Writers of Pro Football Prospectus 2008

09 Sep 2008

2008 Quick Reads: Week 1

Here's the first Quick Reads featuring YAR and EYds. (That's right, the folks who didn't like the name "Equivalent Yards (EqYds)" won -- the name will be changed on all the sheets to "Effective Yards" in the next few days.) Sorry this is so late, we're still getting used to our ESPN filing schedule. As a bonus, it includes both Monday Night games.

Posted by: Aaron Schatz on 09 Sep 2008

84 comments, Last at 12 Sep 2008, 5:32pm by Scott C.

Comments

1
by Lou (not verified) :: Tue, 09/09/2008 - 8:59pm

link redirect to this page. follow link in my name

2
by ZSGhost (not verified) :: Tue, 09/09/2008 - 9:02pm

Are you guys using a different system? I'm assuming I should divide the DPAR numbers you give to QBs by ten to get the old number.

3
by Alex51 (not verified) :: Tue, 09/09/2008 - 9:11pm

Are you guys using a different system? I’m assuming I should divide the DPAR numbers you give to QBs by ten to get the old number.

Yes, they are using a different system. There is no DPAR anymore, just DYAR (Defense Adjusted Yards above Replacement). Divide DYAR by about 14.3 to get DPAR. As Pat pointed out in an earlier thread when they introduced DYAR, the conversion should be about 100 yards/7 points, or ~14.3 DYAR/DPAR.

4
by Gerry (not verified) :: Tue, 09/09/2008 - 9:25pm

Are the EYards for Peyton correct there?

5
by Temo (not verified) :: Tue, 09/09/2008 - 9:28pm

4. He had tons of attempts, so he has a high effective yards. His DYAR will be lower because he was not "that good" above replacement.

6
by Gerry (not verified) :: Tue, 09/09/2008 - 9:30pm

"He had tons of attempts, so he has a high effective yards"

If the math works, the math works. But I would have thought the large number of attempts, mostly ineffective, would cause his EYards to be lower than his passing yards. And given that isn't the case, the comment is really disorienting.

7
by Gerry (not verified) :: Tue, 09/09/2008 - 9:36pm

To give details to what I was saying:

“Need an even quicker and easier way to see how good a player was? Use another stat we call Effective Yards (EYds). Effective Yards takes the player’s performance, adjusted for situation and opponent, and puts it on the exact same scale as standard yardage. If a player has more Effective Yards than Yards, he was better than standard stats make him look. If he has fewer Effective Yards, he was worse than standard stats make him look.”

OK, so if EYards>Yards, player was better than standard stats made him look. Check.

“Although Manning threw for 257 yards, it took him 49 attempts, much of his yardage was gained when he dumped the ball underneath coverage for short, ineffectual gains”

OK, that reads to me like Manning was worse than standard stats made him look.

But his EYards were 40% higher than his yards, which isn’t just better, but much better.

I am not finding this intuitive, at all.

8
by Kaveman (not verified) :: Tue, 09/09/2008 - 9:41pm

Hmm. Cutler had 0.0 rushing DYAR? He ran twice. Once for 11 yards and a first down, and once for a single yard. That equates to 0.0 DYAR? (I assume the three kneeldowns at the end of the game aren't counted.)

9
by Temo (not verified) :: Tue, 09/09/2008 - 9:41pm

7. You're right in this respect: Why is (almost) everyone's Eyds greater than regular yards? Are we to believe that other than Drew Brees, Brett Favre, and Jeff Garcia, every QB from 1-25 was better than his regular yardage would indicate?

However, Gerry, if the answer to the above is "yes", then I don't think Manning's total is wrong.

10
by Kaveman (not verified) :: Tue, 09/09/2008 - 9:44pm

Favre has 2 rushing DYAR and he ran once for 3 yards and once for no gain. Which suggests that something is missing for Cutler's DYAR.

11
by Mike (not verified) :: Tue, 09/09/2008 - 9:51pm

WHat was Brian Moorman's DYAR?

12
by Temo (not verified) :: Tue, 09/09/2008 - 9:55pm

Actually, forget what I said above. It makes sense, I suppose, since defence adjustments aren't in. When they are, I'm sure more QBs will show up with less Effective yards than real yards (while others will end up with even more effective yards).

13
by Lou (not verified) :: Tue, 09/09/2008 - 9:57pm

Favre has fewer EYds than actual yards, but still managed more DYAR than guys with a similar number of attempts who had more EYds than real yards. Does that make sense?

in fact their were guys with both more and fewer attempts than favre that had "good" games (EYd>yards) and still had fewer DYAR.

I'm with Gerry. Theres nothing intuitive about this.

14
by The McNabb Bowl Game Anomaly (aka SJM) (not verified) :: Tue, 09/09/2008 - 10:23pm

How does Brees have fewer EYds than actual yards? Didn't he have a good performance?

15
by Crabbie (not verified) :: Tue, 09/09/2008 - 10:42pm

All clicking this link did was remind me of how much more I liked DPAR than DYAR, which still makes little to no intuitive sense to me whatsoever. Well, I guess it means that I'll have one less thing to read each week, which is probably a good thing.

16
by Scipio202 (not verified) :: Tue, 09/09/2008 - 10:44pm

I think the EYds vs traditional yards comparison only makes sense if the player didn't have any touchdowns. Even unadjusted for context we know TDs are worth something - a WR with 70 yrds and a TD seems more productive than a WR with 100 yrds and 0 TD (and similarly for RBs and QBs). When players have a touchdown I think there needs to be some "raw"/"context free" TDs-to-yards conversion to recover the claim "EYds > Yards means he was better than he seemed" - which would now be "EYds > (Yards + TDsAsYards) means he was better than he seemed".

Say 1 TD is given a unadjusted "yard value" of 50. Then suppsose a player has 100 yrds, 1 TD. Then he'll be "better than standard stats" if EYds > 150, and "worse than standard stats" if EYds < 150.

