extremes on both ends.

There was a fair discussion about this in a previous thread, and I swore at some point I was going to make an easy macro in Excel to work this out... but never did. :)

The question of "what's a hard or easy schedule" is actually pretty difficult, because it's not necessarily independent of the team. Take the listed example (76%, -5%, -3%, -1%, -4%). That looks like "one hard, four easy" - and averages out to "pretty hard".

An average team facing an average schedule should win 2.5 games in 5 games, right? An average team (0%) facing those teams would probably lose the first game, and win about half the other games (on average). We'll just say it's 2 wins total. But 2.0 games out of 2.5 is 0.5 fewer wins than average, so that is "pretty difficult."

Now imagine a good team facing that schedule: they'll still likely lose the first game (their first opponent is fantastic), but let's say they win 75% of their games against average opponents. Against 5 average opponents, you'd expect 3.75 wins. Against that schedule, you'd expect 3 wins - but that means that the schedule lost you 0.75 wins more than average. So this could be considered a worse schedule for a better team than an easier one.

Basically, there's no one number that can perfectly characterize a schedule. Mean tells you the average strength of your opponents. Mode would tell you your most common opponent strength. You could also calculate "wins lost/gained by an average team due to this schedule", which would give another measure of the strength of a schedule. You could also calculate "wins lost/gained by a team with my strength given this schedule", which tells you how your schedule affected your wins/losses.

The last number would explain why San Francisco won so many games last year while playing so poorly, incidentally. San Francisco's schedule last year was really freaking lopsided, since it played in the NFC West.

(I personally like the last number, because it complements 'estimated wins' to determine whether or not a team really is reaching, like Atlanta was, or just the recipients of a lucky schedule, like San Francisco was).

3
by Bob (not verified) :: Fri, 07/07/2006 - 3:02pm

When are the FO projections going to be released?

4
by tim (not verified) :: Fri, 07/07/2006 - 3:14pm

to 1: all i know is that mike anderson can play an entire game.

also: in regards to the dropped passes, are there variations in the average for receivers of different height?

5
by Peter (not verified) :: Fri, 07/07/2006 - 3:49pm

Interesting stuff here. I kind of like the expected wins gained/lost idea in #4... although couldn't you also do an average with, say, easiest and hardest game or two removed? Basically a mixture of median and mode. I mean, ultimately it just has to make sense to the people who look at it, I don't really know what suggestion I like better. It's kind of an "I know a hard schedule when I see it" situation.

This question is actually very similar to the Bell/Anderson discussion... on average Bell is better, though it's because he puts up outlier rush totals of 20 or 30+ yards. I've never really understood why people hate on boom/bust rushing so much, it seems great unless you really need short yardage, in which case you could always swap in a big guy. Shouldn't you just get your highest possible ypa guy in there for most of the carries?

6
by Peter (not verified) :: Fri, 07/07/2006 - 3:49pm

I meant #3 there.

7
by Andrew (not verified) :: Fri, 07/07/2006 - 4:01pm

David #1:

"TATUM BELL IS FAR BETTER THAN MIKE ANDERSON ITS CALL YPC"

David #1 is disqualified for invalid homer comments format, and will suffer a 4 thread suspension of comment privileges.

Correct formatting of rants such as David #1's would be:

TATUM BELL is clearly ranked too low because HE HAS A HIGHER YPC THAN MIKE ANDERSON. JUST RANKING PLAYERS BY YARDS DIVIDED BY PLAYS is way better than this DVOA CRAPOLA. THEE CHEEFS AND RAIDARS SUCK DONKEYS AND CANOT NEVER STOP THE BELL FROM GETING THE ENDZONE - BRONKOS RUL!

We now return you to your regularly scheduled thread comments.

8
by ABW (not verified) :: Fri, 07/07/2006 - 4:17pm

Re: 5

No, you shouldn't just put your highest YPA guy in there. It would be if football was just about gaining yards, but it's not. Check out the graph of the distribution of Bell's carries. That's telling you that if you run with Tatum Bell on first down, about half the time you are going to gain 2 yards or less(and about 8-9% of the time you are going to lose yards). That's not good. You simply can't put your offense is 2nd and 8(or worse) half the time - you will be passing against your opponent's nickel or dime package constantly. Whereas if you sent Mike Anderson out on first down, about 35% of the time you are ending up in 2nd and 8 or worse(and you are ending up in 2nd and 10 or worse about half as often as you were with Bell). That difference turns into drives going longer, putting points on the board, and keeping your defense off the field.

FWIW, Tatum Bell's overall success rate was 43%, or 29th out of 53, and Mike Anderson's success rate was 55%, or 2nd out of 53.

9
by Pat (not verified) :: Fri, 07/07/2006 - 4:56pm

It’s kind of an “I know a hard schedule when I see it� situation.

Do you? Really? Take a look at Philly's schedule in 2004. Most people would call that a freaking easy schedule, right? But take a look at Indianapolis's 2004 schedule as well. Easy or hard, compared to Philly's?

Indy's 2004 schedule sure looks harder than Philly's. The average agrees, too: Indy's is 0.0%, a perfectly average schedule, and Philly's is -4.4%, quite easy. But think a bit: the only team on Indianapolis's 2004 schedule that really should've given Indy a run for their money was New England - likewise, the only team on Philly's that should've given them a run for their money was Pittsburgh. Sure, the aggregate of the other teams was better than Philly's other teams, but they were still all much worse than Philly and Indy.

Now, New England's 2004 schedule was harder, as they faced both Indy and Pittsburgh, and Pittsburgh's 2004 schedule was harder as well. But it's tough to say that Philly's 2004 schedule was significantly easier than Indy's if you place it in context of how good the two teams were.

You can't really boil a team's schedule down to a number, though. Philly definitely played worse teams than Indy, so in that sense, they played an easier schedule. But in terms of the number of wins it should've cost them, it's probably pretty similar. Which description of "schedule strength" you want depends on the point you're trying to make.

10
by David (not verified) :: Fri, 07/07/2006 - 7:17pm

all in know is that if you given mikey anderson 200 carries and you give TATUM BELl 200 carries ...................

TATUM HAS MORE TOTAL YARDS

NUFF SAID

11
by MJK (not verified) :: Fri, 07/07/2006 - 7:21pm

Do strength of schedule calculations (in any form) calculated here take order into account? Beside the obvious case of a good team resting its starters (i.e. New England in Week 17 last year), I would imagine that playing a team right after they have a bye is tougher than they would be otherwise. Also, a team's ability ebbs and flows over the season. Would it make sense to factor in a game based on how the team was playing around that time? E.g. imagine Indy played New England some time. Rather than use New England's total DVOA over the year to judge if it was a "hard" or "easy" game in Indy's schedule, what if you just used some sort of weighted average of the games New England played shortly before and after the Indy game. This would also factor some first pass estimate of injuries a team was suffering into strength of schedule...

12
by Kibbles (not verified) :: Fri, 07/07/2006 - 8:27pm

Re #5: This question is actually very similar to the Bell/Anderson discussion… on average Bell is better, though it’s because he puts up outlier rush totals of 20 or 30+ yards. I’ve never really understood why people hate on boom/bust rushing so much, it seems great unless you really need short yardage, in which case you could always swap in a big guy. Shouldn’t you just get your highest possible ypa guy in there for most of the carries?

One of my favorite arguements is that if you could give me a guy who got 4 yards every single time he touched the ball, he would be the greatest RB in the history of the game. I mean, his ypc is only 4 yards per carry, which is actually very low, but every drive would look like this: 1st and 10, 2nd and 6, 3rd and 2, 1st and 10, 2nd and 6, 3rd and 2, 1st and 10, 2nd and 6, 3rd and 2, TD. Every single drive would end in a touchdown.

Now, let's replace this RB who is getting a "meager" 4 yards per carry with an RB who averages 20 yards per carry by running for 2 yards, 2 yards, 2 yards, 2 yards, and then 92 yards. The drives would look like this:
1st and 10, 2nd and 8, 3rd and 6, 4th and 4 (turnover on downs). Next drive: 1st and 10, TD.

So you see, RB #1 would average 4 yards per carry and score on every drive, while RB #2 would average 20 yards per carry and score on every OTHER drive. Which would YOU rather have? Heck, even if RB #2 always ran for some randomized value between 0 and 100 yards, there would be drives where he failed to score, while there would never be a drive that RB #1 failed to score. Even if you try some clever combination of the two, there will be drives where RB #2 and #1 together fail to score, but there will NEVER be a drive where RB #1 fails to score.

Consistancy is a wonderfully powerful thing in football.

