Writers of Pro Football Prospectus 2008

26 Jan 2008

FO Mailbag: Giants vs. Other 3-6 Seeds by DVOA

This comment was made by reader Gerry in the thread discussing Friday's ESPN article on the Giants' place as one of the weakest teams to ever make the Super Bowl.

One more thing that sticks in my craw: "Out of the 10 teams that required three wins in order to make the Super Bowl, the Giants have the LOWEST average margin of victory in the playoffs."

Perhaps. However, they have had over 40% DVOA in each of their playoff victories. I know you don't have DVOA numbers for all 10 of the teams that required three wins to make the SB, but you do have some. What was their average three game DVOA? Why the sudden return to using conventional stats when they are saying something different than the advanced stats?

First of all, to answer the question posed in the last sentence: When you write for a larger audience, and the conventional numbers and advanced numbers tell the same story, you want to use the conventional numbers. They are simply easier for the general public to understand.

Nonetheless, we can answer this question by looking at the seven teams since 1996 that required three playoff wins to make it to the Super Bowl (any team seeded third through sixth).

What we have listed here is the game-by-game DVOA for each team in their first three playoff victories, along with the average of those three games. Then I've listed the regular-season DVOA and rank of each team, along with the regular-season DVOA and rank of the team they would face in the Super Bowl. Teams that need three wins to make the Super Bowl are certainly not all cut from the same regular-season cloth, and they don't all go into the Super Bowl as underdogs. I've ditched the decimal points for space purposes.

Year Team WC Div Conf Avg Reg Rk vs. Reg Rk
1997 DEN 29% 12% 32% 24% 28% 1 GB 26% 4
1999 TEN 2% 45% 51% 33% 16% 7 STL 46% 1
2000 BAL 77% -3% 91% 55% 30% 2 NYG 8% 11
2003 CAR 46% 30% 97% 58% 0% 17 NE 23% 3
2005 PIT 92% 40% 91% 74% 28% 3 SEA 26% 5
2006 IND 100% 67% 34% 67% 19% 7 CHI 24% 4
2007 NYG 55% 46% 48% 49% -1% 16 NE 52% 1

The Giants have had the most consistent run out of all seven teams, especially if we also consider their final regular-season game, where they were also above 40%. However, it has not been the most impressive playoff run out of the seven. Four teams had a higher average DVOA during the three games. The 2005 Steelers also had 40% in all three games (actually 39.7% against the Colts in the Divisional round) and the 2003 Panthers and 2006 Colts each topped 30% in all three games.

Four of these teams eventually won the Super Bowl, but you'll notice that three of those four teams actually had a higher regular-season DVOA than their opponents did, and the other one (the 2006 Colts) came pretty close. Longtime readers may be surprised to see the Steelers ranked third for 2005, but these numbers are based on the current version of DVOA created after that season, not the less accurate numbers we used in 2005 which had the Seahawks ahead of the Steelers.

The 1997 Broncos had the lowest average DVOA during their three playoff wins, but they are also the only wild card team to ever finish the season with the league's best DVOA. In both 1997 and 2000, the top two teams in the league were both in the same division: Denver and Kansas City in 1997, and Tennessee and Baltimore in 2000. Each time, the two teams played in the Divisional round. Denver won 14-10. Baltimore won 24-10 but, as you can see from that "-3%," DVOA thought Tennessee actually had the better game. The Titans had 23 first downs in that game, the Ravens only six. The Titans held the ball for over 40 minutes. You may remember that the Ravens scored one touchdown on a blocked field goal return, another on an interception return. Al Del Greco was 1-for-4 on field goals for the Titans. The AFC had the top four teams and seven of the top eight teams in DVOA that year, so whichever of those two teams won that game was going to be a heavy favorite against the Giants in the Super Bowl.

I want to point out that none of these articles regarding the Giants' performance during the regular season are meant to suggest that the Giants have no chance to beat New England. Of course the Giants have a chance. At least a 20 percent chance, maybe even a 40 percent chance. That's not what it means to pick the Giants. When somebody in the media like Dr. Z or Dan Patrick picks the Giants, it means that this person believes the Giants have a greater than 50 percent chance of winning this game. Given the historically unprecedented gap between the quality of these two teams in the regular season -- and the fact that the Giants' "hot playoff run" is actually not as hot as other recent "hot playoff runs" by teams like the 2003 Panthers, 2005 Steelers, and 2006 Colts -- that seems like a gigantic leap in logic.

