Comments on Advanced NFL Stats: The 2009 All-WPA Team

I read John Clayton's article about the best q...

2010-01-21T22:45:23.098-05:00

I read John Clayton's article about the best quarterbacks in the history of the NFL the other day and it has stuck with me since reading it.

People will say that it's not about numbers, it's about wins. This ignores good quarterbacks on bad teams. Others will say it's only numbers. Others will say it's about championships.

Thinking about that, I tried to get all smart today and come up with a better way to compare quarterbacks. I was thinking about using a formula with passer ratings, winning percentages, and championships. I came up with this:

(RATING) X [1. + (WP)] X [1.0(#CH WINS)]
For example, 88.9 x 1.641 x 1.02 = 148.81

I was so proud of myself but I know nothing about passer rating so I started thinking about receivers and YAC. Then I asked about sacks. I couldn't find the data I wanted so I searched and found your site. I am amazed at the data you have collected and the scrutiny you give it! This is really an awesome job on your part.

Have you ever statistically concluded who the best QB's of all time are?

I wonder if this approach could tell us the conver...

2010-01-04T01:36:14.635-05:00

I wonder if this approach could tell us the converse about defensive backs?

It would be hard to determine who the really good corners are using pure stats, since qb's simply don't throw in the direction of a good corner. However, I do believe that the play-by-play data indicates who was covering the receiver that made the catch.

Could you sum up the WPA for each completed pass a cornerback was involved in? And sort in descending order? A sort of WPA Hall of Shame list.

The both get the full credit, +0.40.

2010-01-01T13:25:31.966-05:00

The both get the full credit, +0.40.

When a receiver catches a pass that takes the team...

2010-01-01T12:23:32.893-05:00

When a receiver catches a pass that takes the team's WP from .30 to .70 - do the QB and WR both get the .40 WP added or do they split it and get .20 each?

It's primarily context. Wayne's big TD cat...

2010-01-01T11:15:25.622-05:00

It's primarily context. Wayne's big TD catch late in the Jaguar game was worth 0.56 WPA alone. He caught the TD to beat the Patriots, which was worth a lot too. Manning has lots of other targets, which robs Wayne of some EPA. But he's made big catches in critical situations to boost his WPA.

What do you think of R.Wayne? Best non-QB i WPA, b...

2010-01-01T10:59:52.740-05:00

What do you think of R.Wayne? Best non-QB i WPA, but 4. worst WR (3. if don't count J.Cribbs, or R.White) Is it a lot of tagets, or what?

These numbers look quite a bit like the numbers I ...

2010-01-01T10:18:28.005-05:00

These numbers look quite a bit like the numbers I got when I created an advanced stats-based fantasy football scoring system. The idea was to penalize players if they didn't get above about 2.5 yards per play (including targets for WRs) and to (HERE IS A BIG DIFFERENCE) count all fumbles the same whether they were recovered by their team or the other team. Check out the scoring in the league--the link is in my name.

Steven Jackson's EPA doesn't surprise me. ...

2009-12-31T19:00:26.647-05:00

Steven Jackson's EPA doesn't surprise me. He has only 4 touchdowns on 355 touches, by far the lowest rate of any major RB. He also has an average YPC and below average 1st down conversion rate. As Brian said, all the small plays add up negatively with no big plays to offset them. It probably says a lot more about the team than the individual in this case.

And, wow, for some reason my eyes skipped right ov...

2009-12-31T17:38:43.388-05:00

And, wow, for some reason my eyes skipped right over it, but Vincent Jackson was a monumental snub, way beyond what I thought was a big snub before looking into it. Jason Witten should never have been selected based on this.

My only complaint with WPA is that, and correct me...

2009-12-31T17:31:04.990-05:00

My only complaint with WPA is that, and correct me if I'm wrong, if your team immediately gains a large lead or is quickly put in a large hole, you don't have nearly the opportunity to accumulate WPA. I would think this would impact players on poor defensive teams (though I suppose if you are given the chance often to "come back" on offense you have more opportunity to accumulate WPA). Basically, I think WPA, because of its sensitivity to context far outside the players control, doesn't do as great a job of evaluating a player as EPA (of these two). That said, there is a lot to a player who makes a big play when it counts, so having a high WPA shouldn't count against a player, but I don't think a low WPA nails it either. Meh, just some quick thoughts.

The thing with RB and negative EPAs is that RB are...

2009-12-31T17:19:21.698-05:00

The thing with RB and negative EPAs is that RB are called upon to bludgeon themselves for 2 and 3 yard gains too often. Only a modest fraction of runs are actually positive. So paradoxically, the better the RB, the more often he carries the ball, and the more negative plays he accumulates.

Jackson's numbers are correct, but I think the...

2009-12-31T17:11:33.252-05:00

Jackson's numbers are correct, but I think there is a slight negative bias on the EPA numbers and a slight positive bias on the WPA numbers. It's really small, like .0002 per play or something. But it adds up over lots of plays.

I can do average per play numbers, but for this application it seems like total accomplishment is most appropriate.

WPA/LI can be done. I'm not happy with my current LI formula right now. To be honest, I don't understand what it tells us. This is my own failure, not a criticism. I'm just not that familiar.

I'd agree EPA should be a more stable, repeatable stat.

Also - should this be normalized on a per play ba...

2009-12-31T16:41:38.710-05:00

Also - should this be normalized on a per play basis?

I wonder if it's possible to do a WPA/LI stat ...

2009-12-31T15:53:09.727-05:00

I wonder if it's possible to do a WPA/LI stat for football. That would seem to be a great stat to have.

Good stuff. My first guess is that EPA is a more ...

2009-12-31T14:38:38.863-05:00

Good stuff.

My first guess is that EPA is a more "stable" stat than WPA since a couple game winning TD passes by a QB can really add up like 2 WPA. Agree?

Is Stephen Jackson at -34 EPA correct?

Shouldn't Rivers be in bold? And I knew Matt S...

2009-12-31T14:38:36.728-05:00

Shouldn't Rivers be in bold? And I knew Matt Schaub should've gotten in. Better than Brady...