13
by Kibbles (not verified) :: Fri, 07/07/2006 - 8:31pm

Re #10: all in know is that if you given mikey anderson 200 carries and you give TATUM BELl 200 carries ……………….

TATUM HAS MORE TOTAL YARDS

NUFF SAID

Read post #12. All I know is that if you give RB1 200 carries and RB2 200 carries... RB2 has more total yards. His team only scores half as frequently, but he'd have more yards.

Football is a game of TDs, not yards. You could gain 99 yards and it doesn't matter one little bit if you don't come away with any points to show for it. You could get outgained 900-15, but if the scoreboard reads 7-0 in your favor, you've still won. Consistantly good RBs result in stronger offenses, longer drives, more offensive plays, and more points than inconsistantly great RBs.

Heck, Tatum Bell averages more yards per carry that Terrell Davis did in 1998. Would you have started Tatum Bell over Terrell Davis?

14
by Pat (not verified) :: Fri, 07/07/2006 - 9:00pm

Do strength of schedule calculations (in any form) calculated here take order into account?

Nope. That would only be a retrospective schedule strength calculation, since you can't really predict how a team's going to evolve. Weighted DVOA by week's available in the archives - but I think you'd want to reevaluate the weights there since you're not trying to predict, you're trying to find the best measure of a team's strength at a given point in time. You'd want to do something like tweak the weights for past and future games to best correlate with a team's measured DVOA in the game they played that week.

15
by Richie (not verified) :: Fri, 07/07/2006 - 9:34pm

all in know is that if you given mikey anderson 200 carries and you give TATUM BELl 200 carries ……………….

TATUM HAS MORE TOTAL YARDS

NUFF SAID

Judging by your reasoning skills, spelling, grammar, punctuation and capitalization, it does not surprise me that that is all you know.

16
by mattman (not verified) :: Fri, 07/07/2006 - 10:50pm

Was David just disappointed that there was no Oddly Misspelled Hate Mail of the Week in this edition?

17
by Phill (not verified) :: Sat, 07/08/2006 - 5:27am

all in know is that if you given mikey anderson 200 carries and you give TATUM BELl 200 carries ……………….

TATUM HAS MORE TOTAL YARDS

NUFF SAID

The flip side is that to give Tatum Bell 200 carries you will have to have more plays than to give Mike Anderson 200 carries, and will have punted considerable more often. And probably scored less.

18
by Bright Blue Shorts (not verified) :: Sat, 07/08/2006 - 5:38am

Surely whether you put Bell or Anderson out there is about how you want to play football?

Basically Bell breaks big runs perhaps for TDs; Anderson is a steady grinder who will move the chains regularly slowly.

Think back to the Super Bowl ... Pittsburgh had 2 very big plays that scored them TDs and a scramble that took them to the 1-yd line. They win the game 21-10. On the flipside the 49ers won Super Bowl's moving the chains with a WCO.

BBS :)

19
by Becephalus (not verified) :: Sat, 07/08/2006 - 6:36am

Heard a great psuedo stat today and just thought you guys here would love it.

Do you know that 14 of 41 teams who made the playoffs two of three years (sorry cannot remember the time period) made it to their conference championship games? And 3 of them won the superbowl?

Well gee who would have thought...

Lets see 14/41 is roughly 35% and 3/41 is roughly 7%... so teams in this position are more or less exactly a random shot to make it into either of these situations... wow deep insight. So this indicator is just like throwing darts but takes longer and will develop my research skills?

I'll definitely incorporate that into my future wagers...

20
by Fnor (not verified) :: Sat, 07/08/2006 - 10:45am

Chicago's problem is, of course, Orton's bad mojo making his recievers' hands slippery.

21
by the peepshow (not verified) :: Sat, 07/08/2006 - 2:19pm

I knew Hasselbeck was putting glue on his balls. How else could they go from the most drops to one of the least? Maybe loosing slick joe from cleavland will hurt more than they think.

22
by David (not verified) :: Sat, 07/08/2006 - 4:11pm

can mikey anderson run 4 yards guaenteed every time I THINK NOT
IRRELEVANT

TATUM "hurricane man" BELL

IS BETTER
than mikey" moose" anderson\

nuff said

23
by Marko (not verified) :: Sat, 07/08/2006 - 4:47pm

Kibbles, when I read your post, I thought of two running backs who have personified RB #1 and RB #2. This obviously is oversimplified, but I think of Emmitt Smith as RB #1 and Barry Sanders as RB #2. Consistent with your point, RB #1 had much more team success. Of course, the players surrounding him (on both offense and defense) were far superior to those surrounding RB #2.

24
by tinselburn (not verified) :: Sat, 07/08/2006 - 5:25pm

re 13,

I still recall the Steelers losing to the texans fairly lopsidedly despite outgaining them in yards something like 435-55. my details are a bit fuzzy as I was slamming my head in a door by the end of it....

So Indeed. Good Point

25
by Pat (not verified) :: Sat, 07/08/2006 - 5:28pm

can mikey anderson run 4 yards guaenteed every time I THINK NOT

According to the article, he can do it a lot more than Tatum Bell can.

Besides, is a 40-yard run for a TD really twice as good as a 20-yard run for a TD? They're both TDs, and the only difference between them was where the team was initially lined up.

26
by tinselburn (not verified) :: Sat, 07/08/2006 - 5:30pm

aha, they lost 24-6 despite outgaining them 422 - 47. now let us never speak of it again. unless its to send porter into a hulk-like rage.

I still hate you tommy maddox.

27
by Mentos Fillapeedios (not verified) :: Sat, 07/08/2006 - 6:13pm

Houston made three plays all game-
long returns by Kenny Wright and Aaron Glenn (twice). One of the weirdest games of all time.

28
by tinselburn (not verified) :: Sat, 07/08/2006 - 6:36pm

yeah, the recap is linked to my name. now im all pissy about maddox getting a ring. couldnt they have just cut him before the superbowl? and you know lure jeff george out of retirement so he could say he got one?

29
by kleph (not verified) :: Sat, 07/08/2006 - 9:01pm

the beauty of all caps is that it makes people using bad logic so much easier to spot. that way you can avoid the childrens table and go right to the adult discussion.

30
by Michael Knght (not verified) :: Sat, 07/08/2006 - 11:09pm

The Bell or Anderson question is part of a bigger debate: big play backs or consistent backs. The baseball analogy has been made here before, but I've never seen an article that explored it in depth. As baseball statheads know, the high-average hitters tend to be overrated. Whereas the all or nothing sluggers tend to be underrated. Is Tatum Bell currently misunderstood like Dave Kingman or Rob Deer were in their respective sports? Do those occassional big gains make up for the usual negative plays?

31
by jebmak (not verified) :: Sat, 07/08/2006 - 11:32pm

#29 We have also learned something new from this thread, if you glance at a post and see that it ends with, "nuff said", you can also skip it.

32
by Anonymous (not verified) :: Sun, 07/09/2006 - 12:25am

All I know is that if you give Bell (5.32 YPA) 200 carries and Plummer (7.38 YPA) 200 passes, Plummer will have more yards than Bell.

'Nuff said.

33
by Duncan Tuckerton (not verified) :: Sun, 07/09/2006 - 1:48am

Most Drops Fewest Drops
Team Drops Pct Team Drops Pct
PHI 49 8.8% CIN 16 3.3%
NO 39 8.0% SEA 17 3.9%
WAS 30 7.1% IND 19 4.0%
CHI 26 7.0% SF 14 4.1%
MIA 34 6.8% BUF 17 4.2%

What are the Cowboys numbers?

34
by Pat (not verified) :: Sun, 07/09/2006 - 2:47am

The baseball analogy has been made here before, but I’ve never seen an article that explored it in depth.

That's kindof what DVOA does, though. You score points in football by continuing drives, and DVOA measures a team's ability to continue drives - because a positive success value means a team is more likely than before to continue a drive. So no, Bell's long runs don't make up for his short runs, because they don't happen enough (or against good enough teams, probably, either). Bell simply hasn't contributed as many points as Anderson has.

Now, if the question is "which would you rather have on your team: Tatum Bell, or Mike Anderson?" - the answer is "both, silly." Both Anderson and Bell can be used properly, presuming that Bell's 'big play ability' is innate, rather than just a result of the plays where he's given the ball. Trying to use Bell like Anderson will doom a team - trying to use Anderson like Bell will be a waste.

35
by Scott de B. (not verified) :: Sun, 07/09/2006 - 12:12pm

As baseball statheads know, the high-average hitters tend to be overrated. Whereas the all or nothing sluggers tend to be underrated. Is Tatum Bell currently misunderstood like Dave Kingman or Rob Deer were in their respective sports?