17
by DrObviousSo (not verified) :: Tue, 09/09/2008 - 11:00pm

Yeah, how can Brees have the #6 DYAR, but have EYar < Yar? Is EYar/Attempt a useful metric?

So confused.
Captch: doofus

18
by Scipio202 (not verified) :: Tue, 09/09/2008 - 11:02pm

Though, looking at the chart again I'm not sure the EYds vs "regular yards" comparison works even if the QB has zero TDs and zero picks (this week that's Flacco, Kolb, Orton, Brady, Croyle, Bulger). Among those QBs (Eyds - yards) lines up with the ordering based on DYAR - except for Brodie Croyle. He's ranked behind Orton and Brady based on DYAR, but his (Eyds - yards) = 23, while for those two (Eyds - yards) = 18. Looking at (EYds/Yards) is even more out of order compared to DYAR - it correctly places Flacco and Kolb at the top and Bulger at the bottom, but the ratio puts Orton/Brady/Croyle in reverse order. And this isn't being driven by efficiency vs totals, since completion percent order them the same as DYAR (except for Flacco).

19
by Temo (not verified) :: Tue, 09/09/2008 - 11:07pm

Ok, so here's the description for True Yards (which was renamed Effective Yards) from the FO site:

True Yards translate DVOA into a yards per attempt figure. This provides an easy comparison: in general, players with more True Yards than standard yards played better than standard stats would otherwise indicate, while players with fewer True Yards than standard yards played worse than standard stats would otherwise indicate. True Yards are not the best way to measure total value because they are more dependent on usage than DYAR.

This makes more sense. Do a quick calculation of EYards/Pass Attempts and you get an approximation of DVOA (actually VOA, for now) for the QB.

And if you do that, Brees in near the top of the list. The thing is, he has the "real" yardage total to back it up, whereas Philip Rivers and Jon Kitna do not.

20
by Lou (not verified) :: Tue, 09/09/2008 - 11:08pm

17

i imagine its something like before if a player had a low or DVOA but a lot of attempts they could rank pretty high in DPAR. that said their are other players that had a similar number of attempts and had "good" games and still rank below Brees.

am I the only one that thinks DYAR isnt going to bring anyone new to this site. It'll just serve to confuse the current readership?

21
by Lou (not verified) :: Tue, 09/09/2008 - 11:15pm

thats the second straight post where i've used "their" instead of "there." I'm calling it a night.

22
by DrObviousSo (not verified) :: Tue, 09/09/2008 - 11:21pm

Ok, I guess I'll just focus on DYAR for now, and let everything else sort itself out.

Is DYAR still a counting stat, am I'm just totally spun around

23
by jack (not verified) :: Tue, 09/09/2008 - 11:27pm

Brees's EYds were probably lower because a large chunk of them came from 3 plays.

Also: Reggie Bush tied with Andre Johnson in DYAR. Sick!

24
by mm (not verified) :: Tue, 09/09/2008 - 11:29pm

7-OK, that reads to me like Manning was worse than standard stats made him look.

But his EYards were 40% higher than his yards, which isn’t just better, but much better.

I am not finding this intuitive, at all

I think yards and Eyds don't make as good a comparison as they say, because (if I understand correctly) TDs and Ints are also included in Eyds, so most QBs will get somewhat higher Eyds than yards. I also believe rushing will affect this, though that doesn't affect Manning's number.

Peyton Manning passed a lot on Sunday. Yet, for all that passing, he was only 68 yards above replacement (not adjusted for opponent).

For comparison, Brodie Croyle was exactly 'replacement level' against the patriots this Sunday. Given that Manning had 49 attempts, that means that Manning was barely one yard above 'replacement' (ie Croyle-level) for each pass he threw.

For the EYds statistic, you take how many Eyds a replacement-level QB (Croyle) would have gotten in those attempts, then add on the yards above replacement. Because there was nearly 50 attempts, even a replacement QB would have gotten a big number, thus Peyton's big number here doesn't represent a superstar performance. It does show that the offense leaned on the passing game, and that Manning didn't get hurt (we've all been reminded this week that durability has value).

14-How does Brees have fewer EYds than actual yards? Didn’t he have a good performance?

107 yards above replacement is pretty good, so I think this again shows that comparing the size of yards and Eyds isn't as informative as they suggest it is.

I should add that I think Brees number will get bigger once defenses are factored in.

25
by Scipio202 (not verified) :: Tue, 09/09/2008 - 11:32pm

So, looking at the correlation between DYAR and (EYds - Yards) within groups of QBS with the same number of TDs and INTs (as long as there are more than 3 of them): 2 TD, 0 INT: correl = 0.87; 1 TD, 0 INT: correl = 0.54; 1 TD, 1 INT: correl = 0.67; 0 TD, 0 INT: correl = 0.84. So the EYds heuristic is very good for two groups, and only kind of good for the other two groups. Looking at (EYds-Yards)/Att is worse: 2 TD, 0 INT: correl = 0.64; 1 TD, 0 INT: correl = 0.47; 1 TD, 1 INT: correl = 0.63; 0 TD, 0 INT: correl = 0.87. (sorry for the formatting)

26
by Pat (not verified) :: Tue, 09/09/2008 - 11:33pm

1) Manning EYds greater than Yds: That's because Manning did actually have a better day than his traditional statistics would indicate. 257 yards on 49 attempts is horrendously bad. Problem is that 358 yards on 49 attempts is still bad. A great day for a QB would be putting up well over 400 on 49 attempts. This is a case of "better than what it looks like, but still not good."

2) Brees Yds less than EYds - what's so difficult to understand about that? Brees had an interception and a bunch of incompletions on 3rd down (and a 5 yard pass on 3rd and 6).

