Writers of Pro Football Prospectus 2008

21 Jan 2011

The Rivers Index

Neil Paine at the pro-football-reference.com blog has updated his "Rivers Index", which measures how many wins a quarterback should have based on his postseason performance, for a variety of new factors and the 2010 postseason to date.

Posted by: Bill Barnwell on 21 Jan 2011

33 comments, Last at 30 Jan 2011, 9:22am by troycapitated polamalizer


by Mr Shush :: Fri, 01/21/2011 - 5:29pm

Poor old Phillip Rivers. Interesting too that, even excluding 1999 and 2000, Kurt Warner's average playoff performance was (at least by this metric) clearly better than that of any of his illustrious peers.

by Bobman :: Fri, 01/21/2011 - 8:46pm

As TMQ used to theorize, Warner is an alien in human form. Sometimes the fumble-prone Gilligan suit he wore dominated, but when he was on, damn, he was the galactic MVP.

by Toast Patterson (not verified) :: Fri, 01/21/2011 - 8:32pm

So three of the top four playoff QBs over the last ten years (by xWin%)are Kelly Holcomb, Jay Cutler & Vinny Testaverde? Huzzah!

by Spielman :: Fri, 01/21/2011 - 8:59pm

Well, small sample size will do that sort of thing.

by Bobman :: Fri, 01/21/2011 - 8:55pm

More proof that all Peyton Manning does is choke in big games.

With his teammates' hands generally around his throat.

And adding to that weirdness is the fact that Manning's teams lost to Rivers's teams TWICE in the playoffs. Two of the "unluckiest" three faced off multiple times in the postseason, with the same result each time. So Manning's crappy luck trumps that of Rivers. If Rivers had played just as well, but against the Pats or Steelers in those two playoff games, he'd have two more losses and be the absolute, unrivaled monarch of the playoff netherworld. ...instead of just some guy with an index named after him, like Dow Jones.

by Enrique (not verified) :: Mon, 01/24/2011 - 12:39pm

It is actually the opposite. It shows that Manning has been unlucky in results while playing great. Did you actually read the article?

by Bobman :: Mon, 01/24/2011 - 4:22pm

Sorry. I must have muted my sarcasm button.

Just turned it up high... you getting it now?

Actually, now that I look at it, saying he "choked" with his teammates hands around his throat was pretty obvious, no? Ths implication was that they were choking him. Ah, never mind.

by Bionicman :: Fri, 01/21/2011 - 9:16pm

I'm calling BS on the claim that Rivers has had bad playoff luck. Rivers is a perfect example of why stats that measure QB performance need a 'garbage time' adjustment. In both of his recent playoff losses (against the Steelers and Jets), he was mediocre at best when the game was a contest, then threw for a lot of yards and touchdowns after the opponents were playing prevent defense. This is especially true of his game against the Steelers, when he threw for 161 of his 308 yards and 2 of his 3 touchdowns when his team was down three scores in the fourth quarter. He got a 'win' (second playoff game against the Colts) when he was awful. His best playoff performance (in the first win over the Colts) was against a defense that gave up a game winning drive to his backup, Billy Volek, after Rivers got injured. I simply can't find one playoff game the Chargers lost where Rivers's performance warranted a 'win' for him.

by Vicious Chicken Of Bristol (not verified) :: Fri, 01/21/2011 - 10:10pm

Agreed. He was crap in that game, until it had no effect on the outcome. Clearly the formula needs tweeking

by Mr Shush :: Sat, 01/22/2011 - 10:37am

In fairness, those Jets and Steelers had pretty incredible defenses. They both led the league in DVOA, and the Steelers were the best DVOA defense since 2004. He probably got a pretty solid opponent adjustment for those games.

by kamiyu206 :: Sat, 01/22/2011 - 12:19am


It might not matter much when it comes to evaluating regular season games, but playoffs are different. Sample size is just too small. Brady played 18 playoff games and that's the most among his peers.