The difference is that a home run is four times as good as a single (roughly). To match that, the all-or-nothing back would have to rip off really big runs a high proportion of the time. (so 4 yards, 4 yards, 4 yards vs. 0 yards, 0 yards, 12 yards).

36
by Anonymous (not verified) :: Sun, 07/09/2006 - 1:25pm

Re #35: "(so 4 yards, 4 yards, 4 yards vs. 0 yards, 0 yards, 12 yards)."

Actually, Bell meets that criteria, thus the higher average. The relevant stats here are success percentage and his bust out percentage. Bell's success percentage is 43% to Anderson's 55%. It looks like Bell is about 5% more likely to bust out a run of 26+ yards. I'll take 12% (55-43) over 5%.

Really, what we need is some way of valuing a 25 yard run versus a 4 yard run. Obviously the 25 yarder not only accomplishes the goal of the current play, it also eliminates the need for two first downs. What's the value of that?

Of course, even if Bell is more likely to get the team down field, there's another aspect. It's usually better to gain eighty yards in twenty plays than one. Why? Because it brings you closer to the end of the game, thus reducing your opponent's chances to score. If this weren't true, there wouldn't be any point in running at all. Passing is much more effective at gaining yards.

37
by Terry (not verified) :: Sun, 07/09/2006 - 1:54pm

If this weren’t true, there wouldn’t be any point in running at all. Passing is much more effective at gaining yards.
I understand that you're likely saying this just to illustrate a point, but just in case, let's not get carried away here. Passing is the way to score in the new NFL-- but running is still an important element of any offense, if only to keep the defense from completely loading up against the passs.

38
by Pat (not verified) :: Sun, 07/09/2006 - 2:51pm

Actually, Bell meets that criteria, thus the higher average.

No - no he doesn't. The total integrated percentage of Bell's runs over 10 yards are still under 10%. In order to meet what #35 was saying, it'd have to be 33%. He's nowhere near.

Really, what we need is some way of valuing a 25 yard run versus a 4 yard run.

That's what DVOA does. Actually, that's what V+ does, if you read the "stats explained" page, and then VOA is just the ratio of that value over the average V+ per play, with DVOA being the defense-adjusted version.

Figuring out how much more a 25-yard run is worth over a 4-yard run is situation dependent, though. A 25 yard run on 4th and 26 is still a failure. A 4-yard run is worse, of course, but proportionally it's 'less bad' than, say, a 4-yard run on 3rd and 10 versus a 25-yard run on 3rd and 10.

39
by BillWallace (not verified) :: Sun, 07/09/2006 - 2:55pm

"Really, what we need is some way of valuing a 25 yard run versus a 4 yard run."

We have it, it's called DVOA, maybe you've heard of it?

40
by Daniel (not verified) :: Sun, 07/09/2006 - 2:58pm

I don't understand why a denver fan would care if Anderson started over Bell. You need both types of runners. I prefer the Anderson types, I believe over the long run you would win more games with a grind-it-out style. Bill Cowher has utilized both kinds of runners throughout his tenure in Pittsburgh. They went to the SB in '95 with Eric Pegram and Bam Morris. In the last SB Parker got the big TD, but Bettis sealed the game picking up 1st downs in the 4th quarter. Bill Parcells also likes to utilize a combo of big/small runners. Denver should be happy they have two guys who are versitile enough to help them win. Of course splitting time is a pain for fantasy football nuts who don't want to burn three draft picks on one team's RB corps.

41
by Pat (not verified) :: Sun, 07/09/2006 - 3:07pm

You need both types of runners.

Eh, I'd disagree - Bell would be a nice runner to have, but you don't need him. You'd primarily use Bell on 2nd and short, or 1st and 5 after a penalty, or possibly even 3rd down in field goal range.

But teams already have a strategy for those situations. It's called a pass.

A consistent short runner, however, is something that teams need.

42
by Duncan Tuckerton (not verified) :: Sun, 07/09/2006 - 4:03pm

We have it, it’s called DVOA, maybe you’ve heard of it?

I don't know about you, but I was laughing real hard here.

I always thought Emmitt Smith was better than Barry Sanders because Sanders had far more runs for negative yards. This is kind of similar to the Bell and Anderson comments.

43
by princeton73 (not verified) :: Sun, 07/09/2006 - 4:58pm

Interceptions do in fact correlate from year to year, from a defensive standpoint. Not as well as other things, but much better than, say, fumbles.

but the team that LEADS the league in interceptions (especially by an absurdly wide margin, like Cincy last year) DO tend to regress towards the mean, usually quite precipitously

TB 2002--31 int's
TB 2003--20

NE 2003-- 29
NE 2004--20

Minn 2003--28
Minn 2004--11 (!)

Car 2004--26
Car 2005--23

this is to be expected, of course--all of the teams except Minny were still above average the following year, but the difference in DVOA between a team with 31 picks and 20 picks is fairly substantial

44
by Mentos Fillapeedios (not verified) :: Sun, 07/09/2006 - 5:22pm

RE: "The Raiders threw to Jordan 23 times on 3rd/4th down and he had a grand total of 3 first downs. Enjoy such passes as:

20 yards on third-and-21
15 yards on third-and-18
11 yards on third-and-12
10 yards on fourth-and-14
5 yards on third-and-6

Wow.

I actually wrote about something similar in Pro Football Prospectus 2006 regarding Chris Perry of Cincinnati, who is thought of as a great receiver but has a horrible receiving DVOA. He has this same issue with pointless third-down catches, as well as a high number of catches for lost yardage."

(I wish I knew how to italicize quotes like other people do.)

What I don't understand here is this- I'm sure some of those plays were designed just to give the Raiders a few additional yards of field position. Norv Turner was probably looking for the easiest way to get a few more yards before sending in the punt team. When Jordan caught a pass for 20 yards on 3rd-and-21, sure it bumps up his average and winds up looking real nice in the boxscore.

Here, that play is considered bad because Jordan failed to convert a first down. At least, that's the message I'm receiving. Maybe I'm reading it wrong.

The thing is, I don't think those were all bad plays. Some of those were probably good plays, others ho-hum, some poor.

I guess I'm requesting more on the whole Jordan thing.

45
by Pat (not verified) :: Sun, 07/09/2006 - 6:23pm

The thing is, I don’t think those were all bad plays. Some of those were probably good plays, others ho-hum, some poor.

It's not that they're bad: they're just not fantastic plays. A 10-yard gain on 3rd and 14 is not the same as a 10-yard gain on 3rd and 10. It's still a failure.

Now, it's still likely better than the average gain on 3rd and 14, which is where the value over average part of VOA comes into play.

The worst part of Jordan's DVOA on 3rd down is probably the fact that he only caught 10 catches on 23 passes. That's awful. The other catches are probably just average.

46
by Anonymous (not verified) :: Mon, 07/10/2006 - 3:40am

Re: #38, In order to meet what #35 was saying, it’d have to be 33%.

Why 33%? While Bell gets more tackles for a loss or no gain than Anderson, it's nowhere near the 67% of the example. Bell is about 7% more likely than Anderson to get zero yards or a loss. He's about 5% more likely to break a run more than 10 yards. 5% is more than half of 7%.

Actually, that’s what V+ does

No, it isn't. V+ arbitrarily says that a 24 yard run is worth five times as much as a 4 yard run in a 1st and 10 situation. This is no better than David's claim that a 24 yard run is worth six times as much as a 4 yard run. Both are equally arbitrary.

You're essentially saying that because your stat says it, it must be true; this is no different than his claim that because his stat says it, it must be true. The fact that I tend to agree more with the V+ stat than the YPC stat is irrelevant.

What I'm saying is that to answer the question, we need to measure how much more a 24 yard run is worth than the 4 yard run that would have made it 2nd and 6. I suspect that if we did that, the V+ value would be closer to reality than is the strict YPC ratio. However, measuring is what would demonstrate that.

47
by Fnor (not verified) :: Mon, 07/10/2006 - 8:22am

This thread has gotten silly and confusing.

I will attempt to alleviate this by saying- Smith only looks better than Sanders in that respect because, by the time Sanders touched the ball, the defence was in the backfield. Then he had to go all Houdini on them. Smith had an excellent line for most of his career.

48
by Pat (not verified) :: Mon, 07/10/2006 - 10:49am

No, it isn’t. V+ arbitrarily says that a 24 yard run is worth five times as much as a 4 yard run in a 1st and 10 situation.

You do realize that DVOA was constructed to correlate to points, right? It wasn't an arbitrary definition of the point values.