That one actually makes obvious sense: if you look at McNabb and Brees, their yardage and stat lines are kinda close. 21/33, 361 yards, 3 TDs, 0 INTs, vs 23/32, 343 yards, 3 TDs, 1 INT. But even taking into account the interception, McNabb's day was waaay more effective than Brees's.

27
by Fergasun (not verified) :: Tue, 09/09/2008 - 11:34pm

Vince Young: Young's most effective yards were the ones when he walked off the field to allow Kerry Collins to get on. Collins earned 48 YAR on just two pass attempts (65 passing yards).

Nice!

28
by Fergasun (not verified) :: Tue, 09/09/2008 - 11:36pm

Where was Brandon Jacobs? 6th or 7th?

Also... I thought Pac Man Jones was a great CB when his head was in the game... no surprise he could cover Braylon Edwards.

29
by Tom D (not verified) :: Wed, 09/10/2008 - 12:02am

I'm wondering where Matt Forte came in, also I think Aaron is a little harsh on Orton.

30
by Dr. Mooch (not verified) :: Wed, 09/10/2008 - 12:02am

I'm getting a T-shirt that says, "Brian Moorman: Better Than Tom Brady in 2008"

31
by mikeabbott (not verified) :: Wed, 09/10/2008 - 12:05am

the sorting column is easier to find on the far right than in the middle.

that is why newspapers but w/l/t in th middle and whatever the standings are based on at the end.

32
by The McNabb Bowl Game Anomaly (aka SJM) (not verified) :: Wed, 09/10/2008 - 12:53am

I take back what I said about Brees. He obviously did have a good day, just not as good as his raw yards would suggest. Similarly, Manning had a better day than his numbers suggest, but still not a really good one.

Thank you, Temo and Pat, for clearing this up.

So the comparison of raw yards to EYds is still useful, you just have to keep an eye on the attempts. Maybe add columns for raw YPA and EYdsPA so we can see the comparison to each other and also compare player to player?

For example:

Brees Raw YPA=10.71 EYdsPA=9.78
Manning Raw YPA=5.24 EYdsPA=7.31

So we can see how advanced statistics pull Brees down and pump Manning up, but Brees was still better than Manning. I guess you can see that just by looking at the DPAR but this really clarifies the role of EYds, which are almost useless without this analysis. I almost feel like the entire justification for EYds is for use in YPA.

33
by The McNabb Bowl Game Anomaly (aka SJM) (not verified) :: Wed, 09/10/2008 - 1:01am

In fact, this is the real reason DYAR and EYds were invented- to give new readers a stat they can relate to! Aaron just made the mistake of trying to convert DPAR into yards, when the real solution was to convert DVOA into YPA.

I now humbly suggest that the new "FO stat for the masses" should be EYdsPA.

34
by cd6! (not verified) :: Wed, 09/10/2008 - 1:03am

I think there's a problem of trying to gain too much information from the stats. The whole reason that EYds was used was to say "people see yards and judge whether or not a QB had a good day, so let's make a stat that tells more information and keeps it on the same scale."

However, in that realm, Joe Football fan isn't also thinking "attempts." So, if I say Manning threw for 400 yards on Sunday, then most people will assume "he had a good day" without asking the follow up question of "well did it take 50 attempts or 5?"

On that level, then most people assume 400 yards is a great day, 300 is a good day, 200 is average, and anything from 100 on down is lousy.

Based on what I understand of the intent of EYds, then ideally players can be adjusted to fit onto that same scale.

If Romo goes out and throws for 400 yards, even if most of it was garbage yards, and his EYds is 300, its less than his "traditional yards" but still a good day.

If Roethlisberger goes out and only throws for 150 yards, but gets 2 touchdowns and a handful of first downs, than his EYds should be up in the 200-300 range, meaning, he had a better day than his meager yard total would suggest.

The important thing is that you don't immediately know how many attempts there were, but you shouldn't need to, or you've failed the intent of the EYds.

All that in mind, I can't understand the logic behind Peyton Manning's total or the comment on the ESPN page. 250 yards is pretty average. But considering he had so many attempts and passes for short yards, I can't understand why his EYds would be so much higher than his real yards. Saying he had 250 yards is a better description of his day than saying he had 350 Eyds, which would seem to inflate his performance.

Man what an epic that was. Hello, carpal tunnel syndrome.

35
by The McNabb Bowl Game Anomaly (aka SJM) (not verified) :: Wed, 09/10/2008 - 1:24am

cd6!,

The problem is that you can't just say 300 yards is a good day. Manning proves your logic wrong. He did not have a good day. He had an OK day. But his EYds is accurate. He had a 350 yards on 49 attempts day. You can't ignore attempts, it simply doesn't work.

What the Manning comment should have said is "Manning's day looks horrible when you consider the number of attempts. His high EYds shows that it wasn't as horrible as it looked, but his DYAR shows that it wasn't great either." This would be a lot simpler if the chart had his YPA and EYdsPA (see post #32), which easily demonstrate this fact.

36
by Matt (not verified) :: Wed, 09/10/2008 - 2:08am

Braylon Edwards was "victimized" by drops? If he was the victim of the drops, who was the culprit?

But as for DYAR, True Yards, Effective Yards, etc., let me just call it now.

2008: the year FO jumped the shark. Maybe all of these explanations of what Manning's EYds mean are right,and maybe they are not; I have no way of assessing that myself, and Aaron is too busy recuperating from a long day at the IUPUI Barnes & Noble to explain.

37
by cowfez (not verified) :: Wed, 09/10/2008 - 2:24am

Wait a minute... Eyds?
AIDS?!

38
by Mystyc (not verified) :: Wed, 09/10/2008 - 3:02am

People have accepted Yards Per Carry for running backs, so clearly rate stats are not beyond them. I'm sure most people could even tell you that 4 YPC is an average day, 3 or less is bad, and 5 or more is great. The masses might be hung up on completion percentage for QBs right now, but so what? Give them something better.

