NFL Draft Theory

Like the title says, anything and everything else goes here. As long as it follows the forum rules.

Moderator: Raccoon

Post Reply
User avatar
Eigenbasis
My Pie Blown Sky High
Posts: 606
Joined: Mon Jul 27, 2009 9:07 pm

NFL Draft Theory

Post by Eigenbasis » Fri Jul 16, 2010 1:55 pm

I decided to make a new thread rather than go off on tangents in the contest thread, which is crowded already.

For my Networks class last semester, I wrote my final paper on the NFL draft. It borrows heavily from the Massey-Thaler study, if you're familiar with that. Basically, it is mathematically better to NOT have the #1 pick, or if you get it, to trade it for lower picks. The reason for this is that high draft picks are basically guaranteed to get very high rookie contracts without playing a game with no guarantee that they will perform well. Eventually, around the end of the first round, player value/salary maximizes, and those players give the most surplus. This is even noticeable in the early to mid first round. The 2004 draft is a classic example: Eli Manning went 1st, Philip Rivers went 4th, and Ben Roethlisberger went 7th. Obviously Eli got the biggest contract, but he was arguably the worst of the bunch (whether Rivers or Big Ben was better is an exercise left to the reader, but Big Ben was the best value here.

Image

This model leaves out a lot of intangibles, such as ticket sales brought in by #1 picks, but that's a short term thing and in the long term performance becomes more important.
"Have you ever heard the expression, ‘When life gives you lemons, make lemonade, and then throw it in the face of the person who gave you the lemons until they give you the oranges you originally asked for?’"

User avatar
stupac2
Oh my! Guy with Pie!
Posts: 3027
Joined: Mon Dec 08, 2008 10:04 pm
Location: Stanford, CA
Contact:

Re: NFL Draft Theory

Post by stupac2 » Fri Jul 16, 2010 2:07 pm

What I find particularly interesting is why no one in the NFL seems to recognize this. That yawning gap could close pretty easily if they just refused to pay those guys so much. It would take literally 1 season of everyone doing it to close it.

But I guess that's something of a prisoner's dilemma, and some teams would almost certainly think, "I can predict better than those other guys so I'm going to grab people up for a steal." Of course, they're only deluding themselves, but that's what would happen.


Oh, and how certain is it that the #1 really is that much bigger of a performer? Is that graph some kind of average accumulated over many years?

User avatar
Eigenbasis
My Pie Blown Sky High
Posts: 606
Joined: Mon Jul 27, 2009 9:07 pm

Re: NFL Draft Theory

Post by Eigenbasis » Fri Jul 16, 2010 2:15 pm

I think I stole that graph from the Massey-Thaler study.

Some teams have recognized this, but not many. The Colts notably traded their high picks either this year or last year (I don't remember). There's a lot of resistance from the front office to "innovate" because you get a lot more fan backlash if your new-fangled tricks fail than if your standard strategies fail.
"Have you ever heard the expression, ‘When life gives you lemons, make lemonade, and then throw it in the face of the person who gave you the lemons until they give you the oranges you originally asked for?’"

User avatar
top1214
Oh my! Guy with Pie!
Posts: 1959
Joined: Mon Mar 15, 2010 1:10 am
Location: St Louis, MO, USA

Re: NFL Draft Theory

Post by top1214 » Fri Jul 16, 2010 2:21 pm

The Patriots have also been one of the big trade downers in recent years.

It also depends on what you're looking for though. That "surplus" line is the money saved vs. playing similar performing veterans. So on average, you're better off with a draft pick than a veteran, salarywise (although some will be Ryan Leaf). The other point is, if you are looking for the biggest impact player, you want on average a higher draft pick. Limited numbers (of both players available, and players you can put on the field) dictate the higher cost.

