How to Find the Next Greg Olsen at the NFL Combine
Get a free NFL subscription for 3 days.
The NFL combine is finally here! Prepare to take in a bunch of data and measurables and spit hot takes that mostly confirm your existing opinions! I’m being somewhat facetious, of course. But I think the influence of the combine and our ability to synthesize its results are moving at two entirely different paces. We’ve seen great combines move prospects way up NFL draft boards, but are we sure their results are truly going to translate to NFL and fantasy football success? I’ve already attempted to answer the question of which drills really matter for running backs and wide receivers, and now let’s turn to tight ends. Tight ends fill a hybrid position, with blocking and receiving duties both contributing to their NFL value. In fantasy football, we’re concerned only with the receiving side of the equation. I designed a decision tree to identify which combine drills/measurables are most predictive of fantasy football success (defined as at least one top-12 PPR year in a tight end’s first three seasons), and how we should consider those variables in conjunction with each other. Here’s how to read the regression tree nodes. The “yval”, or predicted value, in this case is the likelihood of success (from 0 to 1). The darker the node, the higher the yval. I plugged into the decision tree the following NFL combine measurements for tight end prospects from 2000-2013: height, weight, 40-yard dash, short shuttle, three cone, vertical, broad, and bench. The decision tree does the work of figuring out which variables are most important, and how we can classify different athletic profiles by their chance of NFL success. I also set the “minsplit” parameter to 351 to simplify the output and limit overfitting. Here are the results:
- This means that a node must have at least 35 observations to be split (back)