I find EYPA (or EYPTouch/Target for RBs and WRs) far more useful than straight Effective Yards, and having that for a rate stat and DYAR for a counting stat seems like an excellent pairing.

39
by ABW (not verified) :: Wed, 09/10/2008 - 4:57am

I was willing to wait until the season to pass judgment on the whole DYAR/"effective yards" thing, and I can now say definitely that "effective yards" make no sense to me whatsoever. DYAR is fine, I found DPAR to be a bit more intuitive but whatever, I look at the DYAR numbers and they look about right - McNabb clearly had the best day, Young was terrible, and everyone in between seems to more or less line up right.

Effective yards seem to be just completely random numbers with no relationship with how valuable the player actually was. Please bring back DVOA/DPAR, or at least DVOA/DYAR. The "effective yards" thing just weakens the whole point of advanced statistics, which is to make it so you can tell who provided value to their team and who didn't and how much. It just makes it more confusing - why are Joe Flacco and Tarvaris Jackson so valuable? It seems like if you rank the players by EYds it ranks them more or less correctly, but the numbers just seem to have no relationship to anything on the field. I see 6.5 DPAR, I think "OK, that player was worth almost a TD". It makes sense. 266 Eyds? I have no idea what that actually means.

I really liked DVOA and DPAR and talked them up to my friends and tried to get them into FO(with a modicum of success), but I'm really going to have a hard time selling the "effective yards" thing.

Between this and the half of an AGS this week it is not an encouraging start to another year of FO.

40
by ammek (not verified) :: Wed, 09/10/2008 - 5:02am

I'm with 31: would it be possible to put the figures for rush and pass/rec DYAR in a different color, or not in bold? There's just a jam of figures on the right side of the column which makes it harder to do a "quick read".

41
by steelberger (not verified) :: Wed, 09/10/2008 - 7:31am

It really looks like FO stats are moving towards vastly overvaluing quantity over quality.

FO is losing its way. Sad really.

42
by bengt (not verified) :: Wed, 09/10/2008 - 7:35am

Schaub to Andre Johnson: 10-for-12 (plus a pass-interference call) for 112 passing yards
Schaub to Andre Johnson after it was 35-3 with 2:04 remaining in the third quarter: 6-for-7 (plus a pass-interference call) for 62 passing yards

43
by Boggle (not verified) :: Wed, 09/10/2008 - 8:30am

I don't want to pile on, but I also much preferred DPAR.

I understand the reason for the switch - fans are familiar with yards - but I still disagree with the decision.

Beyond ease of use (DPAR was certainly easier to understand for me), the other thing I liked about DPAR is that you could use it to get an approximate comparison of players across positions. So if a QB had a DPAR of 12 and a RB one of 6, I could say that the QB was twice as valuable to his team. Clearly passing yards and rushing yards are not the same (partly because the variance on passes is much higher), so I guess there is no way to make these comparisons anymore.

44
by cd6! (not verified) :: Wed, 09/10/2008 - 8:31am

35 & 38

For the record, I agree with you. But that's not how "just converting performance into one stat" works for the masses.

With running backs its the same way. If a guy gets 150 yards, its a great day, 100 is a very good day, anything under 50 is lousy.

Obviously, whether or not it took 40 rushes to do it is the second, follow up question, but when you're discussing fantasy results with your buddies or whoever, its generally "LT had 120 yards, hooray" and that's it, everyone then accepts he had a good day.

45
by Temo (not verified) :: Wed, 09/10/2008 - 8:37am

A few things here:

1. It really looks like FO stats are moving towards vastly overvaluing quantity over quality.

That's what DYAR/DPAR is supposed to be, a counting stat! If you want a rate stat, you want DVOA.

2. I now humbly suggest that the new “FO stat for the masses” should be EYdsPA.

What you're really looking for is DVOA :) Effective yards=DVOA translated to Yards per attempt, multiplied by attempts.

3. Effective yards seem to be just completely random numbers with no relationship with how valuable the player actually was. Please bring back DVOA/DPAR, or at least DVOA/DYAR.
To everyone who complained about Effective yards. If you don't like the stat, then don't use it. DYAR is more or less the same as DPAR (as said before, DPAR is about equal to DYAR/7). Effective yards is, again, simple DVOA converted to yards per attempt, multiplied by attempts. That's the relationship.

46
by andrew (not verified) :: Wed, 09/10/2008 - 10:35am

Here’s the first Quick Reads featuring YAR and EYds. (That’s right, the folks who didn’t like the name “True Yards” won — the name will be changed on all the sheets to “Effective Yards” in the next few days.)

In honor of this change, the phrase "True Dat" will be replaced by "Eff Dat".

Also, this will make the FO nickname for Adrian Peterson more appropriate.

47
by MJK (not verified) :: Wed, 09/10/2008 - 11:38am

If you have a time machine, go back in time to an hour before the Patriots game. Bet your friend $100 that Brian Moorman will throw more touchdown passes in 2008 than Tom Brady. Hope your friend doesn't see your escape pod.

"...because, just as sure as you stand there, you're going to get an earful if cider!"

By the way, I think you got "least valuable RB" wrong. I nominate Sammy Morris. Having a 28% success rate doesn't nearly compare to allowing the man you were supposed to block end the season of your reigning MVP star QB.

48
by Led (not verified) :: Wed, 09/10/2008 - 11:38am

Did DO used to include DVOA in the Quick Reads? If so, I wish they'd go back to it. I find it's more informative to know how a QB did on a per pass basis. But I can see the value in the counting stat -- that's why they should include both.

49
by chip (not verified) :: Wed, 09/10/2008 - 12:00pm

Where did Forte finish up? 6th? 7th?

50
by ADS (not verified) :: Wed, 09/10/2008 - 12:12pm

What's with the Orton hate? A game manager is someone who doesn't 'Eff it up for the defense. And that's exactly what he did. He didn't win the game. Everyone knows that. If he finishes in the black on FO's decidedly unintuitive new metrics, most fans will be estatic.

