Trading up in the NFL Draft isn't worth it (1 Viewer)

[Titans_fan]

When a First Pick Isn’t the Best Pick

By RICHARD H. THALER

The league has helped to create this problem in the way it pays rookies. A special salary cap applies to rookies and, unlike the overall salary cap the league uses, it varies across teams. The teams with the first picks get more money to spend on signing their draftees. Since agents know how much a team has been allocated to sign the first-round pick, they demand that amount for their player. As a result, compensation for draft picks declines almost in lock step; the first player gets the most, then the second pick, and so on, with the last player taken in the first round getting only 25 percent of the amount awarded to the first pick.

This is what stuck out most to me. Why doesn't the NFL get rid of the "special rookie salary cap" and makes teams stay under a hard cap? It would make the draft more boring, but the teams that lost players via retirement or free agency or releasing them would likely have more money than teams that are already stocked.This would likely still have the early picks getting paid more as the team's picking need more help, but wouldn't be so outrageous that guys can retire right out of college without ever playing a down or proving themselves.

Just a thought...

 

[Maurile Tremblay]

you need some overpaid players on your team in order to maximize team ability.

No.If you took any NFL team and replaced all of its overpaid players with underpaid players earning the same amount, the team would improve.

The problem is, you can't do what you just said. Because really good, underpaid players, aren't available.

Every player in the league was, at one point, available. A team that drafts only superstars who outperform their salaries, and no busts who underperform their salaries, would not, by doing so, fail to maximize team value. But if it helps make the point, we can replace overpaid players with fairly paid players earning the same amount. The team will still improve."Overpaid" and "underpaid" are relative terms. A player is underpaid if he plays better than the average player earning his salary. He is overpaid if he plays worse than the average player earning his salary.I think it's obvious that you don't need players who play worse than their peers in order to maximize team ability. But if it's not obvious, I could probably supply a formal proof.

 

[CalBear]

There are pretty significant flaws in their analysis.

* They're failing to use any kind of value-over-replacement-player metric. They use number of games started, and Pro Bowl participation to differentiate their populations. As everyone here should know, making the Pro Bowl isn't a great indicator of the value of a player.

* Their stat-based analysis of WRs uses receiving yards, not receiving yards over replacement value, so the value curve is wrong.

* They also fail to account for the fact that there are a limited number of starting positions on a team, so choices which maximize value in the starting lineup provide more value to the team.

Essentially, what all this adds up to is that their study is assuming that two mediocre players have more value than one great player. The games started/Pro Bowl analysis assumes that Peyton Manning is equal in value to David Garrard, and the WR analysis assumes that two WRs who average 800 receiving yards each are more or less equivalent in value to one WR who averages 1600 receiving yards.

Football has too much noise in the statistics to analyze individual player performance in the way that baseball sabermetricians do, but the basic concept of value-over-replacement-player obviously needs to be included in any analysis purporting to measure the value of players. Chris Johnson scored 347 fantasy points in 2009; that's roughly equivalent to Fred Jackson plus Cedric Benson. M+T's analysis assumes that Jackson+Benson~=Johnson, so if Jackson+Benson costs less than Johnson, you should buy Jackson+Benson instead. Or you should buy Malcolm Floyd and Pierre Garcon instead of Andre Johnson. The reality is that Malcolm Floyd is providing very little value to his team relative to a replacement player, while Andre Johnson is providing a lot. A team with Andre Johnson and a replacement-level player at WR will have a better offense than a team with Malcolm Floyd and Pierre Garcon at WR. Unless you're measuring that additional value that Johnson brings, you can't really comment on what his value is.

 

[Mahikoa]

There are pretty significant flaws in their analysis.* They're failing to use any kind of value-over-replacement-player metric. They use number of games started, and Pro Bowl participation to differentiate their populations. As everyone here should know, making the Pro Bowl isn't a great indicator of the value of a player.* Their stat-based analysis of WRs uses receiving yards, not receiving yards over replacement value, so the value curve is wrong. * They also fail to account for the fact that there are a limited number of starting positions on a team, so choices which maximize value in the starting lineup provide more value to the team.Essentially, what all this adds up to is that their study is assuming that two mediocre players have more value than one great player. The games started/Pro Bowl analysis assumes that Peyton Manning is equal in value to David Garrard, and the WR analysis assumes that two WRs who average 800 receiving yards each are more or less equivalent in value to one WR who averages 1600 receiving yards. Football has too much noise in the statistics to analyze individual player performance in the way that baseball sabermetricians do, but the basic concept of value-over-replacement-player obviously needs to be included in any analysis purporting to measure the value of players. Chris Johnson scored 347 fantasy points in 2009; that's roughly equivalent to Fred Jackson plus Cedric Benson. M+T's analysis assumes that Jackson+Benson~=Johnson, so if Jackson+Benson costs less than Johnson, you should buy Jackson+Benson instead. Or you should buy Malcolm Floyd and Pierre Garcon instead of Andre Johnson. The reality is that Malcolm Floyd is providing very little value to his team relative to a replacement player, while Andre Johnson is providing a lot. A team with Andre Johnson and a replacement-level player at WR will have a better offense than a team with Malcolm Floyd and Pierre Garcon at WR. Unless you're measuring that additional value that Johnson brings, you can't really comment on what his value is.

