There's been an inordinate amount discussion about the best draft strategy. Most of the discussion is about whether teams are best served by drafting BPA or for need. In my opinion BPA versus need is really a question of how you balance short-term benefit against long-term benefit (i.e. do you draft for the short term or for the long-term).
1st let's get some consistent terminology. In economic terms we are talking about marginal utility. Marginal utility is the benefit that's derived from one incremental unit. This contributor at advanced NFL stats writes about the concept of marginal utility in a coherent manner.
In plain English about football, this means that if your team has two all-pro wide receivers and zero competent linebackers, and you have a choice between adding a third all-pro receiver or an all-pro linebacker, you'd better go for the linebacker. Adding your first good linebacker will give more net points to your team than adding a third top receiver
No one is disputing the above quote. That raises a question: if everyone agrees with the above statement what is it that people disagreeing about? I believe there are 2 things. First, people disagree about the correct balance between short-term benefit long-term benefit. Second, people disagree about the value of the current players on the roster. This analysis is focused on the balance between short term and long term.
For clarity instead of talking about utility (which is a theoretical concept) I'm going to assume that utility is equal to the change in points that a player produces. For example, the average NFL team will score approximately 400 points per season and will allow approximately 400 points per season. That average team would have a record of 8-8. When I talk about utility I'm going to talk about how much a player would change either total points scored or total points allowed for a season.
For example, let's assume that the Dallas Cowboys are an average NFL football team (not crazy since they were 6-10 last year). Furthermore, let's assume that all the players on the Cowboys are average players except for the following:
-Dez Bryant: 25 points over an average player
-Miles Austin: 25 points over an average player
-Marc Colombo: -50 points below an average player
-Leonard Davis: -25 points below an average player
-DeMarcus Ware: 50 points over an average player
-Alan Ball: -25 points below an average player
Adding up all the pluses and minuses the net points over average is 0 (this actually is not accurate as the offense was above average and the defense was below average … just ignore that). This should produce an 8-8 team.
Now let's consider a couple draft scenarios. Assume that both Von Miller and Tyron Smith are available with the number 9 pick. Assume that Von Miller is in fact equivalent to DeMarcus Ware and is worth 50 points over an average player and that Tyron Smith is worth 25 points over an average player. Using these assumptions Von Miller is the BPA. However one can make a case that Tyron Smith is the better draft pick for Dallas. That case rests not on the value over an average player but the value over the player that is being replaced (in this case I'm going to call that value over substitute).
Let's take a look at a comparison of Von Miller's value over substitute (where he’s substituting for an average Anthony Spencer, 0 points over average) and Tyron Smith’s value over substitute (where he’s substituting for a below average Mark Colombo, 50 points below average). I've assumed that both players have ten-year careers and perform at a constant level over the course of their careers.
This looks great for Tyron Smith. Tyron Smith produces 750 points over substitute over the course of his career while Von Miller only produces 500. Tyron Smith should be the pick.
However it also makes explicit the implicit assumptions that the need-based argument rests on:
-That the value of a draft pick should be evaluated based on their impact in the short term only, and/or
-the value over substitute should be evaluated as if the existing substitute is the alternative for the draft pick’s entire 10 year career.
For example, if we assume that instead of Marc Colombo the substitute for Tyron Smith after year 1 is an average player (i.e. 0 points over average), then the value over substitute for Tyron Smith's career dramatically changes. Here we assume that Tyron Smith would be replacing Mark Colombo in his 1st year but would only be replacing an average player in all subsequent years.
Now Von Miller provides almost twice as many points as Tyron Smith over his career.
I've noticed that the initial scenario, that a draft pick should be evaluated against the existing substitute for their entire career is the position than many people who favor drafting Tyron Smith are making. What I've seen argued is that because the Dallas Cowboys don’t have depth on the offensive line and the free-agent market is thin there is no opportunity for Dallas to acquire even an average offensive lineman and this situation will continue in perpetuity. My personal opinion is that is assuming the worst possible case to back into the conclusion they want to reach. Dallas has been successful in acquiring free-agent offensive lineman who are satisfactory. In addition, Hudson Houck has now been in Dallas 3 years and in that time has developed Doug Free. In my opinion over the long term the most reasonable base case assumption is that your substitute would be an average player. You have injuries, retirements, free agency, player development, etc. You also get nonsensical results, like if AJ Green slipped to be 2nd round it’d be better to take James Carpenter because he replaces Colombo. Again, this shows that BPA versus need is really about finding the right balance between short term and long term benefit. The longer period you consider the more sense it makes to assume an average player as the substitute. I leave it to the reader to make up their own mind about what the appropriate substitute assumption is.
Finally, I’d like to tie this back to 2 posts from OCC:
The underlying assumption of a BPA approach is that drafts are not a short term fix for a couple of team weaknesses, but that the draft is the cornerstone of long term franchise building and team strategy. Over the long run, BPA should in principle give you the best team possible, but injuries, free agency and draft busts can have a significant impact on this. The downside obviously is that a BPA approach does not address team needs. That is why God (with a little help from the NFLPA) invented free agency.
Think of the teams you usually associate a BPA approach with. The Patriots. The Colts. The Packers. The Steelers. Perhaps the Eagles. Maybe a few more. What do all these teams have in common? Over the last few years, they usually haven’t had that many holes on their roster. They had the luxury of mostly following relatively stringent BPA approaches, whereas most teams drafting in the top 15 every year do not have that luxury. Instead, these weak teams must address very specific needs or holes in their rosters and have to resort to at least a ‘BPA at position of need’ strategy, if not an outright need strategy, to fill those holes.
"We always try to draft the best player available, and the best example I can give is Reggie Wayne," he said. "We had a big need for a cornerback that year, but we could not agree on who that should be. So we turned around, traded down, and took Reggie Wayne, who was the best player available at that pick. And it turned out to be the right thing."
"And it always is, you should take the best player. You might be wrong in the assessment of the player, but as long as you take the best player your odds of success are very much greater."
I think the 2nd quote is a possible explanation for the first quote …
What do all these teams have in common? Over the last few years, they usually haven’t had that many holes on their roster.
They don’t have holes because "as long as you take the best player your odds of success are very much greater". Improving your odds leaves you with fewer holes, which gives you the luxury of drafting the best player. Again, short term benefit against long term benefit. Anyway, this probably could have been said in 100 words instead of 2,000 but sometimes it is useful to be precise so we can see the underlying assumptions explicitly.
Have a good weekend.