My anti-spam word was DPAR. Oh the irony.

51
by mm (not verified) :: Wed, 09/10/2008 - 12:38pm

If you only look at one stat, look at YAR. However, YAR and Eyds together give you more information than just YAR by itself. This is especially true once you get to the middle and bottom of the list.

A player can come in a few plays a game, have a good YAR value, but low Eyds. Its quite possible his YAR value would go down if he played more often (while his Eyds might creep up). Some of his success may be as a 'change of pace' that defenses haven't focused on.

Similarly, a slightly above-average RB that can start and consistently take a lot of carries per game without getting injuries is going to pile up a lot of Eyds but may not be getting a huge number of YAR because he's getting focused on & beat up. He's still providing real value to the team. The team doesn't have to focus on high priced backups, and it opens the passing game up because the defense focuses on the consistent (but not game-breaking) running game.

In this case, a team can be successful if they have 1) a strong Defense and/or 2) a QB with a high DYAR but low Eyds. Note that if the RB ran less his Eyds would likely go down, but his DYAR may go up (he's getting less focused on and is less tired & beat up).

Obviously, you would like a player with large YAR and large Eyds. However, some players are good in one and not in the other, indicating they contribute in different ways.

If you have a player with large YAR but low Eyds, he's having a few good plays but isn't contributing on most plays. This may be the case of a QB who doesn't throw much because of a solid running game, your deep-threat backup reciever who's only thrown to twice a game, but makes a huge play every 3 games, or your change of pace running back.
When you look at the season stats, this can also indicate a starter who's good when he's on the field, but can't stay healthy.

The concepts are a bit clearer in baseball (which influenced a lot of these statistics). A pitcher who can go 9 innings and give up 4 runs can be more valuable than a pitcher who can goes 6 innings and gives up 2 runs. The second pitcher may have a higher 'DYAR' equivalent, but the first pitcher has a higher 'Eyds' equivalent, because he pitched an extra 3 innnigs and saved the bullpen.

52
by GlennW (not verified) :: Wed, 09/10/2008 - 12:38pm

> "To everyone who complained about Effective yards. If you don’t like the stat, then don’t use it."

I really don't think "then go pound sand" is the appropriate response to valid critiques of a statistic (EYds) that is being presented to the mainstream at ESPN and is confusing to say the least.

The issue with EYds (as exemplified in the case of Peyton Manning) is that first we have to evaluate what we *think* we know about the player's performance based on the raw stats (not very good) and then compare that with EYds, even though EYds also incorporates TDs, INTs etc. (converted to yards) into the comparison. So is 358 EYds a good performance or not, even given the 49 attempts? Hell, I have no idea, and really don't want to have to make another calculation for all QBs to establish a baseline EYd/Att in order to figure that out, or even to make a rough guess at it. At first (and maybe second) glance EYds simply looks like an alternative counting statistic at which Peyton Manning did very well, when a more meaningful counting statistic has also been provided that shows that not to be the case.

DYAR is basically fine (independent of the DPAR/DYAR debate). But as opposed to its description as "an even quicker and easier way to see how good a player was", EYds is confusing, extraneous clutter. Given the puzzled reaction from the more devoted readers in this forum, I'm afraid that to the average ESPN reader the dual DYAR/EYds presentation is mind-boggling. You're simply rating players on a weekly basis with some inherent degree of error already assumed-- so keep it simple, with Total DYAR. Just my opinion.

53
by The McNabb Bowl Game Anomaly (aka SJM) (not verified) :: Wed, 09/10/2008 - 12:59pm

Temo,

I know that EYdsPA = DVOA. That's the whole point. Aaron and co. were looking for a way to make FO stats more intuitive for new readers. DVOA is not intuitive. EYdsPA looks and functions exactly like YPA for QBs and YPC for RBs, making it far more intuitive to noobies.

See the Manning and Brees examples above (post #32). EYds doesn't really tell us much that's useful. But EYdsPA is a simple conversion of DVOA to traditional and recognizable stats.

54
by GlennW (not verified) :: Wed, 09/10/2008 - 1:30pm

> "The concepts are a bit clearer in baseball (which influenced a lot of these statistics). A pitcher who can go 9 innings and give up 4 runs can be more valuable than a pitcher who can goes 6 innings and gives up 2 runs."

I don't believe this is the proper analogy. The second pitcher has a better ERA, but that's a rate statistic which we can readily differentiate from a counting stat such as VORP or the like. On the other hand DYAR and EYds are both counting stats but with different baselines.

In your example perhaps there'd be an analogous pitching counting statistic called "effective runs allowed" in which we'd rate the second pitcher at say, 2.3 runs, such that we could deduce that his real performance of 2 runs allowed was deceivingly low because of the fewer innings pitched (we would call this pitcher "Daisuke Matsuzaka"). Such a statistic could only be used in the context of that pitcher's actual deployment in the game and wouldn't be terribly useful for comparison purposes against other pitchers' performances.

As such I can understand the call for DVOA (or EYds/Att) as the alternative rate statistic to the counting stat DYAR in "Quick Reads", but think that for single-game ranking purposes presentation of just Total DYAR is sufficient (and at the very least not confusing).

55
by Temo (not verified) :: Wed, 09/10/2008 - 1:32pm

52. Maybe I was too dismissive in my post, but I believe #51. really outlined it pretty well.

What I meant by "If you don't like it, don't use it" is that people are complaining that EYDS isn't an effective measure of "how "good the player did". But it's not MEANT to do that. All it's supposed to do is give you an idea of the overall value created by the player. If you want to know "how good the value provided was", then use DYAR. If you want to know how well he did on a per play basis, then use DVOA (or EYDS/PA... which I now agree, #53, should probably be included). But it doesn't seem logical to complain that EYDS is NOT something it was never meant to be.