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

 

[CalBear]

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

Yes, I think it would change significantly. The performance curve for value-over-replacement-player looks a lot more like the draft value curve than the performance curve for receiving yards does. And frankly, I'm suspicious of anyone who's doing performance analysis without considering replacement value; this area of sports analysis is pretty well understood at this point.

 

[Maurile Tremblay]

the WR analysis assumes that two WRs who average 800 receiving yards each are more or less equivalent in value to one WR who averages 1600 receiving yards.

Can you point to which part of the paper you're getting that from?It looks to me like their equation for estimating value based on yards is non-linear.

 

[Mahikoa]

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

Yes, I think it would change significantly. The performance curve for value-over-replacement-player looks a lot more like the draft value curve than the performance curve for receiving yards does. And frankly, I'm suspicious of anyone who's doing performance analysis without considering replacement value; this area of sports analysis is pretty well understood at this point.

Can you show us that performance curve relative to the draft value curve using VORP? I'm sure a lot of us would love to see that.

 

[Maurile Tremblay]

the WR analysis assumes that two WRs who average 800 receiving yards each are more or less equivalent in value to one WR who averages 1600 receiving yards.

Can you point to which part of the paper you're getting that from?It looks to me like their equation for estimating value based on yards is non-linear.

Actually, never mind. It's almost linear. (They raise yards to the power of only 1.12.) So they are saying that a WR who averages 800 receiving yards should be paid about 46% of what a WR who averages 1600 receiving yards is paid.(Or rather, they're saying that's how free agent WRs actually are paid.)

 

[CalBear]

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

Yes, I think it would change significantly. The performance curve for value-over-replacement-player looks a lot more like the draft value curve than the performance curve for receiving yards does. And frankly, I'm suspicious of anyone who's doing performance analysis without considering replacement value; this area of sports analysis is pretty well understood at this point.

Can you show us that performance curve relative to the draft value curve using VORP? I'm sure a lot of us would love to see that.

I don't think anyone's established a Mendoza Line (replacement level) for NFL players yet. Eyeballing the WRs, it looks like you can get a replacement-level WR who can put up something like 600 yards in 16 games. At 600 yards, you're looking at Malcolm Floyd providing 176 YORP, and Andre Johnson providing 969, so a trivial yardage analysis would suggest that Johnson is worth more than five times what Floyd is worth.I did a quick chart of yardage for 2009 down to 600 yards (WR#66), and the curve looks low and linear from about WR#66 to approximately WR#32, then is more steeply linear from WR#32 to about WR#10, then is asymptotic from there. Which is about what I would expect; the top 10 are a lot more valuable than the decent starters, who are more valuable than the #2WR types.

 

[Maurile Tremblay]

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

Yes, I think it would change significantly. The performance curve for value-over-replacement-player looks a lot more like the draft value curve than the performance curve for receiving yards does. And frankly, I'm suspicious of anyone who's doing performance analysis without considering replacement value; this area of sports analysis is pretty well understood at this point.

Can you show us that performance curve relative to the draft value curve using VORP? I'm sure a lot of us would love to see that.

I don't think anyone's established a Mendoza Line (replacement level) for NFL players yet. Eyeballing the WRs, it looks like you can get a replacement-level WR who can put up something like 600 yards in 16 games. At 600 yards, you're looking at Malcolm Floyd providing 176 YORP, and Andre Johnson providing 969, so a trivial yardage analysis would suggest that Johnson is worth more than five times what Floyd is worth.I did a quick chart of yardage for 2009 down to 600 yards (WR#66), and the curve looks low and linear from about WR#66 to approximately WR#32, then is more steeply linear from WR#32 to about WR#10, then is asymptotic from there. Which is about what I would expect; the top 10 are a lot more valuable than the decent starters, who are more valuable than the #2WR types.

