QB Choke Index (1 Viewer)

[Doug Drinen]

Buried somewhere in one of the Manning-is-a-choker threads, I posted that the Colts in the Manning era have exactly the playoff record you would expect based on their seedings. They have been the higher seed 3 out of 8 times, and they've one 3 of 8 games. Even Steven.Anyway, I thought it might be fun to run those numbers for all QBs with 5 or more playoff games and see whose record exceeds expectations by the most.Here's the list. Fine print is below:EDIT: MarshallRob caught an error in the original. This is a corrected list.

Expected  Actual                          wins     wins     Diff-----------------------------------------------Trent Dilfer              2.5       5      +2.5Tom Brady                 5.0       7      +2.0Mark Rypien               3.0       5      +2.0Jim Everett               0.0       2      +2.0John Elway               12.0      14      +2.0Wade Wilson               1.5       3      +1.5Joe Montana              14.5      16      +1.5Jeff Hostetler            2.5       4      +1.5Drew Bledsoe              3.0       4      +1.0Jay Schroeder             2.0       3      +1.0Mark Brunell              2.0       3      +1.0Brad Johnson              3.0       4      +1.0Dave Krieg                2.0       3      +1.0Steve McNair              4.0       5      +1.0Jim Harbaugh              1.0       2      +1.0Doug Williams             3.5       4      +0.5Troy Aikman              10.5      11      +0.5Stan Humphries            3.0       3      +0.0Vinny Testaverde          2.0       2      +0.0Brett Favre              11.0      11      +0.0Phil Simms                6.5       6      -0.5Peyton Manning            3.5       3      -0.5Kerry Collins             3.5       3      -0.5Kurt Warner               5.5       5      -0.5Rich Gannon               5.0       4      -1.0Mike Tomczak              4.0       3      -1.0Randall Cunningham        4.0       3      -1.0Donovan McNabb            7.0       6      -1.0Bernie Kosar              4.0       3      -1.0Jim Kelly                10.0       9      -1.0Kordell Stewart           3.5       2      -1.5Dan Marino               10.0       8      -2.0Warren Moon               5.0       3      -2.0Jim McMahon               5.0       3      -2.0Neil O'Donnell            5.0       3      -2.0Steve Young              11.5       8      -3.5

Fine print:1. If a team is at home and has the better record, they are given an "expected win"2. If the road team has the better record (which is very rare), each team is given half an expected win (this is where Manning's -.5 comes from. Against the Jets in 02, they were the lower seed but had the better record).3. In the Super Bowl, the team with the better record is given an expected win.4. In the Super Bowl, if the two teams have the same record, they are each given half an expected win.5. A QB is credited with a game if he threw 10 or more passes. In games where two or more QBs threw 10+ passes (e.g. Brady/Bledsoe 2001 vs Pitt), I made no effort to sort out who gets the credit/blame. It counts for both of them. This is a rote database query with no human intervention. (Maybe you homers can alert me to instances where adjustments might be warranted.) 6. A QB is credited with an "actual win" if his team won the game.7. The list contains all QBs whose career started in 1978 or later and played in 5 or more playoff games.I am well aware that my algorithm for determining expected wins is pretty crude. If I find the time, I'll cook up a fancy-schmancy formula that does a better job of computing expected wins based on the team records, the bye (or lack thereof), and home field.I really and truly have no point to make here, just a quick query that I thought was of interest.

 

[Jason Wood]

Fascinating approach, particularly because it cements two of my own personally held beliefs...that Montana is the best ever, and that Marino doesn't deserve to be held among the very highest pantheon of NFL QBs.

 

[Family Matters]

Great post Doug. A couple of names I wasn't expecting in the even to plus catagory where:RypienDilferEverettWilsonKriegAnother thing that caught my eye. 4 QB's totaling +4.5 were coached by Joe Gibbs. Those QB's were Rypien, Williams, Schroeder & Hunphries. When add in Theissman that nakes 5 QB's. Not sure if Joe is a + or - but I would guess a +.

 

[Mr. Vegas]

Buried somewhere in one of the Manning-is-a-choker threads, I posted that the Colts in the Manning era have exactly the playoff record you would expect based on their seedings. They have been the higher seed 3 out of 8 times, and they've one 3 of 8 games. Even Steven.Anyway, I thought it might be fun to run those numbers for all QBs with 5 or more playoff games and see whose record exceeds expectations by the most.Here's the list. Fine print is below:

                       Expected  ActualQuarterback              wins     wins     Diff-----------------------------------------------Joe Montana              11.5      16      +4.5Mark Rypien               2.0       5      +3.0Troy Aikman               8.5      11      +2.5Trent Dilfer              2.5       5      +2.5Jim Everett               0.0       2      +2.0Tom Brady                 5.0       7      +2.0Wade Wilson               1.5       3      +1.5Doug Williams             2.5       4      +1.5Jeff Hostetler            2.5       4      +1.5Mark Brunell              2.0       3      +1.0Jay Schroeder             2.0       3      +1.0Brett Favre              10.0      11      +1.0Brad Johnson              3.0       4      +1.0Dave Krieg                2.0       3      +1.0Steve McNair              4.0       5      +1.0Jim Harbaugh              1.0       2      +1.0Drew Bledsoe              4.0       4      +0.0John Elway               14.0      14      +0.0Vinny Testaverde          2.0       2      +0.0Kerry Collins             3.5       3      -0.5Kurt Warner               5.5       5      -0.5Phil Simms                6.5       6      -0.5Peyton Manning            3.5       3      -0.5Mike Tomczak              4.0       3      -1.0Jim McMahon               4.0       3      -1.0Stan Humphries            4.0       3      -1.0Randall Cunningham        4.0       3      -1.0Bernie Kosar              4.0       3      -1.0Rich Gannon               5.0       4      -1.0Donovan McNabb            7.0       6      -1.0Kordell Stewart           3.5       2      -1.5Warren Moon               5.0       3      -2.0Jim Kelly                11.0       9      -2.0Steve Young              10.5       8      -2.5Neil O'Donnell            6.0       3      -3.0Dan Marino               11.0       8      -3.0

