Using Historical Distribution Scores to Visualize and Quantify Ranges of Outcomes

This article introduces Historical Distribution Scores. It breaks down what they are, how they are calculated, and why they are relevant when considering a player’s range of outcomes. Articles focusing on each position and the calculated scores will be posted in the coming weeks.  At RotoViz, we often preach the importance of considering a player’s range of outcomes. How good can Devin Funchess be in a fantastic season? If he struggles, how bad of a season are we looking at? These are important questions and considering a player’s floor and ceiling is a major part of assembling strong rosters.

Developing Ranges of Outcomes

So how do we develop ranges of outcomes and frame our expectations? One way is to build historical projections. For a detailed walkthrough of this process, please check out the Stat Explorer. In short, historical projections use a player’s average stat line from season ‘n’ (including only the stats that are predictive of fantasy scoring and carry from year to year) to find a list of players that compiled similar numbers in seasons past. The performance of these players in their subsequent seasons, year ‘n+1’ is then used to build a range of outcomes and assign low, median, and high projections for the player in question.

Lost Context

While I really enjoy building projections like these and reviewing the results, one thing that has always bothered me is a little bit of context is lost. Two players could have similar low and high projections but the distribution of the points generated in year n+1 by their comparable players could be considerably different. To visualize this, let’s take a look at Jarvis Landry and Julio Jones.

Subscribe to the best value in fantasy sports

Subscribing is the only way to make sure you don't ever miss an article.

By Dave Caban | @DaveCabanFF | Archive