You're saying that they likely would get a better fit if they tried:$ = a * (x + yards)^bInstead of simply:$ = a * yards^bCorrect?

 

[Chase Stuart]

you need some overpaid players on your team in order to maximize team ability.

No.If you took any NFL team and replaced all of its overpaid players with underpaid players earning the same amount, the team would improve.

The problem is, you can't do what you just said. Because really good, underpaid players, aren't available.

Every player in the league was, at one point, available. A team that drafts only superstars who outperform their salaries, and no busts who underperform their salaries, would not, by doing so, fail to maximize team value. But if it helps make the point, we can replace overpaid players with fairly paid players earning the same amount. The team will still improve."Overpaid" and "underpaid" are relative terms. A player is underpaid if he plays better than the average player earning his salary. He is overpaid if he plays worse than the average player earning his salary.I think it's obvious that you don't need players who play worse than their peers in order to maximize team ability. But if it's not obvious, I could probably supply a formal proof.

Agreed on overpaid and underpaid being relative terms; I think to M-T, a lot of players are overpaid.I disagree with your assumption that players who are paid appropriately paid are available; these guys aren't widgets. How many QBs were available for teams to sign this year? It's not like you can get a perfectly paid guy at $500K, $1M, $2M, $3M, etc., and then overpaid guys at those levels, too. The choices might be: the best QB available who is overpaid at $3M/yr, or an appropriately paid guy at $1M.

 

[GregR_2]

Replying to

I'd be interested in seeing how any sort statistical analysis on this subject valued players like Peyton Manning, Orlando Pace, Deion Sanders and John Elway. Obviously these are players that were worth their draft position, but to have any sort of fair determination of you have to be able to assign value beyond just pass/fail. You can't just say Manning was worth the top pick overall money but Couch wasn't so that's 50/50. Because Manning represented a huge reward, far beyond just one good draft selection. People trade up to the early part of the draft to take a shot at getting a truly great player and while you can find great players anywhere the odds are by far greater with the early picks. For example, if Bradford has a 20% chances of being the next Peyton Manning and a 80% chance of being a bust, I'd still take him #1 overall if I'm the Rams. The benefits of hitting on one of those true difference making players is huge and that needs to be represented in any sort of determination of value of the early draft selection.

And

the WR analysis assumes that two WRs who average 800 receiving yards each are more or less equivalent in value to one WR who averages 1600 receiving yards.

Can you point to which part of the paper you're getting that from?It looks to me like their equation for estimating value based on yards is non-linear.

I skimmed the article and remember seeing this part that touches on these topics. YMMV whether you think it's enough or not.

A1. SuperstarsSome readers of previous drafts of this paper have worried that our results might be caused by ourfailing to capture the true value of the superstar players who single-handedly transform a team....Another reason we are skeptical is that there is considerable convexity in our performance valuesalready. We value the performance of the top percentile of players at more than twice that of players atthe 94th percentile, and in turn value those twice as highly as players at the 72nd percentile.33 This is one ofthe reasons there is no need for an additional, “elite”, performance category.

 

[Chase Stuart]

* Their stat-based analysis of WRs uses receiving yards, not receiving yards over replacement value, so the value curve is wrong. * They also fail to account for the fact that there are a limited number of starting positions on a team, so choices which maximize value in the starting lineup provide more value to the team.Essentially, what all this adds up to is that their study is assuming that two mediocre players have more value than one great player. The games started/Pro Bowl analysis assumes that Peyton Manning is equal in value to David Garrard, and the WR analysis assumes that two WRs who average 800 receiving yards each are more or less equivalent in value to one WR who averages 1600 receiving yards. Football has too much noise in the statistics to analyze individual player performance in the way that baseball sabermetricians do, but the basic concept of value-over-replacement-player obviously needs to be included in any analysis purporting to measure the value of players. Chris Johnson scored 347 fantasy points in 2009; that's roughly equivalent to Fred Jackson plus Cedric Benson. M+T's analysis assumes that Jackson+Benson~=Johnson, so if Jackson+Benson costs less than Johnson, you should buy Jackson+Benson instead. Or you should buy Malcolm Floyd and Pierre Garcon instead of Andre Johnson. The reality is that Malcolm Floyd is providing very little value to his team relative to a replacement player, while Andre Johnson is providing a lot. A team with Andre Johnson and a replacement-level player at WR will have a better offense than a team with Malcolm Floyd and Pierre Garcon at WR. Unless you're measuring that additional value that Johnson brings, you can't really comment on what his value is.