Posted by: Aaron Schatz on 26 Jan 2008

54 comments, Last at 03 Feb 2008, 7:40pm by mush

Comments

1
by TheEvilBillGates (not verified) :: Sat, 01/26/2008 - 7:01pm

Jim Carrey from Dumb and Dumber: "So your saying there's a chance!"

A secretary from work, Boston Native living in California, summed it up best. "Whut does is matta who Bowston plays? They-a gonna beat 'em anyway."

Epitomized my hatred for this perfect team, that quite simply has to win. If they don't, their entire season will be remembered as the biggest waste of everyones time. All the pressure is on them.

As a Boston hater, I almost pity the idea of them losing this game. It will be the biggest choke-job ever if it happens, and the Pats will never, ever, be able to live it down.

2
by David (not verified) :: Sat, 01/26/2008 - 7:22pm

So would you stand by your claim that the Giants are the worst team to ever make it to the Superbowl?

Your own advanced statistics seem to indicate otherwise.

3
by cd6! (not verified) :: Sat, 01/26/2008 - 7:46pm

At least a 20 percent chance, maybe even a 40 percent chance. That’s not what it means to pick the Giants. When somebody in the media like Dr. Z or Dan Patrick picks the Giants, it means that this person believes the Giants have a greater than 50 percent chance of winning this game.

I'll put aside the hilarious "at least 20 and maybe 40 percent chance" to instead comment on the "they believe they have at least a 50 percent chance to win" claim. I think this paragraph goes too far in assuming that people operate like computers. If Dr. Z or Dan Patrick pick the Giants, all that means is that they think the Giants have a greater than zero chance to win.

4
by Arkaein (not verified) :: Sat, 01/26/2008 - 7:50pm

Re 2: David, the chart shows the Giants with a season DVOA of -1%, the worst on the list, so I'm not sure what you see that doesn't support the assertion that the Giants are the worst Superbowl team in at least the DVOA era.

5
by cd6! (not verified) :: Sat, 01/26/2008 - 7:52pm

Sorry for the double post, I wanted to also comment on this part:

Given the historically unprecedented gap between the quality of these two teams in the regular season — and the fact that the Giants’ “hot playoff run” is actually not as hot as other recent “hot playoff runs” by teams like the 2003 Panthers, 2005 Steelers, and 2006 Colts — that seems like a gigantic leap in logic.

The three teams you cited bolster the Giants' case. The Steelers and Colts teams both won the superbowl against teams with comparable or better regular season DVOAs, and the Panthers took a supposedly "much better" team to within a minute of overtime in the superbowl.

Saying that giving the Giants a chance is a taking a "gigantic leap in logic" is laughable.

6
by RickD (not verified) :: Sat, 01/26/2008 - 8:06pm

RE: 3

You really think that if Dr. Z picks the Giants to win, it's possible he secretly thinks that their chance to win is only non-zero? So you think that if he thought the Giants had a 1% chance of winning, he could pick them anyway?

Well, that's certainly a novel argument.

7
by David, FL (not verified) :: Sat, 01/26/2008 - 8:19pm

Re: 5

You are misinterpreting the comment. The article by Aaron does not suggest that giving the Giants a chance is taking a gigantic leap in logic. It states that giving the Giants a greater than 50% chance to win is taking a gigantic leap in logic. No one is suggesting the Giants don't have a chance.

8
by TheEvilBillGates (not verified) :: Sat, 01/26/2008 - 8:21pm

All information aside, each time has a 50% chance to win.

With that in mind, what's wrong with saying, "When somebody in the media like Dr. Z or Dan Patrick picks the Giants, it means that this person believes the Giants have a greater than 50 percent chance of winning this game"?

By picking a team, in any game, all your saying is their chance to win is greater than 50%.

This article concedes the Giants have a 20-40% chance, so the pick is Pats.

Just remember people, it doesn't matter if you win or lose, just if you covered the spread.

9
by PatsSF (not verified) :: Sat, 01/26/2008 - 8:33pm

RE: 6

You really think that if Dr. Z picks the Giants to win, it’s possible he secretly thinks that their chance to win is only non-zero? So you think that if he thought the Giants had a 1% chance of winning, he could pick them anyway?

Sure. He's got nothing to lose ("The Pats won and I was wrong? Well that's happened before.") and everything to gain ("The Pats won! I got pregame headlines and I told you so.")
This, of course, ignores the absurdity of proposing that any NFL team only has a 1% chance of winning a game: That whole "Any Given Sunday" thing.