I understand what Paine tried to find out, but seriously there need some tweaks on the metric.

by dub_rex (not verified) :: Sat, 01/22/2011 - 2:16am

C'mon man. The first sentence, see the words "nothing more:"
"I developed what I called "The Rivers Index" (so named for Philip Rivers), a metric that measured how many games a QB should have won based on nothing more than his own passing performance."

by tuluse :: Sat, 01/22/2011 - 5:25am

On the other hand, if the defense kept the game close, he might have been able to pull off a win at the end.

by Sjt (not verified) :: Sat, 01/22/2011 - 2:20pm

In the first Indy game, Rivers was 14-19-264-3-1 on the road against the #2 DVOA defense that year (including # 2 against the pass). He did that sans LT after the first quarter, and with Gates hobbling around after being "doubtful" all week.

He was "awful" in a game where he threw for 217 yards, had only one turnover, and led his team on 4 critical scoring drives, three of which were ended by rushing TDs after Rivers had passed them down the field? Against a top 10 DVOA defense, without LT and with a hobbled Antonio Gates (noticing a trend here).

The Steelers game was 10-14 at the half, and the San Diego defense didn't do much to help things for the entire game. They gave up long, time consuming drive to start the 3rd quarter. After that, River made his one mistake for the game, when a spectacular play by the Steelers resulted in a deflected interception. The San Diego defense (again) got run the hell over, and the Chargers were quickly down 18 points after having run only 1 play the entire 3rd quarter. Rivers then drove his team down to score and cut the lead, only to have the defense (again!) get run over and give up another TD

In the Jets game his team lead at the half, and his efforts were hurt by a 10 Charger penalties, a non existent running game a defense which eventually wore down against the leagues top running offense, and 3 missed kicks by the most accurate kicker in NFL history.

The Jets and Steelers were #1 in DVOA in their respective years. One went on to win the Super Bowl, the other was leading at the half in the AFCCG.

And while this doesn't fit with the "River index", how bad is your bad luck when you have to go to New England to play a 17-0 team with your best weapon hurt and yourself one week from emergency ACL surgery? And you still make it a close game?

by Nate Dunlevy :: Sat, 01/22/2011 - 4:00pm

I was at that first game in Indy, and while the Colts DVOA was great, that wasn't the Indy D. Freeney was out, and Brock and Mathis were both hurt. There was no rush and he had all day to throw and didn't look that great. His performance in that game is wildly overrated. He left with his team losing.

He was legitimately terrible the next week in New England, and Cowher was calling for him to be benched at half time. Had Volek played, the Chargers might have actually won. Rivers play that day was the weakest link and cost them that game.

by Sjt (not verified) :: Sat, 01/22/2011 - 7:07pm

Freeney had been out half the year, and yet that Colts team still managed to finish with the #2 DVOA and #2 pass DVOA. They were the #2 seed and had a week off to prepare, and they were playing in the loudest stadium in football against a team with major injuries on offense. You'll get no sympathy from me for that defense.

"There was no rush and he had all day to throw and didn't look that great"

14-19-264-3-1. Including one of the prettiest TD passes I've every seen (the ball to Chambers). In 3 quarters. With a hobbled Gates and no LT for half the game. This is what passes for "not that great"?

"He left with his team losing."

Not before leading them to 3 TDs, having Kaeding doink 3 points off the crossbar, and having Cromartie's go ahead TD taken off the board by a bogus holding call. He did his job and put his team in position to win for as long as he could play.

"He was legitimately terrible the next week in New England"

So was Tom Brady.

"Had Volek played, the Chargers might have actually won"

We'll never know. We do know that Rivers, despite his injury and poor play, did keep his team in the game. The failure in that game was in converting in the red zone. But he did get them to said redzone, and but for a hobbled Gates who couldn't run full spead and a great tackle but Seau on a run play, two FGs turn into 2 TDs and the whole game changes.

And that particular game is besides the point, because its actually a game the formula predicts him to lose. Its not bad luck if you play bad and still lose. The bad luck was the torn ACL and the other injuries to his team.

"Rivers play that day was the weakest link and cost them that game."

I love it. The guy plays a great game and his team wins = doesn't deserve credit.