Fine print:1. If a team is at home and has the better record, they are given an "expected win"2. If the road team has the better record (which is very rare), each team is given half an expected win (this is where Manning's -.5 comes from. Against the Jets in 02, they were the lower seed but had the better record).3. In the Super Bowl, the team with the better record is given an expected win.4. In the Super Bowl, if the two teams have the same record, they are each given half an expected win.5. A QB is credited with a game if he threw 10 or more passes. In games where two or more QBs threw 10+ passes (e.g. Brady/Bledsoe 2001 vs Pitt), I made no effort to sort out who gets the credit/blame. It counts for both of them. This is a rote database query with no human intervention. (Maybe you homers can alert me to instances where adjustments might be warranted.) 6. A QB is credited with an "actual win" if his team won the game.7. The list contains all QBs whose career started in 1978 or later and played in 5 or more playoff games.I am well aware that my algorithm for determining expected wins is pretty crude. If I find the time, I'll cook up a fancy-schmancy formula that does a better job of computing expected wins based on the team records, the bye (or lack thereof), and home field.I really and truly have no point to make here, just a quick query that I thought was of interest.

So Mark Rypien is no Joe but He's better then anyone else.Yes this chart works. :rotflmao:

 

[cobalt_27]

Buried somewhere in one of the Manning-is-a-choker threads, I posted that the Colts in the Manning era have exactly the playoff record you would expect based on their seedings. They have been the higher seed 3 out of 8 times, and they've one 3 of 8 games. Even Steven.Anyway, I thought it might be fun to run those numbers for all QBs with 5 or more playoff games and see whose record exceeds expectations by the most.Here's the list. Fine print is below:

Expected  ActualQuarterback              wins     wins     Diff-----------------------------------------------Joe Montana              11.5      16      +4.5Mark Rypien               2.0       5      +3.0Troy Aikman               8.5      11      +2.5Trent Dilfer              2.5       5      +2.5Jim Everett               0.0       2      +2.0Tom Brady                 5.0       7      +2.0Wade Wilson               1.5       3      +1.5Doug Williams             2.5       4      +1.5Jeff Hostetler            2.5       4      +1.5Mark Brunell              2.0       3      +1.0Jay Schroeder             2.0       3      +1.0Brett Favre              10.0      11      +1.0Brad Johnson              3.0       4      +1.0Dave Krieg                2.0       3      +1.0Steve McNair              4.0       5      +1.0Jim Harbaugh              1.0       2      +1.0Drew Bledsoe              4.0       4      +0.0John Elway               14.0      14      +0.0Vinny Testaverde          2.0       2      +0.0Kerry Collins             3.5       3      -0.5Kurt Warner               5.5       5      -0.5Phil Simms                6.5       6      -0.5Peyton Manning            3.5       3      -0.5Mike Tomczak              4.0       3      -1.0Jim McMahon               4.0       3      -1.0Stan Humphries            4.0       3      -1.0Randall Cunningham        4.0       3      -1.0Bernie Kosar              4.0       3      -1.0Rich Gannon               5.0       4      -1.0Donovan McNabb            7.0       6      -1.0Kordell Stewart           3.5       2      -1.5Warren Moon               5.0       3      -2.0Jim Kelly                11.0       9      -2.0Steve Young              10.5       8      -2.5Neil O'Donnell            6.0       3      -3.0Dan Marino               11.0       8      -3.0

Fine print:1. If a team is at home and has the better record, they are given an "expected win"2. If the road team has the better record (which is very rare), each team is given half an expected win (this is where Manning's -.5 comes from. Against the Jets in 02, they were the lower seed but had the better record).3. In the Super Bowl, the team with the better record is given an expected win.4. In the Super Bowl, if the two teams have the same record, they are each given half an expected win.5. A QB is credited with a game if he threw 10 or more passes. In games where two or more QBs threw 10+ passes (e.g. Brady/Bledsoe 2001 vs Pitt), I made no effort to sort out who gets the credit/blame. It counts for both of them. This is a rote database query with no human intervention. (Maybe you homers can alert me to instances where adjustments might be warranted.) 6. A QB is credited with an "actual win" if his team won the game.7. The list contains all QBs whose career started in 1978 or later and played in 5 or more playoff games.I am well aware that my algorithm for determining expected wins is pretty crude. If I find the time, I'll cook up a fancy-schmancy formula that does a better job of computing expected wins based on the team records, the bye (or lack thereof), and home field.I really and truly have no point to make here, just a quick query that I thought was of interest.

So Mark Rypien is no Joe but He's better then anyone else.Yes this chart works. :rotflmao:

I guess this means we have to put Rypien, Dilfer, Hostetler, Chris Everett and Wade Wilson in the pantheon of great QBs, then.

 

[bostonfred]

How about a list of the worst QB ratings in a playoff game vs. regular season QB rating? I'd guess that Manning has two, maybe three of the twenty worst differentials in playoff history.

 

[KnowledgeReignsSupreme]

I'd rather see this expressed as a ratio of actual wins:expected wins, than a differential. Kinda like TD:INT ratio. That would make Mark Rypien the best ever with a 2.5:1 ratio. I can live with that.

 

[boubucarow]

To me, I see Aikman on the plus side and Young on the minus side. I am sure you remember the monkey of off Young's back routine on the sidelines during the Charger super bowl. Both are viewed as sure hall of famers, from what I hear. But also I have heard Aikman questioned for the obvious reasons (#22, #88, OL, no stats).But I am thinking Aikman before Young because of 3 to 1. As in Super Bowls. As in the number of times Aikman outplayed Young in the playoffs. Young self-destructed the first two games. I think playoff performance is often too overlooked for such "caretaker" quarterbacks.