49
by Stung (not verified) :: Mon, 07/10/2006 - 10:58am

Remember Emmitt's line...

Erik Williams
Larry Allen
Mark Stepnoski...

Now, to be fair, I don't have a clue who was blocking for Sanders, so I can't attest to their level of quality or lack thereof. That said, fewer losses is often an indicator of offensive line quality. I'm not saying Sanders is definitely better. I'm just suggesting that you imagine Barry Sanders running behind the Cowboys' line, with the Aikman-Irvin pass keeping the defenses honest. Now imagine Smith running for Detroit...

Given that the way the two look with their real teams is pretty even, I'd say that Sanders is much better, although the flash vs. consistency point still applies.

50
by Peter (not verified) :: Mon, 07/10/2006 - 11:33am

OK, I see when I take a weekend off everything goes crazy. How did you guys get along without me for so many years?

8: You're right, football isn't about gaining yards. It's not about sustaining drives either; it's about scoring points. My argument was not that a 1 yard gain is better than 4 yards; my argument is that on average, Tatum Bell will advance your team further down the field than Mike Anderson. A 40 yard run, no matter what, makes that entire DRIVE a success. Even if you're on your own 20, you've gotten to the point that even one or two successful plays means a field goal, or at the very least you're looking to pin the other team with a punt. Mike Anderson taking 10 plays to get those 40 yards is not considerably more valuable, and he's not especially likely to link together 10 carries like that either. It's not as if Mike Anderson gets 4 yards per carry every time like our mythical Robo-Runner (tm, Robo-Player Corp., if you liked our punters, you'll love our short-yardage runners). I'll get back to this in a second.

9: I appreciate the condescending tone, especially since you end up agreeing with me that schedule strength is a difficult to quantify, subjective idea. As 11 points out, things vary. Consider an example dear to my own heart, the Panthers of 2004. They came in as potential superbowl favorites (damn SI cover curse...). Green Bay beat them the first week of the season, when the team was at full strength. This win would not be weighted more heavily than any other that season, but they were the only team to play the Panthers at their best. The Panthers then went on to lose their top three runningbacks, their best player overall (Kris Jenkins), and the player who we now think is their best player (Steve Smith), along with the other 10 or so players on IR. When they went on the six-game losing streak in the middle, none of those teams faced a serious challenge. Eventually, Jake Delhomme calmed down, the offensive line started to gel, and Nick Goings proved he was bad, but not as bad as you might think, and the team ran out a couple of wins, making their team strength reach a third level, better than week 7, but not as good as week 1 or 2.

This discussion sort of reminds me of college football, where something crazy happens, like Baylor beating #4 or whatever Texas. If Texas, demoralized by that loss, goes on to lose another 2 games, they'll finish the season ranked 20th or so, and everyone would think they sucked. This does not mean that Baylor's win was not an incredible victory over a team that could easily have been top-5 ranked. At the time, Texas had the swagger and the talent of a top 5 team, but by the end of the season, the win seems worse than it was.

All I'm trying to suggest is that schedule strength is something of a subjective idea... in some situations, maybe that Texas team wasn't as good as we thought, and the Baylor game is evidence of that (meaning Baylor's schedule was less strong). In other situations, maybe the team WAS that good, suffered an unlucky loss, and as a result Baylor's game was extremely hard at the time, even if Texas didn't continue to play at that high a level.

Side note: you're right in 41, having a consistent, short-yardage runningback is essential, and a longer runner isn't... all I'm suggesting is that having Tatum Bell plus a good fullback for 3rd and 1 might be just as good or better than Mike Anderson.

12: This is a pretty useless example, and not as fun as Robo-Punter. I mean sure, Robo-Runner that just automatically gains a first down every 3 carries is superior... congratulations on proving that.

To return to real life, perhaps Tatum Bell is too extreme an example (gaining HUGE runs or nothing) for what I'm talking about. How about a theoretical speed back who has a distribution a little more contained... we'll use Mike Anderson as the power back, but break up Tatum's rushing so that he has far more 9-20 yard runs. So he still has more losses/no gains than Mike, but he pops off fairly long runs (making a man or two miss, I suppose) much more often. Is that runner, with a higher average, more valuable? He still kills drives fairly frequently, but he also sustains them a bit better than Tatum, and will occasionally pop off one that makes a drive-altering difference (20 yards or so). I mean we're discussing a theoretical continuum of distributed runs, but given we're talking about real life and not Robo-Runner, what kind of distribution would make you want the occasional deep run?

51
by Duncan Tuckerton (not verified) :: Mon, 07/10/2006 - 11:38am

Sanders did have Lomas Brown, Jeff Hartings, Bill Fralic, Mike Utley and Kevin Glover blocking for him. That was not all the same year, it was spread out. He had Cory Schlesinger and Tommy Vardell as his lead blockers the time when he went for over 2,000 yards in one season. The Cowboys might not even get a Hall of Famer out of Smith's line besides Larry Allen. The lines are equal I do indeed think.
The big difference was the Lions had bad quarterbacks and so the Cowboys could throw better and this helped Smith gain more yards more easy. Wait, I think I'm saying Sanders is better.
Well not really because the Lions opponents could then put eight guys in the box to stop Sanders.
So Smith was better.
The Cowboys also won more Super Bowls with Smith than the Lions won playoff games with Sanders.

52
by Stung (not verified) :: Mon, 07/10/2006 - 12:45pm

And Larry Izzo has won more super bowls than Dan Marino, so clearly is a better player...

Can these peurile 'Player A is on a better team so he's obviously better' arguments be made somewhere else?

53
by Pat (not verified) :: Mon, 07/10/2006 - 1:47pm

I appreciate the condescending tone, especially since you end up agreeing with me that schedule strength is a difficult to quantify, subjective idea.

That wasn't condescending. My point was just that since there's no a priori definition of 'schedule strength', you can't even say "I know it when I see it." First thing you need to do is define what you mean by schedule strength. So when you said "I know a hard schedule when I see it" - what I was saying was "you can't." Because you haven't defined what you mean by a hard schedule. Hard for that team? Hard for an average team? Hard for a bad team? All of these are equally valid definitions of difficulty of schedule.

All I’m trying to suggest is that schedule strength is something of a subjective idea… in some situations, maybe that Texas team wasn’t as good as we thought, and the Baylor game is evidence of that (meaning Baylor’s schedule was less strong). In other situations, maybe the team WAS that good, suffered an unlucky loss, and as a result Baylor’s game was extremely hard at the time, even if Texas didn’t continue to play at that high a level.

The problem with college football is that 'schedule strength' is evaluated usually solely on won/loss results. The computer rankings are forced to only use win/loss results (other than the wacko Billingsley formula which kinda sorta semi uses strength of victory). That's clearly not enough statistics or information.

As a simplistic assessment, if you just look at how strong Texas's wins were against other teams, you can determine whether or not that loss was a fluke loss. If you only use win-loss records, though, you're stuck.

The only reason that schedule strength is really subjective is that in the minds of the people who use the term, they don't define it well. If you define what you mean by the term, you can easily use it non-subjectively.

54
by Scott de B. (not verified) :: Mon, 07/10/2006 - 2:01pm

Why 33%? While Bell gets more tackles for a loss or no gain than Anderson, it’s nowhere near the 67% of the example.

I think my point was that YPC doesn't measure total value of plays, since the four-down system rewards consistency. So the boom-or-bust guy has to be sufficiently consistent in his booming to ensure that he continues as many drives as the consistent back.

Example 1: Back A gets 4 yards each carry. Back B gets 40 yards every nine carries, 0 yards the other eight. Back A is more valuable, even though Back B has more YPC.

Example 2: Back A gets 3 yards each carry. Back B gets 7 yards every other carry. In this case, the booms come close enough together that Back B is more valuable.

So, a true value of a back is how they contribute to converting first downs. That's success rate, more or less.

55
by ABW (not verified) :: Mon, 07/10/2006 - 2:23pm

Re: 50

My argument was not that a 1 yard gain is better than 4 yards; my argument is that on average, Tatum Bell will advance your team further down the field than Mike Anderson.

If you run with Tatum Bell you are going to end up running far fewer plays than you would with Anderson, because Bell will bust and you will be forced to punt. What's better, to have 10 plays with Anderson averaging 4 yds a play or 6 plays with Bell averaging 5 yards per carry and then have to punt? On average, running with Anderson, not Bell, is what is going to move the chains and get you down the field.

A 40 yard run, no matter what, makes that entire DRIVE a success.

Also not true. What if you throw an interception on the next play? What if you miss the FG, or get sacked? A 40 yard run is a very successful play, but one play is not a successful drive unless it scores a TD.