The guy plays a bad game = he's horrible and he's the reason his team lost.

by Bobman :: Mon, 01/24/2011 - 4:38pm

Not quite. Freeney got hurt halfway through week 11--so he contributed 10.5 games, not just eight. 2.5 games doesn't sound like a lot, but it's a pretty big percentage--31% more than the eight games that "halfway" implies. And with him out, the double-teams (and niggling injuries) just wore Mathis down more and more, diminishing his contribution. (They could not afford to rest the 245-lb DE with Freeney out, so the D saw a steady decline after week 11.)

And while I am no fan of Rivers, I agree DZ is a little hard on him there. It seems like he's lost a lot of weapons over the years and hardly skipped a beat.

Your Tom Brady stunk too argument is... flying in from left field on the straw man express and has no place in this conversation.

Now the next season's win over the Colts--regardless of Rivers's performance--was not due to him nearly as much as special teams play (especially Scifres) and, I hate to say it, a lot of horrid non-calls. So the team DID get the win, but not sure how much of the true credit belongs to Rivers. I mean an OT game where the losing team's average starting field position is about 15 yards lower than the winning team? Five or six Colt possessions inside their own 20? That's not Rivers....

by Vicious Chicken Of Bristol (not verified) :: Tue, 01/25/2011 - 10:40am

"The Steelers game was 10-14 at the half, and the San Diego defense didn't do much to help things for the entire game. They gave up long, time consuming drive to start the 3rd quarter. After that, River made his one mistake for the game, when a spectacular play by the Steelers resulted in a deflected interception."

That was about as Rivers apologistic (yes, I made up a word) as you can get.

Philip Rivers started the game with a big pass to Vincent Jackson for a 40+ yard TD, 3 minutes into the game. Yup, he sure started off great.

Then, for the next 45 minutes of game time, his numbers were as follows:

9/17, 73 yards, 0 TD, 1 INT, 2 sacks.

At that point it was 28-10, with 12 minutes to go in the 4th qtr. To say the rest wasnt garbage time is ridiculous.

As the original poster claimed, Rivers is the shining example of empty numbers.

The Chargers D allowed only 4.9 yards per play (the Steelers D allowed 5.7). Blaming the defense? Really? Why not blame the offense being being unable to sustain a drive?

After their first TD drive (and before garbage time) the Charger offense had 7 drives (not counting the kneel down at the end of the first half). 5 of them were 5 plays or less. The average time of possesion for those drives was barely over 2 minutes.

Yeah, it was all the defense's fault.

by Neoplatonist Bolthead (not verified) :: Wed, 01/26/2011 - 9:14pm

Dude, that was the best defense of the 2000s they were playing.

by troycapitated p... :: Sun, 01/30/2011 - 9:22am

The yards allowed per play stat is probably skewed by the big play for an early score followed by very few offensive plays in the middle and ended with frantic catch-up mode where they again hit a few big plays. The Steelers, on the other hand, sustained numerous long drives and probably ran significantly more offensive plays than San Diego.

And, yes those were absolutely garbage scores. In no way were the Chargers ever a threat to come all the way back in that game.

by Sjt (not verified) :: Sat, 01/22/2011 - 2:25pm

And of course, lets not overlook his first every playoff game. He didn't play great, but he should have won anyway. If you switch that result he would be closer to his projected winning percentage.

by Sjt (not verified) :: Sat, 01/22/2011 - 2:36pm

And I really need to proofread.

by DKNJ (not verified) :: Mon, 01/24/2011 - 12:37am

Disagree....the Steelers held the ball for 14 minutes in the 3rd quarter of that game. You are penalizing Rivers for not being able to get on the field. He played very well that game against a historic Pitt D. He had a very nice 1st half too. It wasn't his fault he got the ball back down 2 scores to start the 4th quarters. 2 scores is hardly garbage time....if you want garbage time witness what Hasselback did against the Bears.