The story these stats tell about Manning is that he gave 358 Yards of value on the day (EYDS), which is tied for 4th most in total value for the week. However, he gave only 68 more yards of value than a replacement level player (DYAR), which was 12th most for the week.

On a side note, I blame most of the misunderstanding on the Manning comment, which makes it seem like he did nothing well that day (which is not what was intended to be said, I'm sure). Manning actually passed for 15 first downs on the day, which is a pretty good amount of total value (or, EYDS). However, the fact that he had to take 49 attempts to get those 15 first downs is what depressed his DYAR and makes it so that we know he really had just an OK day and not a great day. If he had only taken 30 attempts and had the same production, his DYAR and DVOA would be much greater, and his EYDS the same amount.

56
by mm (not verified) :: Wed, 09/10/2008 - 1:45pm

I don’t believe this is the proper analogy. The second pitcher has a better ERA, but that’s a rate statistic which we can readily differentiate from a counting stat such as VORP or the like. On the other hand DYAR and EYds are both counting stats but with different baselines.

I'd compare the relation between DYAR and Eyds to Win Shares Above Bench and Win Shares.

57
by mm (not verified) :: Wed, 09/10/2008 - 2:24pm

Some examples of starters who have high DYAR/low Eyds who might not be as good as DYAR indicates:

The QB who only has to throw 15 times in a game because the special team scores, the defense is domininant, or the running game is so consistent. Other players (the RB, the D, or the returners) are providing more value to the team. If his RB goes down and suddenly there is not 8 in the box he may struggle to find open recievers. If his defense gives up points and he has to play from behind, he may collapse when the opponents decide to turn up the pass rush.

A RB in a top passing offense. The defense is gearing up to try and stop the pass, leaving open running lanes. If suddenly the starting QB goes down, he may find the going much tougher.

A #2 receiver. Your #1 gets the toughest coverage and is usually targeted more often. Certainly there are lots of examples of good #2s who get signed away in free agency who utterly fail when they're expected to be a #1.

A player with high Eyds has to be:
1)Durable--not get injuries
2)Able to handle being focused on by the defense and still be better than replacement value (and remember, replacement value doesn't know if you're being focused on by the D)

58
by DB (not verified) :: Wed, 09/10/2008 - 2:37pm

RE: A bunch of posts... (although I think 55 may already have said this)...

Is it just me, or is the almost the whole problem people have with EYds due to Aaron's comment about Manning?

The comment made it sound like Manning had a bad day, despite OK numbers. What EYds says is that Manning had a good day. That's the source of confusion I think.

59
by mm (not verified) :: Wed, 09/10/2008 - 2:51pm

One more example to illustrate differences between DYAR and Eyds.

Lets say you're the GM of a team with a lousy backup QB. You have a choice of 2 QBs to sign for the next 3 years.

Player 1 is guaranteed to get spectacular DYAR in your offense, but will have mediocre Eyds because he's also guaranteed to go down with injuries for large stretches of the next 3 years. We'll call this player Wurt Karner.

Player 2 will take every snap under center for the next 3 years. Despite this we'll say player 2 gets less overall DYAR than Karner because he makes a few too many risky throws. Nevertheless, player 2 is an above average QB who'll have good DYAR and accumulate vastly more Eyds than Karner. We'll call player 2 Frett Bavre.

Who do you sign? Who's more valuable?

For me, it depends on the quality of the rest of the team. If I have a poor to mediocre team, I very well may sign Karner. Bavre would likely win more games overall in the next few years; however, Karner is more likely to put together one transcendent stretch that gets you into the playoffs one year and wins a superbowl. It's still unlikely, but his higher DYAR/game played gives him that chance.

If I have an above-average to great team around the QB, I'm definitely signing Bavre. You won't have any games played with your awful backup. You'll definitely get in the playoffs 3 times with a fighting chance to win the super bowl each of those years.

60
by drobviousso (not verified) :: Wed, 09/10/2008 - 2:53pm

(or EYDS/PA… which I now agree, #53, should probably be included)

I've created a google spreadsheet showing this for all the qb's listed. Clickey Sigy. Top of the list looks like this:
Cutler--16.13
McNabb--14.55
Roethlisberger--14.50
Kolb--13.33
Ryan--12.62
Romo--11.91

Which seems about right. Cutler, McNabb, and Romo all had a significant number of attempts, but a lot of productivity. Roethlisberger, Kolb, and Ryan each had und 20. I didn't see Kolb, but Roethlisberger and Ryan were both very efficient when they did throw.

61
by Temo (not verified) :: Wed, 09/10/2008 - 3:04pm

Actually, mm (can't you get a better posting name!), all a high DYAR over low EYDS gives you is a great DVOA. Go ahead, do the math :p

What you mean to say is someone with great DVOA who doesn't have great EYDS.

62
by Temo (not verified) :: Wed, 09/10/2008 - 3:15pm

According to drobviousso's spread sheet, Manning had a 7.31 EYDS/Att, roughly middle of the pack. This, despite the fact that he was the 4th best in terms of total production for the week. Seems about right.

63
by GlennW (not verified) :: Wed, 09/10/2008 - 3:21pm

> "The comment made it sound like Manning had a bad day, despite OK numbers. What EYds says is that Manning had a good day."

Actually this last statement is entirely unclear (we've had varying opinions on Manning's performance), which I think demonstrates the problem with the EYds counting statistic being presented in this simple weekly ranking format. Manning was entirely mediocre in EYads/Att (or DYAR/Att, either way), but with the EYds statistic we'll give him more "credit" for his contributions in higher volume. EYds only suggests that Manning contributed a lot of offense of some undetermined value, but in football opportunity varies so much between players that this isn't terribly meaningful-- which is the reason that advanced statistics have moved away from comparison with raw statistics in the first place.