10
by Paul (not verified) :: Sat, 01/26/2008 - 9:25pm

PIcking a team to win the Super Bowl is not the same as saying theirs odds to win are greater than 50%. It is similar to playing higher or lower with a deck of cards. If a 7 is showing, odds are better to pick higher, but sometimes, someone may go against the odds and pick lower. The odds are still under 50%, but this one time, they believe it's going to be lower. That's the same with picking the Giants. For this one game, Dr Z believes the Giants are going to win.

11
by Mystyc (not verified) :: Sat, 01/26/2008 - 9:35pm

I find it hard to believe in "Any Given Sunday" anymore with the Patriots. It's part of why I can't stand them - they've removed the drama from the game. I'm not even going to watch, because I know what the result will be.

12
by cd6! (not verified) :: Sat, 01/26/2008 - 9:56pm

6: "So you think that if he thought the Giants had a 1% chance of winning, he could pick them anyway?"

Absolutely.

If he thought they had zero chance to win, he obviously wouldn't pick them.

But if they have a chance, any chance, then he could.

If people only "picked" propositions that were 50% or better, Vegas would not exist.

8: "By picking a team, in any game, all your saying is their chance to win is greater than 50%."

Not true. It's perfectly possible you could think a team only has a 40% chance to win, but pick them anyway, because 40% isn't really that much of a disadvantage.

The superbowl is the one game most likely to be closer to a 50-50 probability split than any other game, I would suspect. The media, the hype, the neutral field, the 2 week layoff, the pressure, etc will have an effect of "evening the playing field" between two good, tested teams.

Last but not least, the pressure is more on the "Win or this season means nothing" Patriots vs. the "happy to be here, especially considering our QB is Eli" Giants. And as nobody is giving them a chance, the Giants get to play the unstoppable "Nobody respects us" card.

Of course, going by past behavior, it's not inconceivable that Brady or HGHarrison won't also attempt the "nobody respects us" card so maybe that one will cancel out.

13
by mush (not verified) :: Sat, 01/26/2008 - 10:00pm

When somebody in the media like Dr. Z or Dan Patrick picks the Giants, it means that this person believes the Giants have a greater than 50 percent chance of winning this game.

You're wrong on this one, Aaron. A lot of media members want to be "cute" with their predictions, earn "I told you so" rights, etc. If they were all making "gun to head" predictions we'd see very few NYG calls. Dr. Z's pick seemed to be at least partially inspired by regret he had for *almost* calling the Jets to win Super Bowl III.

14
by B (not verified) :: Sat, 01/26/2008 - 10:17pm

If Dr. Z or Dan Patrick think that the Giants chance of winning is less than .5, and they "pick" the Giants to win, they are in fact lying about who they think will win.

15
by Ian (not verified) :: Sat, 01/26/2008 - 11:15pm

The variable all you guys seem to be missing out on is this "Wilbon Factor". Or,

The kudos you get for choosing an underdog and them winning.

Every sports journalist does it occasionally. The cost associated with picking the Giants and them losing is almost zero. The kudos if you pick it and they win is fairly large.

16
by Richard (not verified) :: Sat, 01/26/2008 - 11:20pm

The three teams you cited bolster the Giants’ case. The Steelers and Colts teams both won the superbowl against teams with comparable or better regular season DVOAs, and the Panthers took a supposedly “much better” team to within a minute of overtime in the superbowl.

If you accept the validity of Aaron's metrics yet want to make this argument then you need to really take a step back and look at this in terms of causation and correlation. One of the first things they teach in many statistics class is the need to find a causal mechanism to support the evidence of correlation. You causal mechanism seems to be worse teams beat better teams.

17
by Richard (not verified) :: Sat, 01/26/2008 - 11:23pm

Re: #8

All information aside, each time has a 50% chance to win.

All information aside, we don't know what game we're playing, how many teams are playing ... With the most cosmic of possibilities in mind, 50% seems entirely too large!

All-but-minimal information aside, I don't think there's much of an argument to be made that this is truly a 50/50 affair.

18
by Richard (not verified) :: Sat, 01/26/2008 - 11:29pm

#15:

The cost associated with picking the Giants and them losing is almost zero.

I completely disagree. To me, it's the difference between being viewed as a real analyst and somebody like, say, Peter King.