 

[BGP]

I really and truly have no point to make here, just a quick query that I thought was of interest.

This is where I am at with this data.I truly believe wins and losses should be a team stat. Maybe a coaching stat, given certain conditions.A comparison between one player and wins and losses is , while interesting, not something I put much stock in. Especially when it tells me the 5 best include Rypien, Dilfer and Everett.

 

[Cocky Crow]

Doug, why don't you use the spread to determine who should have the "expected win". Even though it is probably most often the same, using records and home field data can be misleading when determining who is "expected to win". Case in point is this week's game between the Pats & Steelers. Using your method, you will flag it as an expected win for Pitt (better record, home field) despite the fact that Vegas, and really thus the nation, feel that the Pats are clearly the favorite based on the spread (expected to win).I am guessing, perhaps, that you don't have this statistic going back in time, no? Anyway, just my thoughts, and very interesting data none the less.

 

[valhallan]

How about a list of the worst QB ratings in a playoff game vs. regular season QB rating? I'd guess that Manning has two, maybe three of the twenty worst differentials in playoff history.

 

[BassNBrew]

How about a list of the worst QB ratings in a playoff game vs. regular season QB rating? I'd guess that Manning has two, maybe three of the twenty worst differentials in playoff history.

And Delhomme might have the best.

 

[GregR_2]

Hey Doug, interesting take on it. One other stat I'd like to see in there is to see how many expected losses they had too. For instance, if Montana has a +4.5 when there were only 5 games he was expected to lose, and Aikman has +2.5 when he had 7 games he was expected to lose, it makes Montana's accomplishment even more meaningful.But on the other hand, if Montana got that +4.5 from 15 games he was expected to lose, and Aikman only had 3 games he was expected to lose, it would be a big bonus for Aikman in considerations like this.

 

[Kilgore Trout]

Here's an article on ESPN's page 2 about the Best Big Game QB's. Its more subjective, but it is interesting none the less.

If NE wins the next two games than you have to start making the Brady/Montana comparisons as far as post-season career status.

 

[MarshallRob]

Fascinating approach, particularly because it cements two of my own personally held beliefs...that Montana is the best ever, and that Marino doesn't deserve to be held among the very highest pantheon of NFL QBs.

I don't think the numbers for Marino are accurate. I count only nine expected wins, not eleven. Two in 1984 (two wins) and 1985 (a win and a loss), and one each in 1983 (loss), 1990 (win), 1992 (win), 1994 (win), and 1995 (win). Seven of his 10 playoff losses were on the road plus the Super Bowl loss which should be considered a road game since it was at Stanford. In the 1992 AFC title game Miami hosted the Bills and lost but both teams had the same record so this is not an "expected win" under the stated criteria. If it is, Marino should get credit for an "unexpected win" in 1999 on the road against Seattle where both teams had the same record.

 

[Doug Drinen]

I don't think the numbers for Marino are accurate.

You're right, there's something fishy.I'll fix up the whole list (if there are other problems) when I get home from work tonight. Sorry.

 

[Doug Drinen]

How about a list of the worst QB ratings in a playoff game vs. regular season QB rating? I'd guess that Manning has two, maybe three of the twenty worst differentials in playoff history.

How about QB rating is 100% garbage? But I can do that for some other reasonable passing stat, like say yards per attempt (with an INT penalty). Check back tonight.

 

[bostonfred]

How about a list of the worst QB ratings in a playoff game vs. regular season QB rating?  I'd guess that Manning has two, maybe three of the twenty worst differentials in playoff history.

How about QB rating is 100% garbage? But I can do that for some other reasonable passing stat, like say yards per attempt (with an INT penalty). Check back tonight.

I'm not a huge fan of QB rating either, but I didn't know what stats you'd have available to you to quantify the dropoff.My case for Manning "choking" has never been that he went into those five playoff losses and gave his all but came up just short. It's been that he fell so far off his personal standard when the team needed him most.

 

[gump]

Jim Kelly, Steve Young, and Dan Marino in the bottom 4??

 

[NE_REVIVAL]

How about a list of the worst QB ratings in a playoff game vs. regular season QB rating?  I'd guess that Manning has two, maybe three of the twenty worst differentials in playoff history.

How about QB rating is 100% garbage?

You cook up a chart that indicates the likes of Mark Rypien, Trent Dilfer, Jim Everett, Wade Wilson, Doug Williams, Mark Brunell, Jay Schroeder were somehow better playoff qbs than Favre and Young and then have the CHUTZPAH to declare the passer rating is garbage I guess this was done in an effort to prop up Manning and show he really isn't that bad in the playoffs. I don't think Manning is a choker but I also don't think he improvises or handles adversity as well as many other qb's who are considered great.

 

[beefcake]

Where's Terry Bradshaw?

 

[NE_REVIVAL]

Where's Terry Bradshaw?

Bradshaw is no Trent Dilfer!

 

[AB in DC]

So basically, the biggest chokers (using a ratio of wins to expected wins) are Neil O'Donnell and Kordell Stewart. Sounds about right to me.

 

[Doug Drinen]

I really and truly have no point to make here, just a quick query that I thought was of interest.

Did people not get to this? Or did they just not believe it? Seriously.

You cook up a chart...

See, I didn't cook up the chart. I cooked up the question. The chart was simply an answer to the question. When I cooked up the question, I had no idea what the chart was going to look like.Look. I like to run statistical queries. I used to just post the results with no commentary. But when I did that, people always assumed I was advancing an agenda. So I started posting disclaimers like the above. It doesn't seem to matter. People still think I'm trying to advance an agenda.