Mike Anderson taking 10 plays to get those 40 yards is not considerably more valuable, and he’s not especially likely to link together 10 carries like that either.

Going 40 yards in 10 plays is more valuable than going 40 yards in one play. For one thing, you can do the first much more consistently than you can do the second. For another, you burn clock and keep the opposing defense on the field.

You are seriously underestimating how frequently you need to break off long runs to make it worth it. Tatum Bell had a total of 10 runs last year longer than 20 yards. He had at least 25 runs(can't tell exactly from the graph, but greater than 15% of his 173 carries) that went for 0 or negative yardage. That's just not a good enough ratio to make it worth it.

Mike Anderson's DVOA was 20.3%. Tatum Bell's was 7.6%. That's telling us that even though Bell broke some long runs which were probably way above 20% DVOA, he had so many crappy runs where he lost yardage or only gained a couple of yards and were worth negative DVOA that it dragged his average contribution on a per-play basis way below Anderson's.

56
by Peter (not verified) :: Mon, 07/10/2006 - 4:17pm

55: Of course turnovers nullify any success. That's outside the realm of the rushing we're discussing; an interception could easily follow ten 4-yard gains as well. If anything the other point should be made, that rushing 10 times more dectuples your chance of fumbling and being injured.

"Going 40 yards in 10 plays is more valuable than going 40 yards in one play. For one thing, you can do the first much more consistently than you can do the second. For another, you burn clock and keep the opposing defense on the field."
The first thing means nothing. Of course you're more likely to get 40 yards if you take 10 plays to do it instead of 1. The 40 yards have the same value. If you give Tatum Bell 10 rushes, all things being equal, he will gain approximately 50 yards; if you give Mike Anderson 10 rushes, he will get 40 yards. The only argument you should be bringing up is that with a limited number of downs to work with (3), it's slightly more likely that a player with a smaller range of potential results (Anderson) will get the next set of downs. My point is that if, as is the case with Bell, a player tends to skip the next two sets of downs by making it 20, he's been rather useful as well.

The second "thing," keeping the opposing defense on the field, is actually a good point and another thing adding to the value of a pounding back. However, it's not as if Tatum Bell doesn't get 3-4 yard gains, or actually does only 3-and-out or rush 20 yards. I am skeptical as to how much difference, in the course of a game, Anderson really makes in terms of time the defense spends on the field.

53: I'm sorry I called you condescending, I usually hate it when people on message boards say something like that and people end up arguing about whether THAT's true for 10 posts.

I wasn't suggesting we discuss the "strength of schedule" typically used by computers... you're right, it's entirely based on W/L records, and thus has problems. I just meant in the way that laymen tend to look at a team's schedule, it depends on whether next to Baylor 17 - Texas 10, you put a little "#4" next to Texas for what they were at the time, or the little #20 for how they finished. Neither really tells the story.

In my opinion the skill of the team playing the schedule should not matter, particularly considering the example of the NFL. A hard schedule for the 49ers is probably a hard schedule for the Colts too, the difference in talent isn't even remotely close to, say, the Royals and Yankees, or Baylor and Texas.

All my example of the 2004 Panthers or the hypothetical Baylor squad are supposed to indicate is that you have to individually consider the circumstances of each game to really get a solid idea of a team's schedule. Any statistical measure will inevitably give a poor picture, whether the problem be strange circumstances, team talent fluctuations, luck, or sample size, and while I admit a human's opinion isn't going to necessarily be better, I prefer it. Hence, my comment that you can sometimes know a hard schedule when you see it, even if the metrics we might use seem dubious.

57
by Pat (not verified) :: Mon, 07/10/2006 - 5:33pm

I wasn’t suggesting we discuss the “strength of schedule� typically used by computers… you’re right, it’s entirely based on W/L records, and thus has problems.

Well, not really, actually. Mathematically, it's very sound - the idea is several hundred years old. :) The main limitation of the W-L based SoS (thus, the computer ranking system at all, since that's all they do) is that they assume that a team's true strength doesn't evolve over the year.

I just meant in the way that laymen tend to look at a team’s schedule, it depends on whether next to Baylor 17 - Texas 10, you put a little “#4″ next to Texas for what they were at the time, or the little #20 for how they finished. Neither really tells the story.

Definitely, although most of the human voters for the AP poll have said that they don't consider something like this. Which is good, because early-season rankings are ridiculously uncertain due to lack of information.

In my opinion the skill of the team playing the schedule should not matter, particularly considering the example of the NFL. A hard schedule for the 49ers is probably a hard schedule for the Colts too, the difference in talent isn’t even remotely close to, say, the Royals and Yankees, or Baylor and Texas.

Oh, no, there I would very much disagree. The NFL has much more disparity than baseball, although nowhere near as much as college football. You can see this pretty easily if you look at the distribution of team records, for instance. Even considering the limited 16-game schedule, the NFL has far too many 12+ win and 14+ win seasons to have the same true strength distribution of MLB. If you take a look at 16-game slices of MLB, you can see that they don't have nearly as many 12+ win stretches and 14+ win stretches as the NFL does. For the 12+ win set it's really clear.

As I mentioned in a different thread, in my mind this is very likely entirely caused by playing multiple games in a row. Since you can't start your best pitcher three games in a row, you naturally tend towards parity since your opponent can start their best pitcher against your worst pitcher, etc.

You can see this in the "estimated wins" statistic, which is an analog of the higher order Pythagenport wins/losses on Baseball Prospectus: Indy last year was playing at a skill level to win 84.4% of their games, whereas San Francisco was playing good enough to win 11.2% of their games. That's a huge difference. And if you take a look at San Francisco's schedule last year, even though it might look relatively average - even hard - if you calculate the "schedule wins exceeding normal" for San Francisco (just using the log5 method to be simple), it's something like 1.4 wins. Which means San Franciso had a very easy schedule - for them - last year. Which is why they went 4-12 rather than 2-14.

58
by Pat (not verified) :: Mon, 07/10/2006 - 5:41pm

If you give Tatum Bell 10 rushes, all things being equal, he will gain approximately 50 yards; if you give Mike Anderson 10 rushes, he will get 40 yards.

Actually, that's not quite true. If you give Tatum Bell 10 rushes 100 times, he'll get a distribution of yards, with a mean of 50 yards. The mode of that distribution will not be at 50 yards, since if you look at the histogram given in the article, his peak is at 1-2 yards.

Which means more often than any other result, when Bell rushes 10 times, he'll gain about 10-20 yards.

When Anderson rushes 10 times, more often than any other result, he'll gain 30-40 yards.

Bell's average will be 50 yards, but not the median nor the mode.

59
by Michael Knight (not verified) :: Mon, 07/10/2006 - 7:10pm

I'm not sure I'd rather have the multi-play drive covering 40 yards or the one and done run.

From "Debunking the Myth of Drive Momentum" in the 2005 Pro Football Prospectus:

"No matter how long the drive has been, there's about a 34% chance that it will stall before the next first down-put a differnt way, there's a 66% chance that an offense will get its next first down."

And...

"A drive is a laborious war of wills and attirtion as much for the offense as it is for the defense. Your job doesn't get any easier until you've put the ball in the end zone and retired to the bench..."

Also...

"These results underscore the importance of both big plays and field position."

The crux of Gerheim and Armstrong's article was there was no appreciable advantage gained from wearing down a foe with a long drive (the so-called "Drive Momentum). But it also implied that first downs aren't that valuable.

We must remember the ultiamte goal in football is not to gain first downs. It is to score points. Is this where a long run is worth more than we have given it credit for?

I think the baseball analogy is relevant psychologically too. In baseball, you have the "small ball" or "station to station" approach. In football, "moving the chains" seems a likely counterpart.

60
by Pat (not verified) :: Mon, 07/10/2006 - 8:12pm

I’m not sure I’d rather have the multi-play drive covering 40 yards or the one and done run.

Of course you'd prefer the 40-yard run right away. That's not the point. Everyone wants a 40-yard run. Heck, everyone wants an 80-yard run from scrimmage the first play.

I hate the question "which would you prefer...". It's not "which would you prefer." It's "which gives a better indication of future performance: 1 40-yard run, or 10 4-yard runs?"

Put a different way, which is a better indication of future performance: 1 150-yard game, or 5 100-yard games?

61
by ABW (not verified) :: Mon, 07/10/2006 - 8:41pm

Re: 56

55: Of course turnovers nullify any success. That’s outside the realm of the rushing we’re discussing; an interception could easily follow ten 4-yard gains as well.

So a statement like "A 40 yard run, no matter what, makes that entire DRIVE a success" would be incorrect then, wouldn't it?

