SITE NEWS: We are moving all of our site and company news into a single blog for Sports-Reference.com. We'll tag all PFR content, so you can quickly and easily find the content you want.

Also, our existing PFR blog rss feed will be redirected to the new site's feed.

Pro-Football-Reference.com ยป Sports Reference

For more from Chase and Jason, check out their work at Football Perspective and The Big Lead.

The Best Player Available

Posted by Jason Lisk on April 19, 2010

If you spend any time listening or reading about the draft, you will no doubt be sick of the phrase "best player available". Writers and teams debate drafting for need versus taking the best player regardless of need. General Managers have vague press conferences where they avow to take the best player available. Part of me wants to know exactly who does not want to take the best player available--it's just identifying that player when you are comparing 22 positions that often have different skill sets that is hard. So I decided to take a look at how often teams actually do draft the best available player.

Lots of things can affect the outcome and make the best talent appear like a bad pick, from injuries to scheme to the input of the other teammates in what is a team game where individual statistics reflect more than the individual. Still, the only way I know to objectively test it is to look at the career value, and this seems like a decent exercise for applying Approximate Value.

I needed to decide on the comparison group when talking about the best available player. I suppose that I could have looked at the entire draft. I decided, though, to focus on the other players who were drafted soon after the player in question, and who might have been thought of as reasonable alternatives to the pick, from the perspective of the knowledge at the time the pick was made. I settled on looking at the next thirty selections after a player was picked. Here's what I did:

1) I used the 1967-2001 drafts (35 total years);
2) I looked at the top fifty selections in each of those drafts;
3) I used the Career Approximate Value to compare draft picks, and noted how frequently each draft slot turned out to have the highest career AV compared to the thirty players that immediately followed.

Here are the number of "best players" selected at each draft slot for that thirty-five year period. I also list the average career AV for the "best players available" actually selected at that draft slot.

pick number best player BPA pct BPA career AV
1 4 0.114 117
2 3 0.086 126
3 2 0.057 130
4 6 0.171 113
5 3 0.086 113
6 3 0.086 87
7 3 0.086 83
8 2 0.057 100
9 1 0.029 97
10 2 0.057 123
11 1 0.029 86
12 0 0.000
13 3 0.086 101
14 1 0.029 93
15 2 0.057 121
16 2 0.057 115
17 3 0.086 112
18 1 0.029 93
19 1 0.029 121
20 2 0.057 115
21 1 0.029 112
22 1 0.029 84
23 2 0.057 85
24 0 0.000
25 0 0.000
26 4 0.114 96
27 1 0.029 146
28 2 0.057 121
29 1 0.029 90
30 2 0.057 75
31 2 0.057 82
32 3 0.086 87
33 4 0.114 113
34 4 0.114 96
35 0 0.000
36 3 0.086 77
37 1 0.029 107
38 6 0.171 85
39 0 0.000
40 4 0.114 100
41 2 0.057 86
42 1 0.029 77
43 0 0.000
44 1 0.029 56
45 0 0.000
46 2 0.057 103
47 0 0.000
48 3 0.086 82
49 1 0.029 93
50 4 0.114 76

So, using my criteria (best player compared to the next thirty picks), teams drafting in the top fifty picks of the draft actually select the best player available 5.7% of the time. That may not sound impressive to you. Keep in mind that if we randomly put 31 players who were drafted consecutively in a hat and pulled one name out, we should correctly select the best player 3.2% of the time. So, it is better than what you should get if it was random. Also, some great picks may not show up on this list because of another great player from the same draft, while some less than memorable players show up as the best available in certain drafts. Bruce Smith was one of the best #1 overall picks of all-time, yet he was in the same first round as Jerry Rice. Meanwhile, Richard Todd isn't exactly remembered as an elite quarterback, but he turned out to have an (approximately) better career than those drafted after him. Let's take that table above, and bunch it into groups to smooth out some of the numbers:

picks best player BPA pct BPA career AV
1 to 5 18 0.103 118
6 to 10 11 0.063 96
11 to 15 7 0.040 103
16 to 20 9 0.051 102
21 to 25 4 0.023 92
26 to 30 10 0.057 101
31 to 35 13 0.074 97
36 to 40 14 0.080 89
41 to 45 4 0.023 76
46 to 50 10 0.057 85

The teams at the very top of the draft hit on the best player more frequently than any other group, over three times better than random chance. Those best players available also had the best career AV. After that, the numbers are up and down and all over the place on best player percentage, while the career AV's start to decline outside the first round. The two slots with the most "best players available" were the 4th and the 38th. Of course, the best players available at #4 were Mean Joe Greene, Walter Payton, Dan Hampton, John Hannah, Edgerrin James, and Russ Washington (average career AV of 113, four Hall of Famers) versus Mike Singletary, Boomer Esiason, Doug English, Levon Kirkland, Mike Rucker, and Flozell Adams at #38 (average career AV of 85, one Hall of Famer).

The Massey-Thaler finding about top five picks being the least valuable in the first round has garnered alot of attention. This data suggests, though, why teams are willing to pay heavily for top five picks. The chances of nailing the pick are higher in the top five, and the "best player available" is simply much better than the best you can usually hope for later in the first round or early second. Brian Burke has recently pointed out in discussing his gladiator versus bricklayer analogy of draft picks in regard to the Massey-Thaler study that "[p]ut simply, if you're not among the very best it matters little how good you are in absolute terms, which is what the Massey-Thaler paper measures. The nature of football puts a premium on singular excellence, and that's why the players with the highest likelihood of great success are prized so highly." Chase (in regard to top five picks at the same position) and myself (in regard to early drafted quarterbacks from the same draft) have both written about the right to choose. Consistent with what Massey and Thaler found regarding the right to choose, we found that the first choice was roughly a 50/50 proposition to actually turn out better. Nevertheless, there is value in having that right to choose, and value in an absolute sense in getting the crack at the best player early in the draft. In 29 of the 35 drafts from 1967-2001, at least one player, and many times more than one player, selected in the first round ultimately finished with a career AV greater than 100, which would equate to a Hall of Fame type career. In contrast, a team with the 40th pick would have had only nine occasions where a player selected within a round of that pick would have a career AV over 100, and then they still had to hit the pick when the occasional opportunity was there.

It is that increased opportunity to nail it and get the best that drives the high price of the top five picks, even if the quest for the Best Player Available is a difficult one. So when you are watching a GM interviewed during the draft on Thursday night, and he talks about how they took the best player that was available, you now have some idea of how likely that will actually prove to be true. To paraphrase George Kennedy from Naked Gun, they actually have a 50/50 chance of getting the best player, though there is only a 10% chance of that.

This entry was posted on Monday, April 19th, 2010 at 7:38 am and is filed under Approximate Value, NFL Draft. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.