I'm not.

I promise.

I suppose I it's my own fault for naming it "choke index." That name was tongue-in-cheek.

I'll admit that I am a Brady/Patriots hater (not nearly as much as I used to be --- a season of Brady and Dillon on my fantasy squad has caused them to grow on me quite a bit), but I could not be more ambivalent about Manning or Marino or anyone else on this list, for that matter. I really don't care if they're chokers or not. Like bostonfred, what I am interested in is the debate and how people choose to weight various bits of data in this discussion.

We've heard countless times that "Brady is undefeated" in the playoffs and "Manning is 3-5" (in quotes because really what's being said is that the Patriots are undefeated when Brady starts and the Colts are 3-5 when Manning starts). Plenty of people seem to think that's relevant data. If that's relevant, isn't it also relevant that most of Brady's games were at home against teams with worse records and most of Manning's were on the road against teams with better records?

I make no statement about whether wins and losses are the best way (or even a good way) to measure QB performance in the playoffs. But I do claim that my list is a better marker than raw W-L record.

Now, I'd be a lot happier with this post if I didn't now have to report that my original chart was in error, as was caught by MarshallRob. My program was misclassifying the "favorite" in super bowls, so guys who played in a lot of super bowls move around a bit. Here are the corrected numbers:

Expected  Actual                                          wins     wins     Diff-----------------------------------------------Trent Dilfer              2.5       5      +2.5Tom Brady                 5.0       7      +2.0Mark Rypien               3.0       5      +2.0Jim Everett               0.0       2      +2.0John Elway               12.0      14      +2.0Wade Wilson               1.5       3      +1.5Joe Montana              14.5      16      +1.5Jeff Hostetler            2.5       4      +1.5Drew Bledsoe              3.0       4      +1.0Jay Schroeder             2.0       3      +1.0Mark Brunell              2.0       3      +1.0Brad Johnson              3.0       4      +1.0Dave Krieg                2.0       3      +1.0Steve McNair              4.0       5      +1.0Jim Harbaugh              1.0       2      +1.0Doug Williams             3.5       4      +0.5Troy Aikman              10.5      11      +0.5Stan Humphries            3.0       3      +0.0Vinny Testaverde          2.0       2      +0.0Brett Favre              11.0      11      +0.0Phil Simms                6.5       6      -0.5Peyton Manning            3.5       3      -0.5Kerry Collins             3.5       3      -0.5Kurt Warner               5.5       5      -0.5Rich Gannon               5.0       4      -1.0Mike Tomczak              4.0       3      -1.0Randall Cunningham        4.0       3      -1.0Donovan McNabb            7.0       6      -1.0Bernie Kosar              4.0       3      -1.0Jim Kelly                10.0       9      -1.0Kordell Stewart           3.5       2      -1.5Dan Marino               10.0       8      -2.0Warren Moon               5.0       3      -2.0Jim McMahon               5.0       3      -2.0Neil O'Donnell            5.0       3      -2.0Steve Young              11.5       8      -3.5

I'll edit the original post and note the correction there.

 

[Doug Drinen]

I'm not a huge fan of QB rating either, but I didn't know what stats you'd have available to you to quantify the dropoff.

I've got all the standard stuff: attempts, completions, yards, TDs, INTs.If that's the data available and I have to concoct one stat to measure QB effectiveness, I'd go with

(Yards + 10*TDs - 45*INTs) / attempts

That stat was cooked up in a book called The Hidden Game of Football, where they make a pretty good case for why it's better than passer rating.

But I'll let you make the call. Let me know what measure you'd like to see and I'll work up lists of individual games with the biggest differences between game stats and regular season stats. Probably won't get to it til tonight, but it's not that tough a query.

 

[cobalt_27]

I'm not a huge fan of QB rating either, but I didn't know what stats you'd have available to you to quantify the dropoff.

I've got all the standard stuff: attempts, completions, yards, TDs, INTs.If that's the data available and I have to concoct one stat to measure QB effectiveness, I'd go with

(Yards + 10*TDs - 45*INTs) / attempts

That stat was cooked up in a book called The Hidden Game of Football, where they make a pretty good case for why it's better than passer rating.

But I'll let you make the call. Let me know what measure you'd like to see and I'll work up lists of individual games with the biggest differences between game stats and regular season stats. Probably won't get to it til tonight, but it's not that tough a query.

According to that formula (and I like it, too), here are the Top-30 performers of all-time (of passers who threw for at least 3000 yards):1. Manning 9.24 (2004)

2. Montana 8.86 (1989)

3. Chandler 8.75 (1998)

4. Jones 8.56 (1976)

5. Marino 8.50 (1984)

6. Young 8.45 (1992)

7. Cunningham 8.45 (1998)

8. Culpepper 8.41 (2004)

9. Young 8.39 (1994)

10. Warner 8.37 (1999)

11. Esiason 8.30 (1988)

12. Young 8.28 (1997)

13. McNabb 8.15 (2004)

14. Warner 8.15 (2000)

15. Montana 8.00 (1984)

16. Rypien 7.95 (1991)

17. Deberg 7.86 (1990)

18. McNair 7.85 (2003)

19. Tittle 7.83 (1963)

20. Brees 7.78 (2004)

21. Young 7.77 (1993)

22. O'Brien 7.74 (1985)

23. Young 7.71 (1998)

24. Pennington 7.69 (2002)

25. Warner 7.69 (2001)

26. Testaverde 7.67 (1998)

27. Unitas 7.66 (1963)

28. Anderson 7.64 (1975)

29. Theismann 7.64 (1983)

30. Aikman 7.60 (1993)

 

[KnowledgeReignsSupreme]

How about QB rating is 100% garbage?