I think these two articles explain it well:
http://www.advancednflstats.com/2009/04 ... rpaid.html
http://www.advancednflstats.com/2009/04 ... ators.html

User avatar
stupac2
Oh my! Guy with Pie!
Posts: 3027
Joined: Mon Dec 08, 2008 10:04 pm
Location: Stanford, CA
Contact:

Re: NFL Draft Theory

Post by stupac2 » Fri Jul 16, 2010 2:23 pm

One of the things that my reading has been saying is that the performance curve should be a lot more flat. Sure, a 1st round pick will normally be worth more than a 2nd round pick (especially in similar positions in the round), but by how much? And how much different is #1 from #5? Even if you consider in a position.

How good are things like the combine at predicting future success? Last thing I read on it said basically worthless, the only test that correlated with success at all was running tests, and even then only for RBs (link).

To me this just smells of people getting paid a lot of money to try and predict something that's fundamentally unpredictable, then having them codify the spaghetti they threw against the wall.

User avatar
top1214
Oh my! Guy with Pie!
Posts: 1959
Joined: Mon Mar 15, 2010 1:10 am
Location: St Louis, MO, USA

Re: NFL Draft Theory

Post by top1214 » Fri Jul 16, 2010 2:40 pm

I think part of the problem w/football and football stats/metrics is the intrinsic team nature. Baseball stats are easy: almost everything is in a bubble (each AB), and being selfish is almost always good for the team. You can see a good hitter/pitcher on a bad team. Can you see a good player on a bad football team? One guy thinks Vinny Testaverde might've been as good as Steve Young, but he never had the talent around him:
http://www.pro-football-reference.com/blog/?p=5449

I think that's one of the reasons looking at one draft pick is hard to do. When you draft a QB, you really do have to surround him with talent in order to succeed. David Carr was on his back more often than the moms of the Hand Turkery All-Stars (that's right, bringing contest drama into this thread ;-) ), but so was Aikman his rookie year. But for Aikman, they kept surrounding him w/talent until he became a HoFer.

Kelemvor
AFH
Posts: 980
Joined: Fri Aug 28, 2009 11:36 am

Re: NFL Draft Theory

Post by Kelemvor » Fri Jul 16, 2010 3:25 pm

What you say about stats in baseball is largely true for hitters, but not so much for pitchers, where there is greater uncertainty about how much of a pitcher's value depends on his fielders. And the interaction between fielders is, as far as I'm aware, completely a mystery (though I haven't followed the research in a couple of years, so maybe it's better understood now).

Anyway, the other thing to consider is volatility, right? It could be that the #1 pick not only performs better on average but also has lower volatility, which has value too.

User avatar
Eigenbasis
My Pie Blown Sky High
Posts: 606
Joined: Mon Jul 27, 2009 9:07 pm

Re: NFL Draft Theory

Post by Eigenbasis » Fri Jul 16, 2010 3:37 pm

If anything the higher draft picks would be even more volatile, I'd imagine.
"Have you ever heard the expression, ‘When life gives you lemons, make lemonade, and then throw it in the face of the person who gave you the lemons until they give you the oranges you originally asked for?’"

Kelemvor
AFH
Posts: 980
Joined: Fri Aug 28, 2009 11:36 am

Re: NFL Draft Theory

Post by Kelemvor » Fri Jul 16, 2010 3:45 pm

As I wrote that, I considered that the higher draft picks probably did have more volatility. But then I also considered that teams who own the #1 pick probably *want* volatility. If you're a sucky team with a bunch of below-average players, getting a guy who is guaranteed to be above-average probably means the difference between 6 wins and 7. Whooopdy-doo. You'd rather take a chance that a guy sucks (because, great, now you still win 6 games) to get the chance that he's awesome (so now you win 9 games, make the playoffs, can sign free agents, whatever).

User avatar
top1214
Oh my! Guy with Pie!
Posts: 1959
Joined: Mon Mar 15, 2010 1:10 am
Location: St Louis, MO, USA

Re: NFL Draft Theory

Post by top1214 » Fri Jul 16, 2010 4:09 pm

Kelemvor wrote:What you say about stats in baseball is largely true for hitters, but not so much for pitchers, where there is greater uncertainty about how much of a pitcher's value depends on his fielders. And the interaction between fielders is, as far as I'm aware, completely a mystery (though I haven't followed the research in a couple of years, so maybe it's better understood now).

