This article is a guest post by Frank DuPont, aka the Fantasy Douche. You can follow the Fantasy Douche on Twitter where he is simply @FantasyDouche. You can also find him on his website where he writes about fantasy football and real football. Lastly, you can check out his book “Game Plan: A Radical Approach to Decision Making in the NFL” on Amazon.
The first thing we should keep in mind when drafting our fantasy team is that primarily we are trying to take advantage of the mispricing of players in the fantasy market. Ideally we want to draft a team that has more value than it actually cost in the way of draft picks. There are two main ways to approach this hunt for mispricing opportunities. We can look for individual players that are mispriced, and we can also take advantage of the mispricing of positions. In this piece I’m going to focus exclusively on figuring out whether the broad fantasy market correctly values the main skill positions of QB, RB, WR, and TE.
Let’s start by looking at how the fantasy market currently values the positions. The graph below shows the current average draft position (ADP) by positional rank. So starting in the upper left side of the chart you can see that RB1-RB3 (Foster, Rice and McCoy) are going off the board first, followed by QB1 (Rodgers) followed by RB4 (MJD) followed by WR1 (Calvin Johnson).
The important thing to pay attention to is the slope of each of the lines involved. The WR and RB slopes are fairly flat, while the TE and QB lines are fairly steep. This reflects the reality that early in the fantasy draft owners will pound the RB and WR position and will generally go for only the top tier players at TE and QB early. You can also see the formation of tiers at each position based on looking at the runs of lines that are long with no dots in between them. For instance, the TE line goes almost 40 ADP spots between TE2 and TE3 (reflecting that drafters are assigning a lot of value to Rob Gronkowski and Jimmy Graham and that there is a steep drop after those two are off the board). One other interesting thing to note is that the value of the WR and RB positions converges at about RB/WR26. By that point drafters aren’t assigning any premium to RBs when compared to WRs.
If we want to see if fantasy drafters are accurately valuing the positions we have to go through a few steps to create our own value model. Those steps are:
1. Create projections for each player for the 2012-2013 season. A number of methods would actually work to create projections. I always want to make sure that my forecasting method is going to smooth out the outlier performances like Peyton Manning’s 2004 season or Chris Johnson’s 2009 season. Those types of seasons are difficult to repeat for the same reason that the Madden Curse exists. Our forecasting method should, at a minimum, include some way to account for reversion to the mean. I like to use a similarity based forecasting system so that I can actually see players that had similar seasons to the player I’m trying to forecast. For instance, when I forecast Cam Newton’s 2012 season, I look at similar players like Duante Culpepper and Michael Vick to see what they did in seasons after they were similar to Cam Newton’s 2011 season.
2. After we create projections we have to figure out a way to compare players from different positions. I convert each player’s projection into a number that is basically the player’s projected points minus the amount projected for the last starter at a position (we’ll call this “Points Over Last Starter” or POLS). So for each quarterback I take the QB’s projected points and then subtract an amount equal to the last, or 12th, QB. I use this methodology because the risk you run when you’re drafting is that if you undervalue a position relative to most drafters, you might end up with the last viable starter at that position. You basically get the scraps. In order to account for the flex position I use the 30th RB and WR as the last starters at those positions.
3. Once we’ve projected each player’s 2012 season and converted the projection to POLS, we can then convert the POLS number back into an ADP-like value by just ranking each player. We want to make a direct comparison between our rankings and ADP in order to find potential mispricing opportunities.
First let’s look at the graph we get for the 2012 projected season based on our POLS projections.
Now let’s look at a graph that will show what happens when we convert POLS into an ADP similar number. This graph should look somewhat similar to the very first graph in this piece.
In order to see whether there are any mispricing opportunities in the market, it helps to look at both the current real ADP and our POLS based ADP on the same graph.
The dashed lines in the graph above are the valuations that ADP gives us. The solid lines are the lines that we got using POLS. Like the very first graph in this piece, the graph starts in the upper left with the first overall pick and then progresses down and right. Here are the implications of the POLS analysis in terms of positional value:
· A value strategy based on POLS means looking at the positional differences in value. We want to identify the ranges of positions that other drafters are overvaluing or undervaluing. We target the areas that other drafters are undervaluing and we let them have the areas that they are overvaluing.
· POLS says that the top tight ends are undervalued by ADP. Because POLS says that the drop off from TE1 to TE12 is steep, the top tight ends should go early. This makes sense given that drafters are currently using mid-first round picks on Calvin Johnson, while Rob Gronkowski is a close approximation of Johnson and he’s being taken in the late 2nd round. POLS says that’s a bargain.
· POLS says that the top quarterbacks are overvalued by ADP. This is the best year I’ve ever seen for executing a strategy that waits on QB. To be clear, I’m sure Aaron Rodgers and Tom Brady will tear it up again this year. But the NFL’s move to a pass happy league means that a lot of quarterbacks are going to put up great fantasy numbers. That means that the drop off at QB isn’t nearly as steep as the drop off at tight end. Drafting a QB in the lower half of the starters means that you can get Tony Romo, Philip Rivers, or Eli Manning in the 6th round.
· POLS implies that after about RB12, running backs are overvalued by ADP. This is a reflection of the reality that there are only a certain number of pass catching running backs that are in their prime. My projections assign a premium to pass catching running backs because they stay involved even when their team is getting blown out. That equates to less team related risk we will assume when we draft a running back. My projections also assign a premium to running backs that are in their prime and have a lowered risk of an age related drop off in production. POLS is basically saying that if you’re looking at running back with some question marks, like Michael Turner at RB15, that you’re better off just taking another position in that range of the draft.
· POLS says that wide receivers between about WR4 and WR13 are undervalued by ADP. If your draft strategy included targeting two of those WRs, POLS says that you would be getting value. That probably means going WR/WR in round three and four after going RB/TE in rounds one and two. I think that for this year that strategy will make a lot of sense as you’ll probably pass on the top three wide receivers in favor of two of the Hakeem Nicks, A.J. Green, Mike Wallace, or Julio Jones-level receivers.
· This analysis only focuses on getting to a full roster of starters. The next step in this analysis would be to consider reserves based on roster limits. Adding in that information could change the relative values of the positions. That’s actually an important concept to keep in mind. When we create models and projections we should always be looking for ways to improve our models and it’s also possible that improving the model will change the results. That’s a good thing.
Primarily I’ve focused on a position-centric view of value based drafting. I haven’t spent a lot of time on individual players, although the reality is that a value based drafting strategy needs to contemplate positional value and individual player value. Ideally what you would do is create a list of players that represent value, figure out where they are being drafted, and then combine that information with the positional value that we got from the POLS exercise.
It’s important to remember that the only way to consistently outperform your leaguemates in fantasy football is to focus on drafting a team that has more value than it actually cost in the way of draft picks. Considering positional value opportunities is one part of implementing that approach. It’s also more than possible that positional value opportunities could change as ADP changes leading up to the fantasy season. Stay on top of the changing value opportunities so that you can stay on top of your league.