Firstly... Out of curiosity, what's your statistical beef with passer rating?Secondly... I'd caution everyone in here passing arguments back and forth not take too much away from this. We can't even beging to determine what this means exactly.Does it mean that QBs near the top played on teams that were "underrated? (teams started off slow but played great down the stretch but are considered the underdog in this system because of total W-L record"Does it mean that QBs near the top are just so good they will their teams to win?Does it mean next to nothing because of the small sample size for most of the QBs on this list?Does it mean a team got lucky a couple times and because of the small sample size can greatly skew the overall result?Will Pats fans find any excuse they can to prop up Brady and bash Manning and call any arguments or stats contrary to theirs "cooked up" or biased?

 

[Wilbur Wood]

I'm not a huge fan of QB rating either, but I didn't know what stats you'd have available to you to quantify the dropoff.

I've got all the standard stuff: attempts, completions, yards, TDs, INTs.If that's the data available and I have to concoct one stat to measure QB effectiveness, I'd go with

(Yards + 10*TDs - 45*INTs) / attempts

That stat was cooked up in a book called The Hidden Game of Football, where they make a pretty good case for why it's better than passer rating.

But I'll let you make the call. Let me know what measure you'd like to see and I'll work up lists of individual games with the biggest differences between game stats and regular season stats. Probably won't get to it til tonight, but it's not that tough a query.

Personally, I'm in the camp that wins/losses, in and of themselves, should not have a lot of weight in the "who's a great" QB discussion, but rather how these QB's performed in the big games.I think that your proposed stat of "passing prowess" would do that when compared on a % basis to the regular season average of the given QB. This would isolate the effect of how given QB's perform during the crucial post-season as a better/worse comparison to their established in-season norms.

Now even this approach can suffer from a lot statistical "noise" as certain QB's may roll up tremnendous stats against an overmatched early playoff opponent and then lay a big egg against a good team in the game which eliminates them.

 

[jurb26]

Where's Terry Bradshaw?

Bradshaw is no Trent Dilfer!

No Bradshaw = FlaVVedHow can he not be in there? I was surprised to see how high Elway ranked in this list.

 

[houndirish]

Fascinating approach, particularly because it cements two of my own personally held beliefs...that Montana is the best ever, and that Marino doesn't deserve to be held among the very highest pantheon of NFL QBs.

I don't think the numbers for Marino are accurate. I count only nine expected wins, not eleven. Two in 1984 (two wins) and 1985 (a win and a loss), and one each in 1983 (loss), 1990 (win), 1992 (win), 1994 (win), and 1995 (win). Seven of his 10 playoff losses were on the road plus the Super Bowl loss which should be considered a road game since it was at Stanford. In the 1992 AFC title game Miami hosted the Bills and lost but both teams had the same record so this is not an "expected win" under the stated criteria. If it is, Marino should get credit for an "unexpected win" in 1999 on the road against Seattle where both teams had the same record.

I don't think SuperBowls can really ever be looked at as a home or road game for either team unless somebody is actually playing in their home stadium (which hasn't happened). That said, Marino's lone Super Bowl appearance shouldn't be considered a road game and thus count as an expected loss because of this. However seeing as the 49ers were favored in the game, I think there's it does qualify as an expected loss for Marion and an expected win for Montana.

 

[MarshallRob]

One QB not included in the original list would finish with a +3.0 differential for six expected wins and nine actual wins. You'd think a guy with that kind of record and the most NFL titles of any QB ever, five (!), might get a mention when people talk about the greatest QB's of all-time. Favre lovers who want to extol their man as the greatest QB ever might want to consider he might not even be the best QB his own team has ever had. Where's the Tom Brady "winning is the only thing" crowd for Bart Starr when he needs it?Of course, considering Starr's playoff record under the Drinen Choke Index reveals the limitations "inherent in the system". Since the Raiders had a better regular season record than the Packers in 1967, Green Bay's win would not be considered an expected win though it was about as expected a win as any game ever played.

 

[Leroy Hoard]

Doug, why don't you use the spread to determine who should have the "expected win". Even though it is probably most often the same, using records and home field data can be misleading when determining who is "expected to win". Case in point is this week's game between the Pats & Steelers. Using your method, you will flag it as an expected win for Pitt (better record, home field) despite the fact that Vegas, and really thus the nation, feel that the Pats are clearly the favorite based on the spread (expected to win).I am guessing, perhaps, that you don't have this statistic going back in time, no? Anyway, just my thoughts, and very interesting data none the less.

 

[Doug Drinen]

How can he not be in there?

7. The list contains all QBs whose career started in 1978 or later and played in 5 or more playoff games.

 

[Doug Drinen]

Doug, why don't you use the spread to determine who should have the "expected win"....I am guessing, perhaps, that you don't have this statistic going back in time, no? Anyway, just my thoughts, and very interesting data none the less.

Good guess. No spread data. If I had it, I agree that it would be the thing to use.I do think that situations like this week's NE-Pit game are probably pretty rare, but I can't say for sure.

 

[houndirish]

I also tend to think the Vegas line is a better indicator to who is "expected" to win. But, like everything else, it has it's holes. Take for example last week's Colts-Patriots game. When the line opened the Patriots were favored. By game time the Colts were the chalk. Granted this very rarely occurs but it just shows that even the point spread isn't definitive. Who gets credit/blame for expectation level in that game?

 

[ConstruxBoy]

Just happy to see Kelly ahead of Marino on some list besides SB losses....

 

[culdeus]

I also tend to think the Vegas line is a better indicator to who is "expected" to win. But, like everything else, it has it's holes. Take for example last week's Colts-Patriots game. When the line opened the Patriots were favored. By game time the Colts were the chalk. Granted this very rarely occurs but it just shows that even the point spread isn't definitive. Who gets credit/blame for expectation level in that game?

