One of the great things you can do with the Similarity Score apps is that you can create custom adjustments with very little effort. I primarily use Similarity Scores when I project players for the coming fantasy season. But one thing I’ve struggled with in the past is how to treat players that will see usage changes. The custom features on the app make quick work of that problem.If I select CJ Spiller’s name from the drop down list (or just start typing “Spiller”) I can then uncheck all of the boxes for games where he didn’t see at least 10 carries. The app will throw those games out of his season long averages, which will change the list of comparables players, and thus change the projection for 2013.
Before we do that though, let’s look at the baseline, which is the projection created before we customize it. Here is the N+1 tab for Spiller if we include all of his 2013 games. The app projects him to score about 11.5 standard points/game, which would be good, but not great. For comparison purposes, the app projects Doug Martin to score 15 points/game.
If we then want to create a custom projection, we just uncheck all of the games where Spiller didn’t get at least 10 rushing attempts. I think that’s a reasonable way to think about his usage going forward as the Bills should want to get him 10 carries a game.
When we do that we get the following projection by looking at the N+1 tab. The similar players changed enough that the yards per game projection increased by 20 yards per game and the touchdown estimate increased as well. Based on this estimate, Spiller would have a top five projection among running backs.
Using the custom feature of the Similarity Score app, it’s possible to dramatically improve an estimate for CJ Spiller, but do it in a way that remains grounded in the idea which drives Similarity Scores.