I'm not saying a real analyst wouldn't pick the Giants. A real analyst, after valid analysis, might feel the Giants will win and pick them. This analysis would be articulated and to the reader/viewership.

Somebody like King would be more likely to be one of your pick-the-Giants-because-hell-why-not?-
I-have-a-feeling-and-Brett's-not-playing-
anymore-types.

19
by Jake (not verified) :: Sat, 01/26/2008 - 11:49pm

Re 5: No they do not. First, they're not that comparable and the teams they faced are not that comparable to the Patriots. Only in 03 was there any real gap.
But, there's a fundamental conceptual error in your thinking. We're discussing the probability of a Giants win. If they lose, the probability of winning was not zero. If they win, it was not 100%. Just because I flip a coin 5 times and it comes up heads 5 times, it doesn't mean that was more probable than other outcomes that did not occur. And just because a past Super Bowl with some passing similarity to the upcoming contest almost resulted in an upset, this does not mean the probability of a NE win was less than expected in that game or in the upcoming game.
Unless there's information indicating that DVOA is a poor predictor of which of the two teams will actually win the game when applied to the Super Bowl, the more important thing to check would be games with a 50% DVOA gap (or 40% in weighted DVOA).

20
by Alex (not verified) :: Sun, 01/27/2008 - 12:48am

It’s perfectly possible you could think a team only has a 40% chance to win, but pick them anyway, because 40% isn’t really that much of a disadvantage.

But if you only need to think that a team has a 40% chance to win in order to pick them, then you could pick both the Giants and the Patriots. When sportswriters make predictions about games, they choose which team they think (given all the specific matchups, emotional factors, "disrespect" wildcards, etc) is more likely to win the game.

If they really think team A isn't more likely to win than team B, and they still pick team A to win, then they're being misleading about their beliefs regarding the game, either to get publicity in case of an upset, or because they just wish that team A had a better shot, or maybe just for the hell of it. But if they're being honest about things, then picking the Giants means they think the Giants have a better chance of winning this particular game than the Patriots.

21
by Quentin (not verified) :: Sun, 01/27/2008 - 10:22am

I'll be a tad more nuanced. If you truly believe that a team is going to win, then you are giving them better than 50% chance to succeed. The less certain you are, then the closer it is to 50%. But you cannot actually think a team is going to win if you're only 40% certain. In that situation, you might be trying to convince yourself that they'll win, but inside you feel otherwise.

22
by mikeabbott (not verified) :: Sun, 01/27/2008 - 11:02am

RE 13,15: I strongly agree.
Who can remember which analyst picked what in the past?
I suspect few do.
If an analyst picks a long shot and hits it they can work it into future columns, if they miss it will be forgotten.

23
by Chris M (not verified) :: Sun, 01/27/2008 - 12:03pm

I think Giants fans are forgetting that the mediocrity of this Giants team in the regular season is precisely what has made this playoff run so much fun.

This was a team that looked a lot like the last two years' models - above average, but often cover-your-eyes level awful. Lest we forget:
-the team quitting against the Packers in week 2
-the parade of false starts against Dallas
-the four picks against Minnesota
-18 for 52 against Washington

And those are just the losses. They looked dead against the Bills for a quarter and a half before turning it on. They struggled mightily against the Eagles and the Bears. This was not a great team.

But they've turned it around the last. Of course, every one of us Giants fans has been waiting for Eli to throw the killer pick for three weeks now and is eminently relieved when each game ends and he hasn't. I don't think you can be a great team when your fan base spends 3 hours a week in deep prayer that your QB doesn't play like he did six weeks ago.

Are they favorites against the Pats? Of course not. Are they historically weak in terms of Super Bowl teams? Yeah, although as Aaron points out, sometimes historically weak teams are able to ride their hot streak (see last year's Colts - but remember they were playing the Rex Grossman bears and not this Pats team). This Giants team are monstrous underdogs - and that's okay. It will only make it sweeter if they win.

24
by Chris M (not verified) :: Sun, 01/27/2008 - 12:05pm

Oh, for clarity, after rereading the above post - I'm a Giants fan. I've watched basically every snap this year. Thinking about the Minnesota game still gives me the shakes.

25
by jimm (not verified) :: Sun, 01/27/2008 - 12:59pm

I think the fan in Aaron is seeking comfort in the stats. I think Aaron the fan is worried about this game because as the season wore on the Pats looked less dominant and in the final game of the season and the playoffs the Giants looked like a different and far better team.

