100 Fantasy Football Tips in 100 Days, Day 27: Problems With Value Based Drafting
In my first book Fantasy Football for Smart People: How to Dominate Your Draft, I propose a version of Value Based Drafting (VBD) as my preferred draft strategy. The concept isn’t without its problems, however, as pointed out by Chris Liss at RotoWire:
(1) It’s hard to determine the real baseline for a given position.
It’s easy to use RGIII as the baseline QB, but given his injury and playing style, most people probably have him projected for only 13-15 games. As such, if you were to draft him, you’d likely get 14 games of RGIII plus two other games of the No. 20 QB, or whomever you managed to pick up that week. Michael Vick presents a similar problem. Moreover, what about the owner who drafts Eli Manning and Philip Rivers and plays only optimal matchups? Or the one who hits the waiver wire and mixes and matches every week? On the downside, some owners will get burned playing matchups, and others will see their signal callers injured in the first quarter in some games. In sum, the baseline is a moving target and so the No. 12 QB’s projected line might not be a good stand-in for it.
(2) It’s unclear whether No. 12 QB/No. 24 RB is really the right level in any event.
For the No. 12 QB to be the real baseline in a 12-team league, you’d have to have a rule that you draft your entire starting lineup before filling out your bench. That way, if you pass on Rodgers early, you’d know the No. 12 QB (Griffin) will be there for your last pick. If you pass on Peterson or some other RB for Rodgers, you know Bradshaw will be there with your last pick. Of course, it doesn’t work that way. People double up on QBs all the time, or take five RBs. By passing on a player at a given position, you’re by no means guaranteed the No. 12 or No. 24 player there. While the last-ranked starter is a rough approximation of positional depth, it’s far from perfect. For example, if RB really drops off terribly after pick No. 30, and people triple up on backs early in the draft, the price for taking Rodgers over Peterson early is certainly steeper than 17 points, so long as good QBs are around.
So what is replacement value, i.e., the baseline player, in a given league? That depends on bench size, roster requirements and owner management styles among other things. The most accurate way to get a sense of it would be to look over the league results the last five years and see what each owner got from each *slot*. The worst starting RB slot on average might be considered the baseline. Same with QB slot. Not individual players but what the owner got from his slot. This isn’t easy to do because most commish services don’t track individual RB slots, and most owners randomly move their players around between eligible slots. Finally, the FLEX spot(s) complicate this kind of calculation enormously.
(3) VBD doesn’t take into account market perceptions
Even if we were able to figure out the proper baselines with some precision and therefore had a good idea of how players should be valued based on their projected production, we also have to account for the perception of players’ value – no matter how erroneous – by the rest of our league. For example, if you’re in a league where owners overvalue quarterbacks, there might be better running backs available in the later rounds than there should be. In that case, you pay a steeper price in QB quality for passing on Rodgers in Round 1 and a less severe one for passing on Peterson than you would in a normal league. If David Wilson is available in Round 6 and Lamar Miller in Round 8, waiting on running backs suddenly makes a lot more sense. While that might sound like an extreme case, leagues do vary widely in the way they value quarterbacks and running backs even with identical scoring systems.
I think the first point is the most damning to VBD, as noted by the guys at rotoViz:
The problem is that we’re now at something of a ceiling with VBD-based thought because there’s only so far you can go with theory and eventually someone needs to design a simulation framework that will solve the VBD puzzle. It’s one thing to say that you just subtract a player’s projection from the projection of the baseline player. It’s another to know what the baseline should be. I could argue that the baseline should be based on some kind of data as to how many games each position give you, but then I’ve opened another can of worms because when we’re talking about how many games each position gives you, do we mean starter level fantasy points? If you’re talking starter level fantasy points, that’s very league size dependent. You can see how difficult this becomes and my suggestion above that only a simulation framework will work is an admission that VBD is a difficult problem.
The choice of a baseline can be arbitrary, but that doesn’t mean that VBD isn’t useful. In my view, VBD has “truth” if we look at it through a pragmatic lens.
Pragmatism is a rejection of the idea that the function of thought is to describe, represent, or mirror reality. Instead, pragmatists develop their philosophy around the idea that the function of thought is as an instrument or tool for prediction, action, and problem solving.
For our purposes, the “truth” of VBD comes in its usefulness. If it’s practical, it doesn’t necessarily matter if the terms are arbitrary.