 

[Chase Stuart]

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

Yes, I think it would change significantly. The performance curve for value-over-replacement-player looks a lot more like the draft value curve than the performance curve for receiving yards does. And frankly, I'm suspicious of anyone who's doing performance analysis without considering replacement value; this area of sports analysis is pretty well understood at this point.

Can you show us that performance curve relative to the draft value curve using VORP? I'm sure a lot of us would love to see that.

I posted a link to that earlier.

 

[Mahikoa]

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

Yes, I think it would change significantly. The performance curve for value-over-replacement-player looks a lot more like the draft value curve than the performance curve for receiving yards does. And frankly, I'm suspicious of anyone who's doing performance analysis without considering replacement value; this area of sports analysis is pretty well understood at this point.

Can you show us that performance curve relative to the draft value curve using VORP? I'm sure a lot of us would love to see that.

I posted a link to that earlier.

Chase, can you please post it again so I don't have to spend a ton of time hunting for it? Thanks.

 

[Mahikoa]

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

Yes, I think it would change significantly. The performance curve for value-over-replacement-player looks a lot more like the draft value curve than the performance curve for receiving yards does. And frankly, I'm suspicious of anyone who's doing performance analysis without considering replacement value; this area of sports analysis is pretty well understood at this point.

Can you show us that performance curve relative to the draft value curve using VORP? I'm sure a lot of us would love to see that.

I don't think anyone's established a Mendoza Line (replacement level) for NFL players yet. Eyeballing the WRs, it looks like you can get a replacement-level WR who can put up something like 600 yards in 16 games. At 600 yards, you're looking at Malcolm Floyd providing 176 YORP, and Andre Johnson providing 969, so a trivial yardage analysis would suggest that Johnson is worth more than five times what Floyd is worth.I did a quick chart of yardage for 2009 down to 600 yards (WR#66), and the curve looks low and linear from about WR#66 to approximately WR#32, then is more steeply linear from WR#32 to about WR#10, then is asymptotic from there. Which is about what I would expect; the top 10 are a lot more valuable than the decent starters, who are more valuable than the #2WR types.

I'm sorry, I should clarify. I can see where you disagree with the methodology, but I'd like to see the same study done with the draft value curve - meaning, the original study, but with VORP, etc. substituted. Is this possible? You seemed to imply that you knew how to do this.

 

[Mahikoa]

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

Yes, I think it would change significantly. The performance curve for value-over-replacement-player looks a lot more like the draft value curve than the performance curve for receiving yards does. And frankly, I'm suspicious of anyone who's doing performance analysis without considering replacement value; this area of sports analysis is pretty well understood at this point.

Can you show us that performance curve relative to the draft value curve using VORP? I'm sure a lot of us would love to see that.

I posted a link to that earlier.

Chase, can you please post it again so I don't have to spend a ton of time hunting for it? Thanks.

Also, I misunderstood - so if you're just repeating what CalBear is saying about yards vs. performance, but not looking at actual draft positions, I'm not really interested. I'm interested in the proof that the original study is wrong, not just that their methodology could be better.Thanks.

 

[Maurile Tremblay]

How many QBs were available for teams to sign this year?

I don't think it matters how many were available through free agency this year. (I mean, it matters to the Seahawks. But it doesn't matter to my point.)Consider a team that already has all the QBs it needs because it drafted well in the past. That team is not suffering from the fact that its players are not overpaid. A team doesn't need any overpaid players in order to maximize its team value.No GM should ever think to himself, "Oh, no, all my players are outperforming their contracts. I need to start overpaying some of them pronto. Either that, or replace them with worse players that I can pay the same amount."As long as there's a salary cap, the key to success will involve getting the most productive players for a given number of dollars. Signing players who outperform their contracts helps achieve that goal. Signing players who underperform their contracts obstructs that goal. There's no such thing as not having enough overpaid players.

 

[SSOG]

Fair enough SSOG, I see your point. Let me be more specific. There are hits on this list and clearly Rat Face did a better job of drafting near the end versus in the middle, but that middle hurt the Broncos bad, it's why they couldn't sustain their dynasty run. Would giving more picks or higher picks from 98 to 04 via trades have made that much of a difference? ( Factor in all of Rat Face's other personnel moves as well, like importing the Browns D line and cherry picking veterans who clearly couldn't help his teams) I'm not picking on the Broncos here, I'm illustrating that having a draft pick isn't a magical elixir or having a high one isn't one either or having many many picks isn't going to save a franchise guaranteed.