These teams played a game 4 weeks ago that meant far more to NE than it did the Giants, and lets face it the Giants could have easily won that game.

I think this will be a very close game and it wouldn't surprise me in the least if the Giants won.

26
by B (not verified) :: Sun, 01/27/2008 - 1:46pm

25: The Giants could easily have won that game? I'm not sure what you're getting at, because when I watched that game, it was even for the first half, then in the second half the Patriots took over, and the game was no longer in doubt by the 4th quarter. The Giants were not the better team in that game.

27
by Temo (not verified) :: Sun, 01/27/2008 - 2:14pm

So many of you are being absolutely irrational in what you say about this game, I don't understand why. This isn't Colts/Pats even!

28
by Aaron (not verified) :: Sun, 01/27/2008 - 2:41pm

Wow... For people that frequent a math geek site (yes, this is a geek site), most of you don't seem to grasp the simple win/loss probabilities.

As long as the Giants win probability is nonzero, they have a chance to win. So let's say it is 5%. That means 5 out of 100 times the game is played, the Giants will win. The fact that these media people pick the Giants reflects solely that they think one of these 5 games will happen.

The other side is that they could believe the Giants are the better team (ha!), or that the week 17 game exposed the Patriots for the Giants and this time Eli & Co. will finish the job. so they think they have better than a 50% chance to win.

But here's the kicker, both of these scenarios are the same when it comes to picking a SINGLE game. It comes down to who you think WILL win, not who will win the most times out of 100. The comments that do not reflect this are just ignorant.

29
by David Mazzotta (not verified) :: Sun, 01/27/2008 - 3:22pm

#12 If people only “picked” propositions that were 50% or better, Vegas would not exist.

Not true. People only pick sub 50% props if a larger payoff justifies it over a >50% prop (or if they are insane).

There is no payoff for Dr. Z's prediction. And I am fairly certain he is not clinically insane. He honestly believes that the Giants are more likely (>50%) to win.

30
by Temo (not verified) :: Sun, 01/27/2008 - 3:29pm

28. See, that's exactly what I'm talking about. If a team has a probability of winning, how can you know exactly what will transpire? You cannot rationally say "I believe the Patriots have a greater chance to win, but I think the Giants win."
That is akin to betting on anything OTHER than a 7 on the roll of two die (most likely outcome is 7). Doing anything else is irrational.
So unless Dr. Z or whoever is irrational, picking the Giants to win is the same as saying they have a greater than 50% chance of winning.

31
by TheEvilBillGates (not verified) :: Sun, 01/27/2008 - 3:39pm

If you pick a team, and believe they have less than a 50% chance to win, I believe you are being disingenuous. I respect Paul Zimmerman as a journalist, and believe he has integrity. I believe his pick of the Giants has little to do with "bragging rights," and is more likely to be his true pick. But hey, maybe I'm a sucker and Dr. Z is taking me on the proverbial "ride."

Also, when did the writers of this site become so sensitive? A whole post about what is acceptable, and a sticky here? I hate to think what will happen if the Pats lose. Are we not going to be able to call Adam out? (Of coarse we wont, this site never predicts a winner, only which has a statistical advantage).

32
by cd6! (not verified) :: Sun, 01/27/2008 - 4:19pm

re: Dr. Z-
A lot of people seem to suggest that if Dr. Z thinks the Pats have a better chance to win, he must pick the pats. That would only make sense if his job (and other pundits) were to forecast who they think has the better chance to win. But either team can win any game assuming both teams have a nonzero chance to win, so he can legitmately predict either team to win, since either team does have A CHANCE.

I'll second what somebody said about "If he picks the underdog and they win, kudos, if they lose, well oh well, they're the underdog." Pundits are paid to incite debate among fans, there's no real incentive for him to pick the favorite.

re DVOA
Unless there’s information indicating that DVOA is a poor predictor of which of the two teams will actually win the game when applied to the Super Bowl
You're looking at it the wrong way. There's no information indicating that DVOA is a good predictor of winning games. It's a statistical analysis of past performance, not a forecasting device. This fact is consistently cited when a lower DVOA team beats a higher ranked DVOA team, though it's also credited as "DVOA was right after all" when a better team beats a weaker team. Hmm. ;)

33
by Temo (not verified) :: Sun, 01/27/2008 - 5:06pm

32. But you're not getting to the point of this matter: what is the virtue of picking the team with the lower possibility of winning? If you roll dice and say the sum of the two will be 2, and it happens to roll out to 2, should be commended for predicting that?
If you pick a stock with a negative expectation over a stock with a positive expectation, and the former happens to do better, should you be commended for that?
If you believe that one team has a lower than 50% chance of winning a specific game and you pick that team to win, you are being irrational. The only rational explanation for picking a team to win is that you believe they have a better chance to win than the other team, otherwise you're might as well be throwing darts at a board.