If anything the other point should be made, that rushing 10 times more dectuples your chance of fumbling and being injured.

Mike Anderson does not take 10 times as many touches to get a given amount of yards as Tatum Bell. If you want to get 40 yards from Bell, you have to give him the ball a lot more than once, unless you are real, real lucky. Also, Tatum Bell fumbles more than Anderson. Not a lot more, but more.

The first thing means nothing. Of course you’re more likely to get 40 yards if you take 10 plays to do it instead of 1.

Consistency does not mean nothing. It means that you can do one thing over and over again, and if that thing is even a little bit good, you can just pile up little bits of good until you have done enough to win. Look at post #12 - does that not make you think that maybe being able to consistently get a few yards is a valuable skill? I'm not saying that being able to bust a 45 yard run isn't a valuable skill either, but Bell simply doesn't break those long runs often enough to counteract the fact that a lot of the time he busts(hence his DVOA being lower than Anderson's). Virtually all boom-and-bust runners don't. That's why these guys are almost always the secondary back.

The only argument you should be bringing up is that with a limited number of downs to work with (3), it’s slightly more likely that a player with a smaller range of potential results (Anderson) will get the next set of downs

This doesn't strike you as a major advantage, that Anderson is more likely to get you that next set of downs? And that difference in success rate(55% vs. 43%) tells us that Anderson is significantly more likely to get it.

My point is that if, as is the case with Bell, a player tends to skip the next two sets of downs by making it 20, he’s been rather useful as well.

Oh, if Bell runs for 20 yards, there's no doubt that's real good. But Bell doesn't "tend" to "skip the next two sets of downs". He "tends" to run for a gain of less than 2 yards a lot of the time, and then bust a long one that brings his YPA attempt up, which helps his team win less than a runner(Anderson) with a lower YPA but more consistency - hence the lower DVOA.

You aren't choosing between "10 4 yard runs" and "One 40 yard run". You're choosing to risk a loss or very short gain against a small chance of long run, or to give the ball to Anderson and get more consistent results. DVOA and the success rate is telling us that Anderson is a much better bet.

Re: 59

I'm not claiming the existence of some mythical drive momentum that leads to points. I'm claiming that dividing carries between your RBs based on YPA is a bad idea.

62
by Peter (not verified) :: Tue, 07/11/2006 - 11:17am

58: I'm not sure why you are using the mode instead of the mean. Obviously when I said 40 and 50 I meant based on their average... Tatum's might go 1, 0, -1, 20, 3, 9, or whatever, but eventually (10 is not a sufficient sample size, obviously) it will be about 50. More than any result, ten rushes will yield about 50 yards... Based on the probabilities, it will be more like one no gain, one loss, three 1-2 yard rushes, three 3-6, and one of 7+... the last is lost in the rounding someplace, but the point is that he doesn't rush his "mode" every time any more than he rushes his mean or median. He'll come out with approximately 50 yards, which is exactly what I said.

57: Once again, we agree... basing things on w/l doesn't account for the an important variable, team strength.

As for baseball and football, I suppose I shouldn't have made that particular comparison. There's no intensity to regular-season baseball, because there are so many games. The win distributions are less spread out, much like baseball, because the best teams aren't really motivated (and that manifests itself also, as you suggest, in playing worse players) and the game is rather random as it is.
In any case, I maintain that strength of schedule should remain independent of team talent... a schedule is either hard or not, and you're either good enough to play it or you aren't.

"Definitely, although most of the human voters for the AP poll have said that they don’t consider something like this. Which is good, because early-season rankings are ridiculously uncertain due to lack of information."
Oh how I wish that were true... Talk to Auburn fans about two seasons ago; poll inertia is an unfortunate reality.

60: I don't think that's the question at all... this isn't a straw man here, ABW literally said: "Going 40 yards in 10 plays is more valuable than going 40 yards in one play." I don't know how future performance is necessarily indicated either... we have enough total runs on both men to know that "future performance" on Tatum Bell is that he will break off long runs every once in a while, and will average more yards per touch than Anderson.

61: "So a statement like “A 40 yard run, no matter what, makes that entire DRIVE a success� would be incorrect then, wouldn’t it?"
I'm sorry, I shouldn't have included the 'no matter what.' If a war breaks out and the stadium is nuked, that drive will also not have worked out. Additionally, if the right guard flinches 4 times in a row, pushing the drive back 20 yards, the drive will also be in trouble. From the context of only considering the running game, a 40 yard play means that a considerable good has been achieved by the offense. They have moved field position to the point that they will either be able to kick a field goal, score the touchdown much more easily, or at the very least put the other team into poor field position.

"Mike Anderson does not take 10 times as many touches to get a given amount of yards as Tatum Bell. If you want to get 40 yards from Bell, you have to give him the ball a lot more than once, unless you are real, real lucky. Also, Tatum Bell fumbles more than Anderson. Not a lot more, but more."
I never claimed Mike Anderson takes 10 times as many touches to get as many yards. I said that a running back doing what you said, rushing 10 times for 4 yards a pop, is. You're right, to get 40 yards from Bell, you need to give him the ball an average of 8 times. To get 40 yards from Anderson, you need to give him the ball 10 times. So 1/4 more. And yes, Tatum fumbles more, but if we had hypothetical backs that had the same fumble rate, taking more carries to get the same distance will result in more fumbles.

"Consistency does not mean nothing. It means that you can do one thing over and over again, and if that thing is even a little bit good, you can just pile up little bits of good until you have done enough to win. Look at post #12 - does that not make you think that maybe being able to consistently get a few yards is a valuable skill?"
I'm not sure why you pointed back to robo-runner. Yes, mathematically speaking, having someone who is guaranteed to get a first down every single 3 runs means he'll score every drive... lovely. I understand that gaining lots of short gains eventually equals a big gain. My point is that if someone gets the big gains often enough, it is as useful, or more useful, than doing it in small chunks.

My problem with the argument that is going on is that we are valuing first downs for the sake of first downs. DVOA cares mostly about situation... getting 4 yards on 3rd and 4 is close to as good as 8 yards on 3rd and 4. Getting the next set of downs is not a guarantee that you will get the NEXT set of downs... field position is always of value, and getting more yards per attempt will help that.

"I’m not claiming the existence of some mythical drive momentum that leads to points. I’m claiming that dividing carries between your RBs based on YPA is a bad idea."
But that's WHY you're saying what you're saying. Your reasoning for why Anderson is better is that he gets more yards and sustains drives... if the slight improvement in drive momentum he brings is not significant to scoring points and winning games, as 59 suggests, then why not go with the man who, on average, will gain more yards when he runs?

63
by Joe Blow (not verified) :: Tue, 07/11/2006 - 11:58am

Hey Peter,

That is a very good post, although it almost made my head explode.

All of you fellows' findings aside, I wonder if the typical NFL head coach, if he had to choose one or the other, would take Bell or Anderson?

64
by Peter (not verified) :: Tue, 07/11/2006 - 2:50pm

If it made your head explode for being hideously long, I'm sorry about that, it's because I'm debating 4 people about 8 things. If it's because you disagree, fair enough, seems like everyone does, hah.

I would bet almost every head coach would prefer Anderson, for two reasons 1) they're conservative, traditional, and risk-averse, and prefer a pounding game (defense wins championships! Running is always best!). 2) They know what they're getting more, and it helps their decision-making to some extent. Playing the lottery sometimes sucks.

65
by Joe Blow (not verified) :: Tue, 07/11/2006 - 3:23pm

Peter,

It almost made my head explode because it was long and because you were "debating 4 people about 8 things."

I agreed with what you wrote.

66
by Pat (not verified) :: Tue, 07/11/2006 - 3:32pm

58: I’m not sure why you are using the mode instead of the mean.

The mode represents the most likely outcome. You said "all things being equal, he'll get 50 yards."

In fact, all things being equal, he'll get 10-20 yards far more often than any other outcome.

You’re right, to get 40 yards from Bell, you need to give him the ball an average of 8 times.

No, and now, I'll have to throw together a simulation to prove this.

Which I just did. Based on the data given here, randomly selecting a run from the distribution given in the article, it takes Bell on average about 9 carries (8.8) for Bell to reach 40 yards. It takes Anderson also about 9 carries on average (9.1) to reach 40 yards.

Bell does not have a symmetric run distribution, so to figure out the average number of runs after X carries, you cannot (and I stress cannot) just multiply "number of runs * yards per carry" for anything less than a ginormous number of runs. Likewise, to figure out the number of runs needed to get a certain amount of yardage, you cannot just divide yardage by yards per carry for anything less than a ginormous amount of yardage.