I'm familiar with the argument- I've said for years that the reason Denver fell off the cliff after 2005 was because they drafted so abysmally from 2002 to 2004 and then had no veteran core. With that said... yes, absolutely, having more picks from 1998 to 2004 would have made a difference. Shanahan's drafting results from 1995-2001 (Davis, Mobley, Pryce, Neil, Greise, Wilson, Gary, Clark, O'Neal, Gold, Kennedy, Carlisle, and Anderson, just to name the regular starters) and from 2005-2008 (Cutler, Marshall, Scheffler, Dumervil, Clady, Royal, etc) demonstrate pretty clearly that Shanahan's not an inherently bad drafter. Worst case scenario, he's an average drafter who flipped a lot of "heads" early and late and "tails" in the middle... that's fine. Even his "bad free agency moves" were very much a "heads/tails" experience- Gerrard Warren wound up justifying the cost of all the Cleveland Browns all by himself, and Bertrand Berry and John Lynch were both phenomenal veteran acquisitions (as was Nick Ferguson and Champ Bailey, if you count Bailey). Give him more coins in the middle, and odds are he winds up with more "heads". There's no guarantee that he winds up with more "heads", but more coins = more chances.It's possible that Shanahan really was cursed for a 4-5 year span and could do no good, and you could give him 50 draft picks and he'd whiff on them all... but I think it's far more likely that a couple of random events wound up grouping in such a way as to suggest that they were something other than random events.

Maybe the NFL should go auction instead of draft...

Yes, but that would only increase the salaries paid to rookies.I do think it's arguable that with a salary floor and a salary cap, a draft is no longer necessary to promote parity. Some teams in less attractive cities/with poorer owners (i.e., lower signing bonuses) might still be slightly disadvantaged, but I don't think parity would be threatened (teams now have bad coaches/GMs, too).

What if, instead of auctioning with real money, they auctioned with some sort of "auction points" system? If you finish 0-16, you get 1,000 "auction points". If you finish 16-0, you get 200 "auction points". They could either come up with an "auction points" formula (i.e. every team gets 200 to start, every loss adds 50 more, maybe a bonus for missing the playoffs, etc), or else they could just come up with fixed "auction point" values for every finish from 1st to 32nd (i.e. the worst team gets 1,000, regardless of record). All players are in the pool, teams bid on them. Obviously it's a pipe dream, but I think it'd be a much better way to reward bad teams.Taking it even further, auction points could be tradeable, and you could even fix a rookie pay scale based on the amount of auction points spent to acquire a player. Say, each rookie gets a $200,000 minimum yearly salary, and every "auction point" raises it by $5,000 (spend 500 "auction points" on a guy, he gets a $2.7 million annual contract). It rewards bad teams, it simplifies trades (i.e. there's no more of that "my first rounder is too high to trade for this guy, but my second rounder is too low" syndrome), and it brings rookie salaries back under control. Plus, there's no way M-T could argue that bad teams were being disadvantaged. Bad teams could either bring in a few "studs" or a handful of bargains, depending on which strategy better suited team needs.

 

[SSOG]

you need some overpaid players on your team in order to maximize team ability.

No.If you took any NFL team and replaced all of its overpaid players with underpaid players earning the same amount, the team would improve.

I think you and Chase are looking at "overpaid" in a different sense. You're looking at "overpaid" in terms of "the dollars spent on this player could be better spent elsewhere"- from that perspective, then the post I just quoted is nothing more than a truism. Chase is looking at "overpaid" from the Kyle Orton vs. Peyton Manning perspective- "Peyton Manning makes 10 times more than Kyle Orton, Peyton Manning is not 10 times better than Kyle Orton, therefore Peyton Manning is overpaid". From that perspective, I completely agree with Chase- a team without any players who are getting "overpaid" is probably a below-average team. I mean, when Chris Johnson's rookie contract runs out, the Titans will be better off "overpaying" him than they will be letting him walk and seeking an accurately-compensated replacement.

 

[Chase Stuart]

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

Yes, I think it would change significantly. The performance curve for value-over-replacement-player looks a lot more like the draft value curve than the performance curve for receiving yards does. And frankly, I'm suspicious of anyone who's doing performance analysis without considering replacement value; this area of sports analysis is pretty well understood at this point.

Can you show us that performance curve relative to the draft value curve using VORP? I'm sure a lot of us would love to see that.

http://www.pro-football-reference.com/blog/?p=527

 

[Maurile Tremblay]

you need some overpaid players on your team in order to maximize team ability.