34
by Eddo (not verified) :: Sun, 01/27/2008 - 6:24pm

re: 32 (cd6!)
You’re looking at it the wrong way. There’s no information indicating that DVOA is a good predictor of winning games. It’s a statistical analysis of past performance, not a forecasting device. This fact is consistently cited when a lower DVOA team beats a higher ranked DVOA team, though it’s also credited as “DVOA was right after all” when a better team beats a weaker team. Hmm. ;)
I don't mean to nitpick your point, because it's a good one. You're dead-on that DVOA is not good at predicting individual game outcomes (or at least no better than a wide variety of measures) due to the inherent luck and variability of a single football game). But when someone says "DVOA was right after all," it's usually in regards to something like this year's Buccaneers or Packers teams; that is, DVOA is good at predicting broader results. It is therefore a very useful tool to show you that a 3-4 team with a 10% DVOA will end up with a better record than a 5-2 team with a -5% DVOA.

35
by MarkM (not verified) :: Sun, 01/27/2008 - 11:47pm

I like it when DVOA gets "tweaked" to show that Pittsburg was in fact ranked ahead of Seattle in 2005.

36
by justme (not verified) :: Mon, 01/28/2008 - 1:04am

35. You say it like something is wrong with that. Aaron isn't trying to pull one over your head, he's just adjusting the formula to fit the data better as more data becomes available.

I feel like FO is having a midlife crisis right now. Is DVOA is really useful, or worth the time to try and figure out odds, or any more valuable than other irrational sites or writers :)

37
by starzero (not verified) :: Mon, 01/28/2008 - 2:42am

so these numbers are rounded, but i'm still trying to figure out how the titans won a game with only 2% dvoa, and how the colts put up 100% dvoa against the chiefs. i don't remember them being so dominant. also, it'd be interesting to know what the sb teams' dvoas were for that game.

38
by Duff Soviet Union (not verified) :: Mon, 01/28/2008 - 4:02am

"so these numbers are rounded, but i’m still trying to figure out how the titans won a game with only 2% dvoa, and how the colts put up 100% dvoa against the chiefs. i don’t remember them being so dominant." Wasn't that 2% DVOA game the Music City Miracle? A lineball game sounds about right to me. As for that Colts/Chiefs games, KC didn't even get a first down in the first half IIRC. Manning threw two or three picks but if not for that it would have been a total massacre.

39
by DolFan 316 (not verified) :: Mon, 01/28/2008 - 8:05am

#31: If the Pats should somehow lose, this site will either temporarily shut down completely for a week or three, or people will get banned from here forevermore if they so much as mention the game or the Pats in general. A Pats loss can produce no other outcome. You can already feel the testiness, for lack of a better word, it's palpable on every thread.

40
by Purds (not verified) :: Mon, 01/28/2008 - 8:43am

#39:

If the Pat lose, we can be sure that we won't have another link to the idiot Bill Simmons that week, because like last year after the AFCC, the "national" columnist will be too childish to write anything.

41
by Yardape (not verified) :: Mon, 01/28/2008 - 9:03am

Simmons wrote a column after the AFCC last year. It was something about Manning taking away from the Pats. It was after the Super Bowl that Simmons went into his snit and wouldn't write a column.

42
by TGT (not verified) :: Mon, 01/28/2008 - 10:51am

@28 You're bastardizing statistics. If in reality, the giants have a 5% chance to win the game, but someone thinks that one of those 5 in a 100 games is going to occur, that person does not believe the team has a 5% chance to win. That person believes the team has a much higher percent chance to win.

@35 Wasn't DVOA tweaked because of how the colts ridiculously weak schedule affected DVOA that year? The resultant DVOA was more accurate. Pittsburgh over Seattle was not the thrust of the tweakings.

@5 Since when are 26% and 24% "comparable or better" than 52%? I think you need to reread the charts.

