Friday, February 15, 2013

Finding players who are undervalued: part 1

Sorry for the long post, but I found this topic particularly fascinating and insightful for picking my team.

For this first post in a series of several, I am going to explore the idea of finding players who project to score more points per game than the average player for that price range. To do that I used my database to do a regression looking at projected points per 90 minutes (PP90) and price. Again, I just want to remind my readers that these projections are 100% based on last year's performance and adjusted as if all players played every minute during the season.

This initial chart shows when I ran the regression using all players with 1,000+ minutes. The key column to look at here is the "residuals" column. Positive numbers represent players who's performance last season is better than the average for all players of the same price in this year's game. The larger the number, the more undervalued they are. The important takeaways are:

1. Gordon and Higuain are significantly more undervalued than any other player based on their PP90 output from last season and this season's pricing.
2. Forwards tend to be more overvalued than the other two positions.
3. Defenders tend to be undervalued.
4. Midfielders seem to be more accurately valued, on average, than the other two field positions.
5. Defensive midfielders tend to be overvalued. This was the question I was originally trying to answer - general fantasy soccer knowledge says that d-mids aren't worth owning. However, with the additional options to earn points with CBI, recoveries, key passes, etc. That claim needed to be re-examined. It still holds true, in general, as many of the league's defensive midfielders have a negative residual.
6. Some of the most undervalued players also happen to be some of the most expensive in their positions - Henry, Collin, Donovan, Zusi, Bernardez. These are the types of players we want to spend our money on and find suitable cheap options for our other players (Woolard, MacDonald, Tierney, Saragosa).

I have several more posts that are related to this that I will hopefully find time for this weekend. I plan to do this same regression idea of this post except specialized for each position and sort by price to find some potentially undervalued players in each price bracket. I Also plan to explore which attribute(s) (goals, assists, CBI, key passes, etc.) is most indicative of a high PP90 for each position.

No comments:

Post a Comment