No.If you took any NFL team and replaced all of its overpaid players with underpaid players earning the same amount, the team would improve.

I think you and Chase are looking at "overpaid" in a different sense. You're looking at "overpaid" in terms of "the dollars spent on this player could be better spent elsewhere"- from that perspective, then the post I just quoted is nothing more than a truism. Chase is looking at "overpaid" from the Kyle Orton vs. Peyton Manning perspective- "Peyton Manning makes 10 times more than Kyle Orton, Peyton Manning is not 10 times better than Kyle Orton, therefore Peyton Manning is overpaid". From that perspective, I completely agree with Chase- a team without any players who are getting "overpaid" is probably a below-average team. I mean, when Chris Johnson's rookie contract runs out, the Titans will be better off "overpaying" him than they will be letting him walk and seeking an accurately-compensated replacement.

Thanks. I knew we had to be using terms differently. That's usually the case when somebody takes issue with what I'm pretty sure is a truism.I don't think either definition of "overpaid," however, implies that a team needs overpaid players. If an overpaid player improves sufficiently, he'll become non-overpaid under either definition. (For example, suppose Peyton Manning becomes 20 times as good as Kyle Orton.) But his improvement will only help his team, not hurt it.

I think what Chase might be getting at is that a team needs really good players -- not just really underpaid players. And he may view M-T as saying only the latter.

But with a salary cap (ceiling and floor), you can't help but have some really good players if they're all underpaid. There's no way to spend the full amount required and have your players all outperform their contracts without having some true studs in the bunch.

(Also, I don't think the "X is ten times as good as Y" idea of value is directly relevant to the M-T paper, since their measure of value is highly non-linear. As GregR points out, a player in the top percentile is worth twice as much as a player in the 96th percentile, according to M-T.)

There's another idea relevant here, which is also relevant to FF auctions: the quality of your starters matters more than the quality of your backups. If two teams have the same level of total talent, but Team A has both superstars and scrubs while Team B is entirely middling, Team A should win more games. That's something that Massey and Thaler didn't address specifically. But as a practical matter, I don't it detracts from the reliability of their conclusion that the #5 pick is worth less than the #25 pick. Those guys are nearly equally likely to be starters.

(Not that I think their conclusion is correct. I just don't think that's the reason why it's incorrect.)

 

[Mahikoa]

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

Yes, I think it would change significantly. The performance curve for value-over-replacement-player looks a lot more like the draft value curve than the performance curve for receiving yards does. And frankly, I'm suspicious of anyone who's doing performance analysis without considering replacement value; this area of sports analysis is pretty well understood at this point.

Can you show us that performance curve relative to the draft value curve using VORP? I'm sure a lot of us would love to see that.

http://www.pro-football-reference.com/blog/?p=527

So you're using career value? What about the argument that the earlier you're picked in the draft, the more chance you're given to pan out? A terrible player picked #1 might actually play for four years. A terrible player picked #240 will get cut in camp.Is that accounted for?

 

[LawFitz]

The draft no longer gives an advantage to the lower end teams in round one, where the best players typically come out. This risk/reward just doesn't seem to match unless you are VERY sure of a player's future stud potential.

Then why don't the Lions swap picks with the Jets? I'd be down with it.

If I was in a draft room picking anywhere in the top 10 - unless I was VERY high on a prospect both on and OFF the field I would trade down to the highest bidder. If the highest bid was just the next available pick or a later pick a few spots down of a similar prospect for less money, I would consider taking it.You're using the Jets to #1 overall as an extreme example and I would say it is unlikely that at least one better trade offer wouldn't be available to me as the Lions to select ahead of your proposed #28 or whatever straight up for my #1 with just salary relief as my incentive alone.The main point is that the difference in value will fluctuate wildly based on who is actually there at a given pick in a given year versus another. The trade chart doesn't account properly for the salary/risk relief of picking later in rd 1 or this random variability of the actual prospects relative to one another from one year to the next.Based on the current system I would nearly always trade down, but you better believe I'd shop around for the highest bidder since we know we don't have 32 geniuses running these teams.

 

[FUBAR]

(from the Bradford thread)

1st round (1st overall) - Peyton Manning

2nd round (32nd overall) (TRADED) FREE AGENT - Drew Brees

2nd round (33rd overall) (TRADED) - Brett Favre

1st round (24th overall) - Aaron Rodgers

UDFA (FREE AGENT) - Kurt Warner

1st round (4th overall) (TRADED) - Phillip Rivers

6th round (199th overall) - Tom Brady

UDFA - Tony Romo

3rd round (90th overall) (TRADED) - Matt Schaub

1st round (2nd overall) - Donovan McNabb

Brees was Traded?

got me. Free Agent, same idea.