You always go off the closing lines for those sorts of questions. Unlike horseraces the odds on your ticket doesn't change but the going off line is what you measure ATS record against, not the opening or day of line.

 

[culdeus]

I also tend to think the Vegas line is a better indicator to who is "expected" to win. But, like everything else, it has it's holes. Take for example last week's Colts-Patriots game. When the line opened the Patriots were favored. By game time the Colts were the chalk. Granted this very rarely occurs but it just shows that even the point spread isn't definitive. Who gets credit/blame for expectation level in that game?

You always go off the closing lines for those sorts of questions. Unlike horseraces the odds on your ticket doesn't change but the going off line is what you measure ATS record against, not the opening or day of line.

And if it is a pick'em then the home team becomes the favorite for purposes of later review.

 

[MarshallRob]

Where's Terry Bradshaw?

Bradshaw is no Trent Dilfer!

No Bradshaw = FlaVVedHow can he not be in there? I was surprised to see how high Elway ranked in this list.

I took the liberty of figuring out where Bradshaw places on the Choke Index. He has both 14 Expected Wins and 14 actual wins for a 0.00 differential.

 

[cobalt_27]

Where's Terry Bradshaw?

Bradshaw is no Trent Dilfer!

No Bradshaw = FlaVVedHow can he not be in there? I was surprised to see how high Elway ranked in this list.

I took the liberty of figuring out where Bradshaw places on the Choke Index. He has both 14 Expected Wins and 14 actual wins for a 0.00 differential.

We really need to change the methodology here. It's really only speaking to who outperformed his expectancy and who underperformed. Obviously, it doesn't speak to much else. Doesn't say anything about that group in the middle.

 

[Maurile Tremblay]

No spread data. If I had it, I agree that it would be the thing to use.

Spread data wouldn't work. The only reason the Colts were underdogs in those five losses in the first place is that the handicappers already took into account that Manning is a choker. Seriously, I doubt there's any good way to reliably determine that one QB is more of a choker than another. The sample size of "big games" is probably never going to be big enough to make "choking" a better explanation than random variance.

 

[Chase Stuart]

No spread data. If I had it, I agree that it would be the thing to use.

Spread data wouldn't work. The only reason the Colts were underdogs in those five losses in the first place is that the handicappers already took into account that Manning is a choker. Seriously, I doubt there's any good way to reliably determine that one QB is more of a choker than another. The sample size of "big games" is probably never going to be big enough to make "choking" a better explanation than random variance.

 

[bostonfred]

No spread data.  If I had it, I agree that it would be the thing to use.

Spread data wouldn't work. The only reason the Colts were underdogs in those five losses in the first place is that the handicappers already took into account that Manning is a choker. Seriously, I doubt there's any good way to reliably determine that one QB is more of a choker than another. The sample size of "big games" is probably never going to be big enough to make "choking" a better explanation than random variance.

The only way to reliably measure an individual player's choke is to look at the increase or decrease in that one player's performance in each playoff game, and compare it with their regular season averages. Some downturn should be expected, because playoff games are usually played against tougher defenses and played in winter weather, and it may make sense to account for that somehow. Manning will have some of the better playoff games (against KC and Denver), and some of the worst (against NE and NYJ), of all time.

 

[KnowledgeReignsSupreme]

The only way to reliably measure an individual player's choke is to look at the increase or decrease in that one player's performance in each playoff game, and compare it with their regular season averages.

You'd have to do that with a through analysis of game tape by experienced... game tape watchers. Coming up with a statistic to try and prove an individual player "choked" proves nothing.

Some downturn should be expected, because playoff games are usually played against tougher defenses and played in winter weather, and it may make sense to account for that somehow.

Not only should downturn be expected, but unexplainable ups and downs should be expected. Sometimes things happen because of dumb luck and through no skill or fault of a person, but it's human nature to assume there is a cause. In this case the assumption is that Manning "chokes" when he plays New England.

 

[MarshallRob]

The only way to reliably measure an individual player's choke is to look at the increase or decrease in that one player's performance in each playoff game, and compare it with their regular season averages. 

Some downturn should be expected, because playoff games are usually played against tougher defenses and played in winter weather, and it may make sense to account for that somehow.

I don't think even this would work. If one team falls substantially behind early, then the QB is going to have to throw lots of passes to get his team back in the game. At the same time, the chances those passes will be effective will have declined as the defense knows the other team will be abandoning the run (a running game that probably wasn't working anyway) and the QB will have to take more chances to try to make big plays. Completion percentages will fall and INT's will rise as a matter of course in this situation. a situation that happened quite a few times to someone like Dan Marino who played with some weak defensive teams and never had much of a running game game at any time. In Dan's one Super Bowl appearance, he threw 50 passes. In his 3 playoff losses to the Bills he threw 49, 45, and 64 passes respectively. He was 1-6 in playoff games where he had to throw more than 40 passes. When a QB has to throw that many passes, it's unlikely his stats are going to be impressive or that his team is going to win. (Playoff record when throwing 40 or more passes in a playoff game: Montana: 1-2; Aikman 0-2; Young 0-2; Elway 0-1; Manning 0-3; Favre 0-4; Kelly 0-3; Brady 3-0 ; Warner 2-2; McNabb 0-2).

Plus, a "choke" might not even be related to a QB's statistical performance. Just a couple of bad plays, or even one play alone could constitute a choke. If a QB overthrows a wide open receiver on a potential game-winning play or takes a key sack that costs his team a makeable game-winning field goal, that might be called a choke though the QB may have ended up with great stats for the game.

P.S. I also see I've inadvertently come across some evidence for the "Brady is God" crowd.

 

[Maurile Tremblay]

The only way to reliably measure an individual player's choke is to look at the increase or decrease in that one player's performance in each playoff game, and compare it with their regular season averages.