67
by Pat (not verified) :: Tue, 07/11/2006 - 3:52pm

In any case, I maintain that strength of schedule should remain independent of team talent… a schedule is either hard or not, and you’re either good enough to play it or you aren’t.

What's your definition of strength of schedule? You don't give it anywhere. That's the entire point of what I'm trying to say.

Is it just a measure of opponents faced? If so, that definition is independent of team talent - but you can't use that measure to say something like "the reason team X went 10-6 is because they faced an easy schedule" because the impact of a schedule on a team is dependent on team talent.

Most people think about schedule strength as being team-independent, but then they try to apply that statement to the team itself. That, you can't do.

68
by Peter (not verified) :: Tue, 07/11/2006 - 4:07pm

"In fact, all things being equal, he’ll get 10-20 yards far more often than any other outcome."
All things being equal meant in terms of defensive awareness (obviously if you run 10 times in a row they're more likely to stop you earlier), play calling, etc. I have no idea why you would assume that meant that the run will have the same result 10 times. Having 10 1-2 yard runs from Tatum Bell is actually terribly unlikely (.33^10), as you know.

Anderson doesn't have a symmetrical run distribution either... only Robo-Runner does. I'm not sure why you went to the trouble to prove mathematically that using random variation will cause slightly different results when speaking of a small sample size (10), going towards an arbitrary goal (40 yards). The issue with the simulation, and with my statement of 8 carries for 40 yards, is that most of the time when Tatum breaks off a big one in this little contest for 40 yds, he'll skip right past it (for instance, rushes of -1, 4, 0, 2, 65, or he gets to 39 and then breaks off 15).

You're right, I didn't do enough math. If it will make you happier, I'll amend my statement: to get to 1000 yards, Tatum Bell will need about 190 carries, and Anderson about 240. Yay.

69
by Peter (not verified) :: Tue, 07/11/2006 - 4:31pm

My definition of strength of schedule is a measure purely of opponents faced. If you play a lot of very good teams, you have a difficult schedule. If you play easy teams, you have an easy schedule. Simple enough. The same lineup of teams is hard or easy, independent of how good your own team is. If Baylor plays X, Y, and Z, and Texas plays X, Y, and Z, both played a hard schedule, even if it would be impossible for Baylor to win those games and Texas might be expected to win one or two.

I'm not sure I follow why I can't then say that a team that finished, say, 14-2, but played an extremely easy schedule, is most likely not as good as their record would suggest. A team that goes 10-6 while playing a number of extremely quality opponents might appear to me as good or better. Again, college football might be a helpful example... just because Northern Illinois goes 10-1 doesn't mean they're better than an 8-4 Auburn team.

Usually I can see the other person's argument, and the fact that I don't understand what you're saying at all suggests that I'm wrong and am not thinking about this properly. From what you've said, given my definition, I should look at a team that plays a group of weak opponents and finishes 10-6 but not be able to say that their relatively mediocre finish suggests they are not a particularly good team? I don't know why.

Pat, just curious, do you, like me, have a terrible job with nothing else to do all day? Because the two of us in the last week are posting like 10x everyone else combined.

70
by Pat (not verified) :: Tue, 07/11/2006 - 5:11pm

Anderson doesn’t have a symmetrical run distribution either… only Robo-Runner does.

You'd be surprised. Sample 10 runs from Anderson's distribution, and it's incredibly symmetric. Sample 10 runs from Bell's distribution and it's very asymmetric.

Bell's distribution is much less regular than Anderson's.

I’m not sure why you went to the trouble to prove mathematically that using random variation will cause slightly different results when speaking of a small sample size

Actually, I screwed up two of the bins (Anderson has a non-zero 26-40 bin, and Bell's numbers didn't add up to 100%). Add that in, and Anderson actually takes less runs (8.8) than Bell (9.1) to reach 40 yards.

It's not much effort (plus I wanted to make sure that I knew how to do it). It's just sampling multiple times from a predefined distribution.

And this isn't slightly different results (and it's not statistics limited, either - it's 10K runs). For large numbers of runs, Bell will win out on average. For smaller numbers of runs, Anderson will win out on average. You were saying that Bell would win out on average. He won't.

The thing is, it takes more than the carries in one game for Bell's distribution to start winning. Which is really, really bad for a team: they might win the game where he rips off multiple long runs, but they'll lose the next three when he doesn't.

I’m not sure I follow why I can’t then say that a team that finished, say, 14-2, but played an extremely easy schedule, is most likely not as good as their record would suggest.

You can: if that schedule definition is dependent upon the team.

Otherwise, you have to ask: if the schedule's easy, doesn't that mean that for any team that plays it, they should be worse than their record indicates? But that won't be true. Why? Because you can only lose 1 game. So what if San Francisco - the worst team in the league - played Indianapolis (near the best in the league), rather than, say, Philly (average). At most they can lose one game.

Take San Francisco's 2005 schedule. SF was worse than their record indicated because of their schedule. But had Seattle, for instance, played that schedule, they'd be better than their record indicated. Same schedule, two different teams, two different effects on the teams.

Pat, just curious, do you, like me, have a terrible job with nothing else to do all day?

There's an simulation that's taking an obscene amount of time to run for some weird reason, and I don't want to interrupt it. It's taken like 3 days and 3 attempts to run it. This is the last day, though - after this I'll run it elsewhere and leave it running.

Plus this doesn't really take much time anyway.

71
by Earl (not verified) :: Wed, 07/12/2006 - 2:31am

I think you guys are failing to take into account that one other major issue is that Shanahan really doubted Bell's stamina. The biggest knock on Bell was not his inconsistency, it was the fact that he was only good for about 15 touches a game, where as Anderson was good for 25-30. Anderson could run the whole game, but Bell would get his bust out after 5 or 6 touches, and maybe another one in the 10-15 range, but that was all that he could be counted on.

That said, Shanahan still believes Bell can improve upon that, but he makes it pretty clear that Bell has not yet made it to that level where he can be an every down back, and until he can be, he's likely doomed to be the very important #2 in a two headed system.

72
by Crushinator (not verified) :: Wed, 07/12/2006 - 4:19am

Tatum Bell vs Mike Anderson -

All this talk here is of the more consistant back. Part of why DVOA says Anderson is a signifigantly better back is because big runs are not something predictable. In the overall scheme, they're outliers.

Some earlier said 3 yards a carry vs 7 yards every other carry, but even 7 yards every other carry is relatively consistant. The question sort of becomes in regards to Tatum Bell - you know he's going to break that one or two long ones - but does that even out for being relatively average? Is it better to have a good back who is more reliable in moving the chains, or your homerun hitter who might bust that long score?

73
by Peter (not verified) :: Wed, 07/12/2006 - 12:21pm

"The thing is, it takes more than the carries in one game for Bell’s distribution to start winning."

This is the point I thought might eventually come out (but I certainly wasn't going to do the math to prove it). In my opinion, that is inconsistent enough that his value is substantially reduced, enough that Anderson is preferable. I was asking in an earlier post at what point the inconsistency becomes worth it, but I think almost everyone would agree that if it's not even once a game, it's not good enough. That was why I also tried to sort of shift off to a theoretical runner that had one or two fewer huge runs, and a few more semi-long runs (10-20). Y'know what I really would love? To see a graph like that for every RB in the league. I couldn't tell you what the distributions would look like for most RBs (except, say, Bettis, or Westbrook) but I'm extremely curious.

"Take San Francisco’s 2005 schedule. SF was worse than their record indicated because of their schedule. But had Seattle, for instance, played that schedule, they’d be better than their record indicated. Same schedule, two different teams, two different effects on the teams."

Again, I'm not completely following. I guess we're assuming here that Seattle cleans up against this crappy schedule, and you're saying I would look at them and say they're worse than their record indicates because they played a crappy schedule, even though they're pretty good. I don't think I would, though... if anything, I would say that they have proven themselves to be at least a decent team (better than the teams on their schedule) but that they hadn't proven whether they were a very good team or not, because they hadn't played elite competition.

Strength of schedule can only tell you how good the team performed against a given spread of teams, it does not necessarily prove anything. You can point at the schedule and say it was hard or easy, so perhaps the team that played it is better or worse than their record indicates, but not definitively so. I still don't know if what I'm saying is somehow logically inconsistent to you.

Your key point as I undestand it is that to a bad team, a loss to a mediocre team is the same as to a good team, because they were at a disadvantage in both games. To a good team, a loss to the mediocre team is a disappointment, and a loss to a good team is understandable since they were roughly equal. I don't know what impact that is supposed to have on skepticism or dismissal of a team based on schedule strength.