 

[Chase Stuart]

Do you think the results of this study would change significantly, and would the conclusions be completely different if they used sabremetric statistics?

Yes, I think it would change significantly. The performance curve for value-over-replacement-player looks a lot more like the draft value curve than the performance curve for receiving yards does. And frankly, I'm suspicious of anyone who's doing performance analysis without considering replacement value; this area of sports analysis is pretty well understood at this point.

Can you show us that performance curve relative to the draft value curve using VORP? I'm sure a lot of us would love to see that.

http://www.pro-football-reference.com/blog/?p=527

So you're using career value? What about the argument that the earlier you're picked in the draft, the more chance you're given to pan out? A terrible player picked #1 might actually play for four years. A terrible player picked #240 will get cut in camp.Is that accounted for?

If you start for four years, you will definitely be given some credit, and more than if you never played in the NFL.http://www.pro-football-reference.com/players/P/PinnAn00.htm

Andrew Pinnock, the 229th pick for the Chargers, struck around for five seasons, but has a career value of just 4 since he barely played.

 

[FUBAR]

M-T would probably also say that Tiger Woods is overpaid. Winning the Masters brings in 4-5x as much money as coming in 5th place in the Masters. M-T would probably argue that there's no way you can justify that, and that the Masters champion was not 4-5x as good as the guy who came in 5th place, so why should he be paid 4-5x as much?

Are we reading the same article? They're talking about value to their team, paying proven players, the value/price of moving up in the draft, confidence that the player paid more is better than the lower drafted/paid player, and making the organization better through the draft. The article does mention performance per dollar, but I take that in this context that performance means how much the player helps his team win, not simply "he plays X position Y much better which is worth $Z" - maybe that is all he means, but it's still a team sport vs. an individual event. Tiger is a horrible comparison to a football draft and to hiring CEOs. I don't think they'd say anything of the sort.

The entire article focuses on player salary. M-T admit that early players are better than late players; the argument is that the drop-off in talent is not commensurate with the drop-off in price. Tournament pay, and therefore golf pay, is an apt comparison. Just like hiring a CEO is. (One difference is that in the golf scenario, we're dealing with certainties, not probabilities, but that does not change the analysis.)

Many differences, the first is like you said - certainty, the golfer WON the tournament (could argue the #3 to #6 drop off shouldn't be much) - and that is a major difference, when hiring a CEO you aren't dealing with winning an event either, just how that hire helps your organization. The other is with tournament play it's only the one person or business that matters to that entity. I assume you realize it's not always the most talented person that wins a tournament, a lot of factors come into play on that day. Individual events that pay a set amount depending on your actual performance will be seen much differently than a salary on a complete unknown, when you're paying him for his future performance. If we could pay football players for the previous year, on a set scale, then you would have a point - still a team sport, but it's closer.

 

[Chase Stuart]

The draft no longer gives an advantage to the lower end teams in round one, where the best players typically come out. This risk/reward just doesn't seem to match unless you are VERY sure of a player's future stud potential.

Then why don't the Lions swap picks with the Jets? I'd be down with it.

If I was in a draft room picking anywhere in the top 10 - unless I was VERY high on a prospect both on and OFF the field I would trade down to the highest bidder. If the highest bid was just the next available pick or a later pick a few spots down of a similar prospect for less money, I would consider taking it.You're using the Jets to #1 overall as an extreme example and I would say it is unlikely that at least one better trade offer wouldn't be available to me as the Lions to select ahead of your proposed #28 or whatever straight up for my #1 with just salary relief as my incentive alone.The main point is that the difference in value will fluctuate wildly based on who is actually there at a given pick in a given year versus another. The trade chart doesn't account properly for the salary/risk relief of picking later in rd 1 or this random variability of the actual prospects relative to one another from one year to the next.Based on the current system I would nearly always trade down, but you better believe I'd shop around for the highest bidder since we know we don't have 32 geniuses running these teams.

If you could have any pick in this year's draft -- and were not allowed to trade after getting this pick -- which pick would you choose?

 

[LawFitz]

The answer to this question is not easy b/c it requires a thorough analysis of the draft class and the other teams to make a best guess as to who will go where. I don't have enough info to say for this year but I'd imagine in most years the pick would be outside the top ten and some years where I was really high on a particular individual, it would be in the top ten.