Correlations that don't signify anything real exist all over the place. If you break down every QB's performance according to how well each has played after I've eaten a pastrami sandwich the previous Wednesday, you will find that certain QBs have done much better than normal in pastrami weeks while others have totally choked. If I haven't eaten pastrami all that often, the effect for certain players will be dramatic. Not because my lunch choices actually affect their play -- but just as the result of random variance applied to a small sample size.There are ways to test for statistical significance by looking at standard deviation for weekly performaces generally, sample sizes of pastrami weeks and non-pastrami weeks, and how big the pastrami effect actually is.I'm suggesting that, with the small sample sizes of playoff games along with high standard deviation of weekly QB performance in general, any kind of "playoff choking effect" we find will probably have no more statistical significance than the pastrami effect.

 

[Joe Bryant]

The only way to reliably measure an individual player's choke is to look at the increase or decrease in that one player's performance in each playoff game, and compare it with their regular season averages.

Correlations that don't signify anything real exist all over the place. If you break down every QB's performance according to how well each has played after I've eaten a pastrami sandwich the previous Wednesday, you will find that certain QBs have done much better than normal in pastrami weeks while others have totally choked. If I haven't eaten pastrami all that often, the effect for certain players will be dramatic. Not because my lunch choices actually affect their play -- but just as the result of random variance applied to a small sample size.There are ways to test for statistical significance by looking at standard deviation for weekly performaces generally, sample sizes of pastrami weeks and non-pastrami weeks, and how big the pastrami effect actually is.I'm suggesting that, with the small sample sizes of playoff games along with high standard deviation of weekly QB performance in general, any kind of "playoff choking effect" we find will probably have no more statistical significance than the pastrami effect.

What MT said.J

 

[bostonfred]

The only way to reliably measure an individual player's choke is to look at the increase or decrease in that one player's performance in each playoff game, and compare it with their regular season averages.

Correlations that don't signify anything real exist all over the place. If you break down every QB's performance according to how well each has played after I've eaten a pastrami sandwich the previous Wednesday, you will find that certain QBs have done much better than normal in pastrami weeks while others have totally choked. If I haven't eaten pastrami all that often, the effect for certain players will be dramatic. Not because my lunch choices actually affect their play -- but just as the result of random variance applied to a small sample size.There are ways to test for statistical significance by looking at standard deviation for weekly performaces generally, sample sizes of pastrami weeks and non-pastrami weeks, and how big the pastrami effect actually is.I'm suggesting that, with the small sample sizes of playoff games along with high standard deviation of weekly QB performance in general, any kind of "playoff choking effect" we find will probably have no more statistical significance than the pastrami effect.

I understand the concept of correlation without causation, but you can't just assume there's no causation, either. If Manning has two or three of, say, the 15 greatest gaps between regular and postseason performance in history, and they're all in the playoffs, then the correlation is pretty strong. I don't know how to measure a standard deviation for his season average or career average performances, but you'd have to think his 4 INT game, his 137 yard game, and his 238/0/1 game in a season when he averaged over 300 yards, 3.5 TDs and .6 INTs per game through his last 16 games (and hadn't had a 0 TD performance through those previous 16 games) would be several standard deviations away from his expected value, no? So let's say use your pastrami sandwich example. There's a pastrami sandwich place down the street. Usually, you like pastrami sandwiches. There's good pastrami, and there's bad pastrami, but you usually like them. And there's this great pastrami sandwich down the street that everyone raves about - your friends have been there 16 times recently, and usually go 16 times a year. You go to this pastrami sandwich place, and the first time you eat there, it's not that good. You go there again, and it's so bad you actually get sick. Undaunted, you try again and it's still not very good, and definitely below average. Now your friends are trying to convince you to go there with them, and you finally cave and when you get there you get a great sandwich. Not just good, but great. And you go again, and it's even better. Which is why it's so surprising when you go again right away, and the pastrami is so rancid, the place gets shut down for a year. But you, the glutton for punishment (or at least a glutton for pastrami) hear that the place has been better than ever before. Your friends have been there 16 times, and not only have they not gotten a bad sandwich, they've had 15 straight great sandwiches. And you're rewarded with a very good sandwich. Just as you're trying to convince your friends that the place is good again, though, lo and behold, they serve you up a poop sandwich on rye. After all this, do you reasonably expect a good sandwich next time? It's really only a few data points, and you've had some good sandwiches in the past. It can't be just you, can it? I mean, they're just making sandwiches, it's not like they know it's you. But maybe there's a difference between what your friends are calling a great sandwich, and what you are. Maybe your friends are a little more forgiving, and it's easier for them to call it a great sandwich when you're not around because they're not that picky. Or it might just be bad luck.

 

[Joe Bryant]

The only way to reliably measure an individual player's choke is to look at the increase or decrease in that one player's performance in each playoff game, and compare it with their regular season averages.

Correlations that don't signify anything real exist all over the place. If you break down every QB's performance according to how well each has played after I've eaten a pastrami sandwich the previous Wednesday, you will find that certain QBs have done much better than normal in pastrami weeks while others have totally choked. If I haven't eaten pastrami all that often, the effect for certain players will be dramatic. Not because my lunch choices actually affect their play -- but just as the result of random variance applied to a small sample size.There are ways to test for statistical significance by looking at standard deviation for weekly performaces generally, sample sizes of pastrami weeks and non-pastrami weeks, and how big the pastrami effect actually is.

I'm suggesting that, with the small sample sizes of playoff games along with high standard deviation of weekly QB performance in general, any kind of "playoff choking effect" we find will probably have no more statistical significance than the pastrami effect.

I understand the concept of correlation without causation, but you can't just assume there's no causation, either. If Manning has two or three of, say, the 15 greatest gaps between regular and postseason performance in history, and they're all in the playoffs, then the correlation is pretty strong.