43
by zip (not verified) :: Mon, 01/28/2008 - 11:53am

As long as the Giants win probability is nonzero, they have a chance to win. So let’s say it is 5%. That means 5 out of 100 times the game is played, the Giants will win. The fact that these media people pick the Giants reflects solely that they think one of these 5 games will happen.

It kills me when people chide others for not "grasping probabilities" and then trot out crap like this.

44
by Herm? (not verified) :: Mon, 01/28/2008 - 12:13pm

Is the -1% regular season average bumping up to the 49% postseason average not the most astonishing/ inexplicable statistic at this site this year?
Isn't this why we're here? Isn't this why people should watch the game this Sunday?
I'm watching for that reason.

If some of you want the Patriots to lose because you hate a woman's accent, well that's your choice, but be prepared for disappointment. No amount of Giant winning is going to change her. And I'm not sure you'll want to replace her with a Giant fan, either.

45
by weaponx (not verified) :: Mon, 01/28/2008 - 12:43pm

What a lot of folks are overlooking what actually has to happen in those 5 chances out of 100. 1 is a large chunk of space debris (manufactured or natural) falling on the Patriots sideline in the first quarter. Another is when 4 (unnamed at this time) members of the NE offense turn heel on a 3rd qtr drive and throw a beatdown on Brady before pulling their jerseys off, revealing NWO Tshirts. The others are just /too/ out there, involving mass fan participation with gazoos and or silly string.

Don't get your hopes up Hateriots, Gmen fans, and members of that undefeated team.

46
by Zac (not verified) :: Mon, 01/28/2008 - 12:47pm

I think the answer is that when the Giants lose, Dr. Z and Dan Patrick need to be reminded of it at every opportunity. The only way to get these pundits to stop making stupid predictions is to call them on it.

47
by Herm? (not verified) :: Mon, 01/28/2008 - 1:11pm

Speaking more realistically, don't you mean that the defensive calls would have to be correct more often than not?...and when wrong the defensive players have to make up for it by stopping the run or pass while outnumbered? Wouldn't the pass rush have to be effective, while the defensive backs have to cover for just long enough for the complementing pass rush / coverage to slow down or stop the passing game?
Won't the offense have to control the ball with a mixed run/pass ratio on every down? and not turn it over or commit bad penalties? This almost worked in week 17...And if this all adds up to a 40% DVOA or better, doesn't this mean that the Giants at least have a chance to win the game?
Besides the -1% jump to 49%, doesn't the consistency of 55%, 46%, 48% against other playoff teams wake you up a little?
If you're holding Dr. Z (I'm pretty sure he is not a real doctor) and Dan Patrick to a prediction their editor told them to make, then perhaps it's time to switch to decaf, and maybe check for sticks in places they shouldn't be.

48
by pete (not verified) :: Mon, 01/28/2008 - 4:08pm

I like the analysis, because i love recent NFL history, but i think this data really helps support the notion the Giants are going to get creamed in this game.

Of the four that won the superbowl, 3 of the 4 came from conferences that had to be considered the best conference of the season. I do not recall anyone thinking in 2000, 05 or 06 that the NFC had any real shot of winning the super bowl. In all three cases, sports pundits called earlier playoff games the de facto super bowl.

the evidence really goes to support what Aaron said two weeks ago. this team really is like the 03 Panthers. They got hot against stronger competition. Carolina put up a remarkable fight against NE.

49
by Alex (not verified) :: Mon, 01/28/2008 - 5:26pm

I love the whole "you don't have to think the team has to have a probability of winning higher than 50% to pick them to win" arguments. I mean, seriously, how would that sound on a pregame show?

"So, John, who do think wins the Super Bowl this year, Giants or Patriots?"

"I'm going to go with the Giants, Mike."

"Really? You think they're going to pull the upset?"

"Yes. They've been on an impressive hot streak, and they just beat some of the best teams in the league on their way to the big game."

"So, just how confident are you in that prediction, John?"

"I'm glad you asked. I'm not very confident at all, Mike. In fact, I'd say the Giants only have about a 5% chance of winning the game, given the dominant Patriots team they're facing and their own struggles throughout the regular season."

"Wait, you only think they've got a 5% chance of winning, but you still think they're going to win?!?"

"Exactly. I'm just going with my gut here, and it's telling me that the Giants are going to beat the odds this time. It's also telling me to get some turducken."