I now understand what the EYds statistic means; I have to admit that I didn't fully before this discussion, given Aaron's explanation (again, basically because "better/worse than standard stats make the player look" completely depends on how one has previously interpreted the "standard stats"). I just think there's such a thing as presenting too complicated or too much information to a particular audience, and this is one such case.

64
by Tom D (not verified) :: Wed, 09/10/2008 - 3:50pm

Re 62:

I just did calculated the average of all passes this week and the average was 7.66, so Peyton comes in a little below average. Eli is almost exactly average with 7.6 EYds/At. 20 QBs come out above average, including Kyle Orton, but he doesn't play defense so he didn't help the team.

65
by Temo (not verified) :: Wed, 09/10/2008 - 3:57pm

63. I just think there’s such a thing as presenting too complicated or too much information to a particular audience, and this is one such case.

I would probably agree. The progression of thought should probably go:
A)Manning had a very high EYDS and a low actual Yards. That means he played better than his conventional stats would otherwise indicate.
B) But wait, his DYAR is pretty average. That means that although his conventional stats are worse than they should be, it's not like he had a great day or anything.

But how many people are going to go from A to B, considering you have three different yardage totals to weigh? I know a lot of people in this thread didn't (me included) at first.

66
by mm (not verified) :: Wed, 09/10/2008 - 4:02pm

Actually, mm (can’t you get a better posting name!), all a high DYAR over low EYDS gives you is a great DVOA. Go ahead, do the math :p

What you mean to say is someone with great DVOA who doesn’t have great EYDS.

Obviously, they're closely related. But we're trying to explain what the 2 counting stats are. So I feel duty-bound to point out one situation where that wouldn't be the case, and should further clarify what the stats measure.

Because its a counting stat, looking only at DYAR will eliminate a few people with just 1 or 2 plays in a game. For example, a #4 WR who is thrown to once in a game for 20 yards on 3rd and 15 may get a large DVOA, but not accumulate much DYAR. Note he probably faced single coverage in this example.

Add a few decent (but not so crucial) catches to that, and his DVOA would drop, but his DYAR and Eyds would both increase. This might be a good #2 or #3 receiver who would be included in what I wrote earlier. Almost certainly the defense isn't focusing on him.

Now target him a bunch more and, as the secondary focuses on him, a few incompletions result, along with several decent catches (which is only marked as 'replacement level', even though its against double or triple coverage), along with some good catches.

His DYAR might not change much; it might even drop, but this reflects the fact that he may now have a better corner on him and certainly will see safeties quicker to add help. His DVOA would also drop, but his Eyds would increase. Note that now other receivers are now facing weaker coverage, and the running game may find more holes available.

67
by Jimmy (not verified) :: Wed, 09/10/2008 - 5:56pm

Drobvioso's chart might be the best chart for analysing a week's QB play that I have ever seen. Seriously it really seems to chime with how I saw the players play (which isn't the best test, but isn't the worst, and it was quick).

If that chart were the one FO used and get updated with opponent adjustments it would be a great statistical tool. It would get my vote for the preferred method of presenting the new FO stats, it is clear and makes great sense.

68
by The Ninjalectual (not verified) :: Wed, 09/10/2008 - 6:25pm

I just divide DYAR by 10 and pretend that I'm reading about DPAR. 10 is much easier than 14.3 and this way I can still pretend that everything makes sense.

I will never stop reading FO as long as it exists, but I am on the bandwagon labeled, "DYAR is stupid; DPAR made sense." Long live DPAR!!!!!

69
by The McNabb Bowl Game Anomaly (aka SJM) (not verified) :: Wed, 09/10/2008 - 6:58pm

Re: 67

Actually, drobviosso's chart is nothing more than a ranking of QBs by (D)VOA, the standard rate stat used by FO to measure everything. EYds/attempt is the same thing as DVOA, just in a more intuitive format. (I'm not saying they should get rid of DVOA, it's far more useful if you understand it, but EYds/attempt is something a new reader can digest easily.)

FO has never ranked skill players by DVOA in a weekly column, always preferring DPAR/DYAR, and I think that makes sense. We want to judge weekly performances by who contributed the most value above replacement (DPAR/DYAR), not who contributed the most raw value (EYds) or who was best per attempt (DVOA or EYds/attempt).

Think of Ben Roethlisburger's first couple of years, when he was great when he threw, but he only threw rarely. Was he providing the best performance each week? No, because his team's offense relied so heavily on their running game. He was providing a great but limited performance, and DPAR showed that better than DVOA.

70
by The Hypno-Toad (not verified) :: Wed, 09/10/2008 - 7:15pm

I'm sure I'll come around, but for now I'm having trouble with the new stats. When I first started reading this site, I didn't understand DVOA or DPAR particularly well, but when I looked at them, I could tell there was really good information there, if I could just be smart enough to understand it. I'm not really getting that feeling from Eyds. Maybe that's just because I've been reading these advanced stats for a few years now, so the sense of revelation isn't really there anymore.
Wow. Awesomely pointless post. I win the gold in the "wasting your time and mine" olympics.

71
by Temo (not verified) :: Wed, 09/10/2008 - 9:04pm

70. For what it's worth, EYDS is more or less useless to me, and I'm a big stat nerd. That doesn't mean it's meaningless... I just don't use it much.

69. Alternatively, David Garrard of 2007. Although it's easiest to look at DVOA vs. DYAR when look at RBs/WRs, since most QBs who pass well will be asked to pass more often (barring injury or good defenses and young qbs).

72
by Jimmy (not verified) :: Wed, 09/10/2008 - 9:05pm

#69

I do understand the other stats, and think they are pretty good. I do think that the Eyds/att is easier to look at than DVOA though. Percentages are all well and good, but what is it a percentage of? Isn't it just easier to see how many effective yards each pass a certain QB throws?