I don't know how to measure a standard deviation for his season average or career average performances, but you'd have to think his 4 INT game, his 137 yard game, and his 238/0/1 game in a season when he averaged over 300 yards, 3.5 TDs and .6 INTs per game through his last 16 games (and hadn't had a 0 TD performance through those previous 16 games) would be several standard deviations away from his expected value, no?

So let's say use your pastrami sandwich example. There's a pastrami sandwich place down the street. Usually, you like pastrami sandwiches. There's good pastrami, and there's bad pastrami, but you usually like them. And there's this great pastrami sandwich down the street that everyone raves about - your friends have been there 16 times recently, and usually go 16 times a year.

You go to this pastrami sandwich place, and the first time you eat there, it's not that good. You go there again, and it's so bad you actually get sick. Undaunted, you try again and it's still not very good, and definitely below average.

Now your friends are trying to convince you to go there with them, and you finally cave and when you get there you get a great sandwich. Not just good, but great. And you go again, and it's even better. Which is why it's so surprising when you go again right away, and the pastrami is so rancid, the place gets shut down for a year.

But you, the glutton for punishment (or at least a glutton for pastrami) hear that the place has been better than ever before. Your friends have been there 16 times, and not only have they not gotten a bad sandwich, they've had 15 straight great sandwiches. And you're rewarded with a very good sandwich. Just as you're trying to convince your friends that the place is good again, though, lo and behold, they serve you up a poop sandwich on rye.

After all this, do you reasonably expect a good sandwich next time? It's really only a few data points, and you've had some good sandwiches in the past. It can't be just you, can it? I mean, they're just making sandwiches, it's not like they know it's you. But maybe there's a difference between what your friends are calling a great sandwich, and what you are. Maybe your friends are a little more forgiving, and it's easier for them to call it a great sandwich when you're not around because they're not that picky.

Or it might just be bad luck.

Hi bf,I guess it can be anything you want it to be. I think it's more along what MT said when he started:

I'm suggesting that, with the small sample sizes of playoff games along with high standard deviation of weekly QB performance in general, any kind of "playoff choking effect" we find will probably have no more statistical significance than the pastrami effect.

But I know the chances of you thinking anything other than Manning is a huge choker is zero so not sure why I'm even replying in this one other than to say I agree with Maurile.J

 

[cobalt_27]

The only way to reliably measure an individual player's choke is to look at the increase or decrease in that one player's performance in each playoff game, and compare it with their regular season averages.

Correlations that don't signify anything real exist all over the place. If you break down every QB's performance according to how well each has played after I've eaten a pastrami sandwich the previous Wednesday, you will find that certain QBs have done much better than normal in pastrami weeks while others have totally choked. If I haven't eaten pastrami all that often, the effect for certain players will be dramatic. Not because my lunch choices actually affect their play -- but just as the result of random variance applied to a small sample size.There are ways to test for statistical significance by looking at standard deviation for weekly performaces generally, sample sizes of pastrami weeks and non-pastrami weeks, and how big the pastrami effect actually is.

I'm suggesting that, with the small sample sizes of playoff games along with high standard deviation of weekly QB performance in general, any kind of "playoff choking effect" we find will probably have no more statistical significance than the pastrami effect.

I understand the concept of correlation without causation, but you can't just assume there's no causation, either. If Manning has two or three of, say, the 15 greatest gaps between regular and postseason performance in history, and they're all in the playoffs, then the correlation is pretty strong.

I don't know how to measure a standard deviation for his season average or career average performances, but you'd have to think his 4 INT game, his 137 yard game, and his 238/0/1 game in a season when he averaged over 300 yards, 3.5 TDs and .6 INTs per game through his last 16 games (and hadn't had a 0 TD performance through those previous 16 games) would be several standard deviations away from his expected value, no?

So let's say use your pastrami sandwich example. There's a pastrami sandwich place down the street. Usually, you like pastrami sandwiches. There's good pastrami, and there's bad pastrami, but you usually like them. And there's this great pastrami sandwich down the street that everyone raves about - your friends have been there 16 times recently, and usually go 16 times a year.

You go to this pastrami sandwich place, and the first time you eat there, it's not that good. You go there again, and it's so bad you actually get sick. Undaunted, you try again and it's still not very good, and definitely below average.

Now your friends are trying to convince you to go there with them, and you finally cave and when you get there you get a great sandwich. Not just good, but great. And you go again, and it's even better. Which is why it's so surprising when you go again right away, and the pastrami is so rancid, the place gets shut down for a year.

But you, the glutton for punishment (or at least a glutton for pastrami) hear that the place has been better than ever before. Your friends have been there 16 times, and not only have they not gotten a bad sandwich, they've had 15 straight great sandwiches. And you're rewarded with a very good sandwich. Just as you're trying to convince your friends that the place is good again, though, lo and behold, they serve you up a poop sandwich on rye.

After all this, do you reasonably expect a good sandwich next time? It's really only a few data points, and you've had some good sandwiches in the past. It can't be just you, can it? I mean, they're just making sandwiches, it's not like they know it's you. But maybe there's a difference between what your friends are calling a great sandwich, and what you are. Maybe your friends are a little more forgiving, and it's easier for them to call it a great sandwich when you're not around because they're not that picky.

Or it might just be bad luck.

Hi bf,I guess it can be anything you want it to be. I think it's more along what MT said when he started:

I'm suggesting that, with the small sample sizes of playoff games along with high standard deviation of weekly QB performance in general, any kind of "playoff choking effect" we find will probably have no more statistical significance than the pastrami effect.

But I know the chances of you thinking anything other than Manning is a huge choker is zero so not sure why I'm even replying in this one other than to say I agree with Maurile.J

And, I agree with Joe.