50
by GlennW (not verified) :: Mon, 01/28/2008 - 5:55pm

> Not true. People only pick sub 50% props if a larger payoff justifies it over a >50% prop (or if they are insane).

Factoring in vigorish with pointspreads and Vegas's standard practice of shorting both sides of a money line as well as longer-term proposition odds, the odds of winning (in all formats) are less than 50% relative to payout. People still wager. People know going in that the odds of winning at blackjack are 48% but they'll still play the game fully expecting that they'll come out ahead against those odds. Human beings are funny that way.

As such it seems completely reasonable to me that someone like Dr Z can say (also somewhat irrationally), hey, I know that if they played this game 5 times, the Patriots would win it 4 out of those 5, but I'm still going to pick the upset for the one time the game will actually be played. And as others have suggested, his payout comes in the form of receipt of a correspondingly larger number of kudos for intentionally picking a major upset, as opposed to picking the Patriots just like everyone else.

The actual money line on the Super Bowl is around a solid +400/-500 on NYG/NE respectively (about 4 to 1 for the Giants), and I can see a prognosticator (similar to an actual bettor) thinking to himself, hey, I think the Giants really do have a better shot than that 20%, so I'm going to make that bold prediction...

51
by Oswlek (not verified) :: Mon, 01/28/2008 - 7:12pm

jmm,

I vehemently disagree. The Giants are clearly a much better team than they were for much of the season.

However, NE was a much better team on that day. NE scored on 7 of 9 possession while holding NY scoreless on 5 of 9 possessions. NE dominated the TOP and their shortest drive of the first half was equal to NY's longest. NE held Jacobs to 7 yards rushing in the entire first half.

Rewatching the NY/GB game, I was surprised at just how poorly both teams played. Of course the weather affected things, but if NY plays like that again they would lose to just about every AFC playoff team.

NE played a mediocre game, but it was almost entirely Brady. The OL played fine in the first half and was dominant in the second. NE's RBs were excellent. NE's defense proved that they can win a game against a good team even when the offense has an off day.

The last one is particularly enlightening because it is now known that Brady was both injured and sick going into that game. in the latter part of the year, NE purposely took away the big play to force teams into long scoring drives. They did this knowing that a mistake would ultimately be made by a team playing out of their comfort zone due to keeping up with NE's offense. Against SD, they played a little differently and leaned on the D more, and it worked.

52
by Alex (not verified) :: Mon, 01/28/2008 - 7:58pm

it seems completely reasonable to me that someone like Dr Z can say (also somewhat irrationally), hey, I know that if they played this game 5 times, the Patriots would win it 4 out of those 5, but I’m still going to pick the upset for the one time the game will actually be played.

But if someone thinks that the Giants will win the game that is about to be played, then they don't think the Giants have a 20% chance of winning that particular game, they think (somewhat irrationally) that the Giants have a 100% chance of winning that game.

They might also (perhaps irrationally) think that the Giants would only win 1 out of every 5 games against the Patriots if they played the Patriots many times, but then we're not talking about the probability of winning the Super Bowl, we're talking about the probability of winning a bunch of other hypothetical games.

As an analogy, if I had a gun that only had 1 bullet left in it, and I pulled the trigger, I'd think that the probability of the gun firing would be somewhere around 99%. But if I pulled the trigger 100 times, then I'd think that the probability of the gun firing on any given trigger pull would be about 1%. But that doesn't mean that I think the probability of the gun firing on the first trigger pull is 1%, just like Dr. Z (presumably) doesn't think that the probability of the Giants winning the Super Bowl is below 50%. He might think that, if the Giants played the Patriots many more times, the Patriots would win most of the games, while also thinking that there is something special about this particular game that makes the Giants more likely to win.

53
by Richie (not verified) :: Mon, 01/28/2008 - 9:26pm

When an "expert" predicts the winner of a game, the IMPLICATION is that they think the team has a greater than 50% chance of winning the game.

Of course, most experts have nothing at stake in their picks. Instead, their only gain can be by picking the underdog. If they pick the favorite and the favorite wins, then nobody cares. If they pick the underdog and the underdog loses, then people think he was just making a crazy pick.

BUT, if he picks the underdog and the underdog wins, then people will think "Wow, how did he know that?"

54
by mush (not verified) :: Sun, 02/03/2008 - 7:40pm

Bob Costas admitted on this week's Inside the NFL that he's picking the Giants for the style-and-credit points. He was up front about it, I'll say that. I'm sure he's not alone in this theme.