73
by Jerry (not verified) :: Wed, 09/10/2008 - 9:15pm

So Hank Baskett with his 2 catches was the 5th most valuable receiver this weekend huh? Why? Even if you take his TD away, Philly still wins by what, 28 points. Dante Rosario with the Panthers on the other hand had 7 catches (out of 8) for 96 yards, 5 for first downs and the winning TD in the final second of the game. But Baskett with his two catches was more valuable than Rosario huh? Sorry, but I just don't get it.

74
by Temo (not verified) :: Wed, 09/10/2008 - 9:35pm

73. Baskett had the 5th most valuable performance given the 3 attempts he had. In your example, I'm sure Rosario would have the higher Effective Yards, but not the higher DYAR.

75
by Stevie (not verified) :: Wed, 09/10/2008 - 11:21pm

I looked forward to Quick reads every week last year but the new stats seem to go against common sense and I admit I cant wrap my head around the Manning thing. If this is the response from the FO community, what are the howling masses at ESPN saying?

76
by Pat (not verified) :: Thu, 09/11/2008 - 12:44am

and I admit I cant wrap my head around the Manning thing.

I really, really don't understand why it's difficult to get the Manning thing.

Look at his traditional statistics. Good or bad? Not so good. Sub-300, for one, even if you ignore the ginormous number of attempts. Look at his effective yards. Good or bad? Much better - well over 300. So he played much better than his conventional statistics would indicate.

Now, did he have a good game? Look at YAR. It's ridiculously middling: 68 YAR. In other words, Joey Harrington probably would've put up nearly 300 effective yards given the same number of opportunities. So no, it wasn't a particularly good game.

Imagine if someone tells you "Manning wasn't the problem. He put up 257 yards or so, which is fine for a QB." That's true. And you can't really refute it by saying "those were a lot of junk passes," because, well, they weren't. The way you refute it is by saying "yeah, but he did it on 49 passes" - and you can see that from his YAR.

Now imagine if someone tells you "well, Garrard wasn't too terrible. He still had over 200 yards passing." The problem was that they were a lot of junk passes coupled with two interceptions, and you can see that from his effective yards.

You can see that he was crap in YAR, too, but you can't immediately see why.

77
by andrew (not verified) :: Thu, 09/11/2008 - 7:31am

Tarvaris Jackson had the best rushing dvoa of any qb in week 1, even better than Flaco's which included a long TD run.

Just about every time he took off he got a first down, sometimes with a long ways to go.

However, some of those yards would be what I term "stupid yards", plays in which he got the first down and tried to get a few more and took a big hit for doing so instead of sliding.

Its probably not feasible to chart this, but I wonder how such a statistic would compare out, given the proposition that any yards beyond the first down marker when there are defenders within say three yards of a quarterback are "stupid" yards, as long as it isn't some situation with less than two minutes left ala Steve Young. (you could also make cases for stupid yards where players who could run out of bounds with time an issue and instead turn upfield, and plays like Bigby's trying to runback an INT with the game won).

78
by jimm (not verified) :: Thu, 09/11/2008 - 7:53am

Regarding T Jackson - his long passing is horrendous. MNF pointed out the Vikings completed just 4 passes over 20 yards last year - two of those weren't even Jackson - one wasn't even a QB (Rice to Shiancoe).

Against GB Jackson had Berrian deep twice with inside position and no safety - he threw both over the wrong shoulder and one 10 yards long the other a couple of yards short. Half decent passes lead in the proper area would have likely resulted in huge plays.

Don't blame Berrian for having a so so game. The QB blew it.

79
by cjfarls (not verified) :: Thu, 09/11/2008 - 11:45am

#74's example of Baskett vs. Rosario comparison for YAR vs. EffYards is the first thing that has ever made any sense to me re: why we would want to have EffYards.

EffYards/Attempt makes a lot more sense to me as a "user friendly" stat that tells something I didn't obviously know. Effyards by itself may occassionally show value (Rosario vs. Baskett example), but particularly if you're not showing Rosario on the table, it really makes no sense to me on why you would publish the stat at all... it just makes me confused.

Just DYAR, with or without EffYards/Attempt I think would be much clearer.

80
by Reinhard (not verified) :: Thu, 09/11/2008 - 6:26pm

Manning played like a mediocre quarterback (DYAR = 68) but racked up production by being given so many opportunities (358 EY on 49 attempts)

81
by Temo (not verified) :: Thu, 09/11/2008 - 8:37pm

80. Wow. The entire thread summed up in 21 words. Thanks :(

82
by Staubach12 (not verified) :: Fri, 09/12/2008 - 2:59am

Okay, I have to agree with those who don't like EYDS. Look, if you are hoping to present a stat that newbies can understand, then EYDS is not it. I think DYAR and DVOA are both very easy to understand.
DYAR=how many more (or less) yards this player earned than an average (replacement) QB would have earned in the same situation
DVOA=How much better or worse (as a percentage) a given player is per play than average

I think most people can understand it if you say Joe Quarterback threw for 150 yards more than the average NFL QB could have in that game & Joe was 15% better than the average QB on a per play basis.
However, I can't really conceive of a concise way to explain to an average reader what EYDS means. That tells me that this new stat will not make FO's ideas more newbie-friendly.
My advice: keep DYAR (it is much easier for newbies to understand), scrap EYDS, and add DVOA (percentages are pretty easy to understand) to Quick Reads.

83
by Staubach12 (not verified) :: Fri, 09/12/2008 - 3:18am

Followup: I'm not saying that EYDS should be eliminated from FO entirely. I'm saying that it shouldn't be part of quick reads because it is just too darn hard to understand. I get that EYDS does provide unique information, I just don't think that this is the sort of info that newbies are looking for. DYAR and DVOA should suffice for ESPN readers.

84
by Scott C. (not verified) :: Fri, 09/12/2008 - 5:32pm

Oh, "Effective Yards" won? ! That makes me happy. Better than "Equivalent Yards" (equivalent to what?) or "True Yards" (aren't real yards true yards?).