71: Yes, that's most likely true... using real-life examples like this has a lot of flaws. I'm willing to bet that not only is stamina and situation a concern, but playcalling also isn't fairly distributed. Perhaps there are plays Bell runs that Anderson might run even better, but isn't given the chance to because of Bell's supposed advantages (or vice versa). That's why in an earlier post (I believe Pat) put up the qualifier "assuming both players are being utilized optimally."

72: Yes, that is pretty much our discussion. Do you have an answer?

74
by Pat (not verified) :: Wed, 07/12/2006 - 1:07pm

Again, I’m not completely following. I guess we’re assuming here that Seattle cleans up against this crappy schedule

No. Look at that schedule for San Francisco - it's not a crappy schedule at all! They played 6 eventual playoff teams (and Seattle twice). Think about that - nearly half the teams they played ended up in the playoffs. That's way above average difficulty.

Seattle wouldn't've cleaned up. Seattle would've ended up likely 12-4, maybe 11-5. But they would've been better than their record indicated, even though San Francisco (playing the same schedule) would be worse than their record indicated.

Why? Because it doesn't matter that San Francisco played 7 games against playoff teams. They likely would've lost those games against Philly, Minnesota, Dallas, Oakland, Baltimore, and Miami. Trade the 6 playoff teams for those guys, and their schedule gets much easier, but they won't gain much more wins. The main thing for San Francisco is the fact that they played St. Louis twice and Houston.

75
by Peter (not verified) :: Wed, 07/12/2006 - 2:17pm

Sorry, I didn't look at it... I assumed since the 49ers got more wins than they were expected to, their schedule was easy. I guess that's your point, though; a few beatable teams made the overall record look better than they were as a team, while they got SLAUGHTERED against the good squads. A team that is slightly better than the 49ers might get the same record against the same schedule, but play much closer in the losing efforts, and thus be better than performance against schedule indicates. Got it.

I guess that's a trap to look out for when looking at a team's performance against schedule... you're right, I can't say (with my team-independent definition) that they did well against a schedule unless I consider margin of victory (or to be more accurate, something like DVOA).

76
by cjfarls (not verified) :: Thu, 07/13/2006 - 1:11pm

re: #50
To return to real life, perhaps Tatum Bell is too extreme an example (gaining HUGE runs or nothing) for what I’m talking about. How about a theoretical speed back who has a distribution a little more contained… we’ll use Mike Anderson as the power back, but break up Tatum’s rushing so that he has far more 9-20 yard runs. So he still has more losses/no gains than Mike, but he pops off fairly long runs (making a man or two miss, I suppose) much more often. Is that runner, with a higher average, more valuable? He still kills drives fairly frequently, but he also sustains them a bit better than Tatum, and will occasionally pop off one that makes a drive-altering difference (20 yards or so). I mean we’re discussing a theoretical continuum of distributed runs, but given we’re talking about real life and not Robo-Runner, what kind of distribution would make you want the occasional deep run?

This actually describes Clinton Portis pretty well, who type-wise I think was exactly what Shannahan looks for... but these folks are hard to come by, and in a pinch, you can use a 2-headed monster like Anderson/Bell. I think this is what Shanny hopes Bell can develop into, and in the meantime will use Anderson/Dayne as the stopgap.

Alternatively you just get RoboRunner (aka Terrell Davis, pre-injury) to grab 4 yards EVERYTIME.

77
by Peter (not verified) :: Thu, 07/13/2006 - 2:30pm

Yeah... like I said, I would really love to have a distribution of all RBs in the league, to get a feel for styles in a way that's impossible to do ordinarily.

78
by JMM (not verified) :: Sun, 07/16/2006 - 1:14pm

I looked at the differences in the distribution of Bell and Anderson run history, and how they wold differ in a 25 carry game. The difference is Anderson replaces one run for a loss and two 1 to 2 yard runs with three, 3 to 4 yard runs. Bell extends one Anderson 9 to 10 yard run with a run longer than 21 yards, and maybe longer than 40 yards.

Which is the better distribution? Hard to say without knowing what the defense has to sacrifice in pass protection to play one vs the other. Also the timing of the plays also matters. 3 losses on 3rd and short or at the goal line or a 40 yard run from the 40 or from your own one... very different as we all know.

79
by Jake S. (not verified) :: Tue, 07/18/2006 - 5:28am

The argument that mean schedule strength is less indicative then median is kind of baffling. While it's true that the most likely single result of an average team playing 1 amazing team and 4 below average teams is 4 wins and 1 loss, the same is true if an average team played 1 above average team and 4 below average teams. The reason the mean SOS is more indicative is that team perfomance varies, bad teams sometimes beat good times and more of the time beat average teams, etc. It's a similar concept to expected value. Nobody argues that the best teams are likely to will all 16 regular season games. There's a probability they'll win each game and the sum of those probabilties is the expected number of wins.

80
by Pat (not verified) :: Tue, 07/18/2006 - 11:16am

While it’s true that the most likely single result of an average team playing 1 amazing team and 4 below average teams is 4 wins and 1 loss, the same is true if an average team played 1 above average team and 4 below average teams.

That's not true. That depends on the win function. Besides, quantizing the mode is kinda silly - you're really looking for the mean of the distribution of many repeated seasons playing the same schedule.

The reason the mean SOS is more indicative is that team perfomance varies, bad teams sometimes beat good times and more of the time beat average teams, etc.

Uh, yeah, which is why you need a function which tells you how often a bad team beats a good team. There's probably a different one for football using DVOA or estimated winning percentage, but no one's worked one out yet, so hey, log5 it is.

There’s a probability they’ll win each game and the sum of those probabilties is the expected number of wins.

Exactly. But the mean SOS is not going to track that, because it allows one game to weight the distribution as much as all the others combined. This is the equivalent of giving a team 1.5 average wins for playing the 49ers, or -0.5 average wins for playing Indianapolis.

81
by Rich Conley (not verified) :: Wed, 07/19/2006 - 5:48pm

78.

The issue with that whole argument is that on average, Mike Anderson gets more carries per game, because less drives stall.

If you want to figure it out, you have to use a Per Drive metric.

IE run like 10K simulated drives using their distributions for every play.

Tatum bell would have a higher YPC, but Mike Anderson would have more runs, more total yards, more first downs, more touchdowns, and less fumbles.

82
by Peter (not verified) :: Thu, 07/20/2006 - 10:39am

Well, no doubt he'd have fewer fumbles, since he has wonderful ball security compared to Bell. Given his short gains, he'd undoubtedly have more first downs, just because he has more opportunity to get them.
I'd question whether he'd necessarily get more yards or touchdowns. I also question whether this simulation would prove anything, because we know Bell and Anderson shifted off many times on a drive, rather than alternating drives or games.

I also tried, several times, to discuss whether someone with a more modest distribution (fewer extremely long runs, more sort-of long runs) would be preferable. Hence wishing I had a breakdown of all the RBs in the NFL.

83
by Rich Conley (not verified) :: Thu, 07/20/2006 - 12:16pm

RE 82

He would most certainly have more yards, and touchdowns.

Thats what success rate measures. Bell would fail much more often than Anderson. Every one of those failures is essentially a stalled drive.

i'm going to try to run the simulation tonight, and I'll post results.

84
by JMM (not verified) :: Fri, 07/21/2006 - 11:17pm

81 & 83

Looking forward to your simualtion. The purpose of 78 was to incorporate the distribution data into the discussion. It seems to me it is more valuable than either the median or the mode and to point out that the distributions were not that far apart.

85
by Quinton (not verified) :: Fri, 05/25/2007 - 12:38am

You described the following catches by RB Jordon as "pointless third down catches":
20 yards on third-and-21, 15 yards on third-and-18, 11 yards on third-and-12,
10 yards on fourth-and-14, 5 yards on third-and-6

But I think you know better. 5 catches for 61 yards holds tremendous value even if they don't produce a single first down. Those significant chunks of yards could have been just enough to get in FG range or ensure a successful FG. Those 61 yards could be the difference between winning and losing in a close game where the importance of field position is magnified. The other team may have fallen one yard short of the end zone to lose the game because of Jordon's last reception before the punt. The 10 yards Jordon gains before the punt (instead of, say, no yards) may become the 10 yards of field position his team needed to score on the ensuing possession. You have thrown out the field position factor in favor of the all-or-nothing first down measure of success. It's the same as saying, well, since they were one yard short of the first down and they have to punt, they could have taken a sack for a 10-yard loss just the same. Jordon's gains added no value in your OPINION because no first down was achieved. I think you know better. Every yard counts, but some count a little more than others.