Anyway, the other thing to consider is volatility, right? It could be that the #1 pick not only performs better on average but also has lower volatility, which has value too.
I think a pitcher's dependence on fielders isn't very great. If we compare Derek Jeter (a bad defensive SS) and compare him to Ozzie Smith or Luis Aparcio (considered a couple of the best defensive SSs), the difference they make is about 30 runs/season, so maybe about 4 runs while a pitcher is on the mound. SS is the most valuable densive position, so maybe ~6 runs a year can be around the limit a team's defense can contribute to a pitcher's ERA, which is ~10% of a pitcher's ERA (or ~.60 earned runs/9 innings).

Kelemvor
AFH
Posts: 980
Joined: Fri Aug 28, 2009 11:36 am

Re: NFL Draft Theory

Post by Kelemvor » Fri Jul 16, 2010 5:36 pm

If you can save 4 runs at SS, I have a hard time believing that you can only save 2 more at every other position (especially at catcher, which is considered to be harder than SS, but is rather poorly understood).

Even so, there are other variables to consider. The effect of a Jeter/Ozzie swap, for instance, would be magnified for a groundballing pitcher (handedness too, but I can never remember which hand causes more GB to the left side). A guy like Derek Lowe has a GB rate roughly three times average, and so would benefit much more from having strong infielders than a guy like Johan Santana, who has a GB rate around 80% of average.

User avatar
top1214
Oh my! Guy with Pie!
Posts: 1959
Joined: Mon Mar 15, 2010 1:10 am
Location: St Louis, MO, USA

Re: NFL Draft Theory

Post by top1214 » Fri Jul 16, 2010 6:12 pm

At SS I was comparing the all-time best vs. a guy considered terrible (by the "serious" statisticians). Realistically, you're not going to see a 30 run/year difference at SS. But yeah, it does depend on the pitcher's style.

Looking at a few people at baseball-reference.com, a gold glover at C, SS, 2B, or 3B saves about 20 runs each (a little less for 2B and C is really variable). OF and 1B are about 10 each (more for CF, less for the corners). So if you field a team of all gold glovers, you save maybe 120 runs/year, or about 0.75 run per game. It's harder (for me) to figure out who the terrible fielders are, but sampling guys that I didn't think were very good at SS, 3B, and 2B, they were about 10 runs below average, or about 30 runs difference from a gold glover. Assuming this scales to the other positions, a team of bad (but major league) fielders can cost a team maybe 60 runs/year, or about .37 runs/game.

So an average pitcher with an all-time great defensive team will have an ERA roughly 0.58 runs lower (assuming he averages 7 innings) than it should be, and on the terrible defensive team, it will be 0.29 runs higher. This even assumes that the differences are only in earned runs (plays made or not), and not related to errors. Given that the best pitchers have an ERA around 2.50-3.00, it's only 20% of the ERA that seems to be influenced by the fielding (by my back of the envelope calculations).

Raccoon
Oh my! Guy with Pie!
Posts: 1975
Joined: Thu Feb 01, 2007 10:33 am
Location: somewhere on the West Coast

Re: NFL Draft Theory

Post by Raccoon » Sat Jul 17, 2010 12:03 am

top1214 wrote:
Kelemvor wrote:What you say about stats in baseball is largely true for hitters, but not so much for pitchers, where there is greater uncertainty about how much of a pitcher's value depends on his fielders. And the interaction between fielders is, as far as I'm aware, completely a mystery (though I haven't followed the research in a couple of years, so maybe it's better understood now).