Rivers has faced the #1 pass D 3 times, the #2 pass, the #5 pass D twice and the #9 pass D. No easy games to amplify his stats.

by Jerry :: Mon, 01/24/2011 - 2:36am

The Chargers' win probability, according to Brian Burke's advancednflstats.com, never exceeded 10% in the fourth quarter. Neil Paine's win percentages are just based on how often a team whose QB puts up a particular set of passing stats wins, regardless of context. So when, as in the case of the Steelers-Chargers playoff game, Rivers puts up big numbers in the fourth quarter of a game that's over for all intents and purposes, it improves his "win percentage" without making his team that much more likely to win.

No, this isn't meant to denigrate Philip Rivers, who's far from the only quarterback to "benefit" from a good passing performance while desperate. It just points up the limitations of what Neil is looking at here.

by coltrane (not verified) :: Mon, 01/24/2011 - 1:33pm

But the Steelers had the ball for 14 minutes of the 3rd quarter. Rivers had a very nice 1st half against Pitt. It is unfair to critique him when Pitt pounds the clock.

The Steelers team played from ahead all year and no one put up "garbage stats" on them. I don't think 2 TDs are garbage stats either....he put the SD D in a position to make a meaningful stop.

by Jerry :: Mon, 01/24/2011 - 7:54pm

The one play the Chargers offense ran in the third quarter was a tip and interception, which isn't much. but doesn't help Rivers' case.

Last night's game is a little fresher in my mind right now, but my recollections of that Charger game are Rivers having an OK first half, the Steelers dominating the third quarter, and being able to enjoy Sproles zipping through the Steeler defense for the last TD since it didn't mean anything. The next regular season, there was an even more lopsided game at Heinz Field where Rivers put up good numbers in a second half where Burke never had San Diego's win probability above 9%. Again, I'm not trying to insult Rivers. It's just an issue with taking this statistic too seriously.

by Vicious Chicken Of Bristol (not verified) :: Tue, 01/25/2011 - 10:46am

"You are penalizing Rivers for not being able to get on the field."

Again, this is just wrong. Rivers had SEVEN drives between TD's. 5 of those drives were 5 plays or less. 4 of them used less that 2 minutes of game time.

He got on the field. He just didnt do anything when he got there.

If anything, the offense screwed the defense over by not letting them get OFF the field for any meaningful amount of time.

by SuperBearsSuperBowl (not verified) :: Fri, 01/21/2011 - 10:56pm

That's awesome. It is now statistically shown that Matt Ryan turns into Chris Simms when the calendar turns to January.

MATTY ICE!!!!!!!!!

by Bobman :: Mon, 01/24/2011 - 4:43pm

Damn, too many Chris Simms jokes I could make. I started with the spleen, then the less-than-articulate father, and then the Cheech and Chong-mobile arrest. (Interesting that he has elected to go to trial instead of havng a traffic infraction on his record; he must really feel he's innocent... or he's still stoned.) I don't really have time to pick the perfect angle.

by RickD :: Sat, 01/22/2011 - 6:55pm

The problem with an analysis like this is that the author never really validates the soundness of using AYAAA as a predictor of whether a team should win or not. It just isn't statistically sound to presume the model is meaningful and that deviation from the lines established by regression are happening by chance.
I'm particularly suspicious of any analysis that attempts to calculate an expected winning percentage from an offensive statistic alone. Defensive statistics really do matter, too. Sure you can plot a regression line using only one offensive statistic, but it seems absurdly unsound to claim that the deviation from the line you see is due to luck, when bad modeling is just as likely to be the cause.

by Sjt (not verified) :: Sat, 01/22/2011 - 9:32pm

I don't think the author ever makes such a claim. He simply ran a logit regression with win/loss results at the dependent variable and QB stats, weather, and defensive adjustment as the independent variables. The result merely stats that, historically, a QB with stat line X is likely to be on the winning side Y percent of the time.

by Mr Shush :: Sat, 01/22/2011 - 10:17pm

But isn't quite a large part of the point that the defense's performance is luck, from the quarterback's perspective?

I mean, obviously this is a pretty silly, knockaround stat for shits and giggles, but I don't think it does too bad a job of being what it is.

by tuluse :: Sun, 01/23/2011 - 5:25am

I think this is less analysis and more just having some fun.