Ultimately, I'd select the highest possible pick where I felt there was a good balance between the risk and reward associated with the players likely to be there versus the dollar committment necessary to pick in that given year.

It's a good question and the answer is a function of the talent tiers vs dollar amount tiers associated with the different sections of the round in a given year and each year could be very different from any other year. I guess it's sort of like VBD in a twisted way.

 

[twistd]

I think many of you are missing the entire point of this article. All they are saying is that every player drafted is a risk. It is a roll of the dice whether that player will turn out to be a Pro Bowl player or a bust. But to take that risk, in the top five in particular, you have to risk a lot more money in a contract. In 2009 the top five picks all commanded over $60 million in a contract. Orakpo got $20 million at 13, and everyone else except Josh Freeman, below him got less. When you commit that top five money to a rookie it takes a bigger chunk of your cap. That means you have problems resigning your own players, and it limits your ability to sign free agents. And I believe that weakens your team. Here is a list of the first round players who went to the Pro Bowl in the last ten years and what slot they were drafted in. The first two years have the contract terms:

2009

Brian Orakpo-13 5 years,20 million ($12.1M guaranteed)

Brian Cushing-15 5 years,18 million ($10.435M guaranteed)

Percy Harvin-22 5 years,14.25 million ($8.4M guaranteed)

Clay Matthews-26 5 years,$13.2 million ($7.1M guaranteed)

2008

Jake Long-1 5 years,57.5 million ($30M guaranteed)

Ryan Clady-12 5 years,17.5 million ($11.5M guaranteed)

Dominique Rodgers-Cromartie-16 5 years,15.1 million ($9M guaranteed)

Chris Johnson-24 5 years, $12 million ($7M guaranteed)

2007

Joe Thomas-3

Adrian Peterson-7

Patrick Willis-11

Darrelle Revis-14

Brandon Meriweather-24

Jon Beason-25

2006

Mario Williams-1

Vince Young-3

D’Brickashaw Ferguson-4

Vernon Davis-6

Cutler-11

Haloti Ngata-12

Antonio Cromartie-19

Davin Joseph-23

DeAngelo Williams-27

Nick Mangold-29

Joseph Addai-30

2005

Ronnie Brown-2

Braylon Edwards-3

DeMarcus Ware-11

Shawne Merriman-12

Jammal Brown-13

Aaron Rodgers-24

Roddy White-27

Logan Mankins-32

2004

Eli Manning-1

Larry Fitzgerald-3

Phillip Rivers-4

Sean Taylor-5

Kellen Winslow Jr.-6

DeAngelo Hall-8

Ben Roethlisberger-11

Jonathan Vilma-12

Tommie Harris-14

Shawn Andrews-16

Will Smith-18

Vince Wilfork-21

Steven Jackson-24

2003

Carson Palmer-1

Andre Johnson-3

Terence Newman-5

Jordan Gross-8

Kevin Williams-9

Terrell Suggs-10

Marcus Trufant-11

Troy Polamalu-16

Willis McGahee-23

Dallas Clark-24

Larry Johnson-27

Nnamdi Asomugha-31

2002

Julius Peppers-2

Bryant McKinnie-7

Roy Williams-8

John Henderson-9

Dwight Freeney-11

Jeremy Shockey-14

Albert Haynesworth-15

Javon Walker-20

Ed Reed-24

Lito Sheppard-26

2001

Michael Vick-1

Leonard Davis-2

LT-5

Richard Seymour-6

Koren Robinson-9

Dan Morgan-11

Marcus Stroud-13

Santana Moss-16

Steve Hutchinson-17

Casey Hampton-19

Nate Clements-21

Deuce McAlister-23

Michael Bennett-27

Reggie Wayne-30

Todd Heap-31

2000

Lavar Arrington-2

Chris Samuels-3

Jamal Lewis-5

Corey Simon-6

Thomas Jones-7

Brian Urlacher-9

Shaun Ellis-12

John Abraham-13

Bubba Franks-14

Deltha O’Neal-15

Julian Peterson-16

Shaun Alexander-19

Keith Bulluck-30

1999

Donovan McNabb-2

Edgerrin James-4

Ricky Williams-5

Torry Holt-6

Champ Bailey-7

David Boston-8

Chris McAllister-10

Duante Culpepper-11

Javon Kearse-16

Damien Woody-17

Antoine Winfield-23

Patrick Kerney-30

Al Wilson-31