Anyway, the other thing to consider is volatility, right? It could be that the #1 pick not only performs better on average but also has lower volatility, which has value too.
I think a pitcher's dependence on fielders isn't very great. If we compare Derek Jeter (a bad defensive SS) and compare him to Ozzie Smith or Luis Aparcio (considered a couple of the best defensive SSs), the difference they make is about 30 runs/season, so maybe about 4 runs while a pitcher is on the mound. SS is the most valuable densive position, so maybe ~6 runs a year can be around the limit a team's defense can contribute to a pitcher's ERA, which is ~10% of a pitcher's ERA (or ~.60 earned runs/9 innings).
I'm not sure this is right. If you look at FIP (fielding independent projected ERA), which considers only the three things fully within a pitcher's control -- strikeouts, walks, and home runs -- you see some large variances with ERA. Chad Qualls has an 8+ ERA but his FIP is around 4. He's been abnormally unlucky this year in terms of batting average on balls in play.
[img]http://i169.photobucket.com/albums/u218/mad_hamish/raccoonsig.jpg[/img]

Kelemvor
AFH
Posts: 980
Joined: Fri Aug 28, 2009 11:36 am

Re: NFL Draft Theory

Post by Kelemvor » Sat Jul 17, 2010 1:33 am

A few things:

1. 20% is a lot! It's the difference between a league-average pitcher (4.50 ERA) and a bad pitcher (5.40 ERA) or the difference between average (4.50) and borderline all-star (3.60).

2. Also consider that defense has more of an effect for average pitchers, who generally tend to strike fewer people out, and so put more balls into play.

3. Finally, ERA is probably not a particularly good measure of pitching effectiveness, since ERA is also strongly influenced by factors like management (does your manager know when you're tired, or will he leave you in to give up 3 runs in the 7th?), relief pitching (when you put two baserunners on and leave the game, does your replacement promptly give up a 2-run double, or does he strike out the side?), and groundball tendencies (groundballs have higher error rates, so more runs that score will be unearned).

User avatar
stupac2
Oh my! Guy with Pie!
Posts: 3027
Joined: Mon Dec 08, 2008 10:04 pm
Location: Stanford, CA
Contact:

Re: NFL Draft Theory

Post by stupac2 » Sat Jul 17, 2010 1:41 am

Quick! Someone ask Nate Silver!

User avatar
top1214
Oh my! Guy with Pie!
Posts: 1959
Joined: Mon Mar 15, 2010 1:10 am
Location: St Louis, MO, USA

Re: NFL Draft Theory

Post by top1214 » Sat Jul 17, 2010 11:26 am

Blergh. I opened up a can of worms apparently.

20% is the absolute maximum. That's a team of all best defenders (Ozzie Smith, Brooks Robinson, Willie Mays, Bill Mazeroski, Ivan Rodriguez, Keith Hernandez, Barry Bonds [he won gold gloves playing LF]), and comparing that to Derek Jeter at SS, Chipper Jones at 3B, and Alfonso Soriano at 2B (2 of those 3 are no longer playing IF positions). That's just using Total Zone Total Fielding Above Replacement as listed on baseball-referencs.com

As for stats like FIP and BABIP, I don't like them too much. They treat pop ups and line drives the same (is a double that bounces off the Green Monster really "in play"?). They are also something of a luck statistic. I looked at a few great, and a couple not so great pitchers over their careers, and all the starters had almost no difference. One of the biggest differences I saw was Mariano Rivera. His ERA is 0.5 lower than his FIP (which is a lot...career ERA 2.22, FIP 2.77). And this is a guy who strikes out a lot of people. But the balls put in play are generally weakly hit (pop ups or slow rollers) b/c of the way he pitches.

User avatar
Eigenbasis
My Pie Blown Sky High
Posts: 606
Joined: Mon Jul 27, 2009 9:07 pm

Re: NFL Draft Theory

Post by Eigenbasis » Sat Jul 17, 2010 12:20 pm

You guys can make a baseball thread if you want. :P
"Have you ever heard the expression, ‘When life gives you lemons, make lemonade, and then throw it in the face of the person who gave you the lemons until they give you the oranges you originally asked for?’"

Post Reply