## Who's 'Clutch' in 2010?

Two of the most clutch QBs of 2010 face off tonight as the Falcons host the Saints. Although 'clutch' performance may not be a persistent skill in players, there undoubtedly exists clutch play itself. Due to the varying leverage created by the combination of score and time, some players will have their better moments when they matter most, and some players will have their worst moments at just the wrong time.

Here's one way to measure which QBs are most 'clutch' this season. WPA accounts for the leverage of score and time while EPA ignores it. We can plot each QB's Win Probability Added (WPA) against his Expected Points Added (EPA), creating a baseline expected WPA for each QB.

Next, we can measure the difference between each QB's expected WPA and his actual WPA. This difference could be considered clutch-ness. This is illustrated in the graph below. The vertical red line happens to be the clutch-ness of the most clutch QB of 2010, Matt Ryan.

The table below lists the QBs from most clutch to least so, according to their differences between actual WPA and expected WPA. Matt Ryan's has over-performed his expected WPA by about 2 full wins. We could say that the guys at the top of the list are playing over their heads to some degree, while the guys at the bottom of the list have been snake-bitten. I would expect to see some regression-to-the-mean effects at some point for guys like Ryan and Flacco, and I think we've already seen Sanchez's play fall back down to Earth in recent weeks.

The guys in the middle of the list have posted WPA numbers about what we'd expect given their EPA numbers. That doesn't mean they're poor clutch performers at all. It just says that they perform about equally well in the clutch as they do in most other situations.

 Player Team G WPA EPA Exp. WPA 'Clutch' 2-M.Ryan 14 4.43 101.9 2.40 2.03 5-J.Flacco 15 3.53 84.2 1.94 1.59 6-M.Sanchez 15 1.62 17.7 0.19 1.43 1-T.Smith 6 1.34 8.1 -0.06 1.40 9-D.Garrard 14 2.42 50.2 1.05 1.37 9-D.Brees 14 4.30 136.2 3.30 1.00 5-J.Freeman 15 3.04 94.8 2.21 0.83 5-B.Gradkowski 6 0.22 -11.2 -0.56 0.78 8-S.Bradford 15 0.48 1.4 -0.23 0.71 2-J.Clausen 12 -1.70 -79.0 -2.34 0.64 3-M.Moore 6 -0.38 -26.2 -0.96 0.58 5-D.McNabb 13 0.48 10.7 0.01 0.47 6-J.Cutler 14 1.63 59.1 1.28 0.35 8-M.Hasselbeck 14 -0.20 -9.9 -0.53 0.33 12-A.Rodgers 14 3.64 138.8 3.37 0.27 10-E.Manning 15 1.44 55.2 1.18 0.26 3-D.Anderson 12 -0.89 -32.0 -1.11 0.22 10-V.Young 9 0.86 35.4 0.66 0.20 8-J.Campbell 12 1.06 45.7 0.93 0.13 11-A.Smith 10 -0.41 -8.8 -0.50 0.09 12-T.Brady 15 3.26 140.2 3.40 -0.14 12-C.McCoy 7 -0.03 17.5 0.19 -0.22 3-J.Kitna 10 0.60 42.3 0.84 -0.24 5-K.Collins 9 0.02 21.4 0.29 -0.27 18-P.Manning 15 2.67 130.5 3.15 -0.48 4-K.Kolb 6 -0.25 21.0 0.28 -0.53 7-M.Cassel 14 1.61 92.0 2.14 -0.53 7-B.Roethlisberger 11 1.55 90.1 2.09 -0.54 7-M.Vick 11 1.86 102.5 2.42 -0.56 4-B.Favre 13 -1.55 -26.2 -0.96 -0.59 9-T.Romo 6 -0.23 30.8 0.54 -0.77 8-M.Schaub 15 1.64 103.7 2.45 -0.81 14-R.Fitzpatrick 13 -0.88 11.1 0.02 -0.90 17-J.Delhomme 5 -1.83 -24.9 -0.92 -0.91 8-K.Orton 13 0.12 50.3 1.05 -0.93 7-C.Henne 14 -1.16 14.7 0.12 -1.28 17-P.Rivers 15 2.06 140.2 3.40 -1.34 14-Sh.Hill 10 -0.65 43.7 0.88 -1.53 9-C.Palmer 15 -0.64 53.1 1.12 -1.76

A logical next step would be to measure the consistency of clutch play, to begin to probe the question of whether there is anything to the notion that it's a persistent skill or something else. With the small sample sizes and large variance in outcomes, I doubt it's measurable even if it exists.

Note: Stats are through Sunday's (12/26) games.

### 8 Responses to “Who's 'Clutch' in 2010?”

1. Tarr says:

The obvious thing to do is to just check the year-to-year (or even first half of the season to second half of the season) correlation of "clutch". I think it's probably safe to re-label that column "luck"...

I knew Clausen was bad, but wow, I had no idea. Hey, at least he's been pretty clutch...

2. randomdude2228 says:

I don't see the difference between "clutch" and "luck" here. I like the premise though, simple yet effective. Seems comparable to an FIP vs. ERA comparison.

3. bigmouth says:

randomdude, there probably is no difference, which is why he distinguishes between clutch as a skill vs. clutch performances.

However, the question is still technically open because (1) no one has actually tested whether such performances are consistent from year to year (which could suggest clutch is a skill), and (2) it's possible that the skill exists but can't be measured with the data available.

I wonder about that last point, though. One thing that's always intrigued me about WPA as a stat is its potential to test this question of whether performance in high-leverage situations is a skill.

4. Jim Glass says:

The obvious thing to do is to just check the year-to-year...

Right. Last year to this?

5. Anonymous says:

Interesting idea. One possible problem is that EPA already has some "clutch" built into it in the form of 3rd down conversions. So this measure would favor late game clutch over 3rd down clutch.

6. Anonymous says:

Wouldn't using WPA/LI be a better way in determing "clutch"

7. Jim Glass says:

Wow, Sanchez is #3 by clutch -- which I take to be mostly luck -- and FOer recently said he laps the league in having 13 dropped would-be interceptions, far more than any other QB, which has got to be happy random chance.

As someone on another site said, if he's this bad when he's this lucky how bad is he going to be when his luck runs out?

8. Gitelson says:

Matt Ryan plays for a balanced ball control offense that does not rack up a lot of yards but tends to finish drives with points. (5th in Scoring Offense)

When things are going to plan ... Matt is more of a game manager ... But has the tools and weapons to pull out the comeback if necessary.

Also ... Even back from his BC days ... He tends to focus better under pressure. I am sure you understand that as an ex-Navy pilot.

Both the system he plays in as well as his inherent personality traits are possible reasons for the gap in statsical performance you are capturing.

Neither have to do with luck.

Incidentally ... I believe the Falcons' style of play is penalized by your predictive model since yards are traded (on both sides of the ball) to gain leverage in other aspects of the game. Also the Falcons have excellent ST which I believe are ignored in your model.