Projection Accuracy: Late March Hitter Rate Stats

I’ve been slowly working my way through the hitter projections and that journey comes to an end today as I examine how each projected hitter rate stats stand up. Besides batting average, I turn each of the counting stats into a rate by dividing by plate appearances. Finally, I adjust each value to the actual league rates. Again, any combination of projections stick out along with the BAT.

For reference, here are the projections used.

• Steamer (FanGraphs)
• ZIPS
• DepthCharts (FanGraphs)
• The Bat
• The Bat X
• Davenport
• ATC (FanGraphs)
• Pod (Mike Podhorzer)
• Masterball (Todd Zola)
• PECOTA (Baseball Prospectus)
• RotoWire
• Razzball (Steamer)
• ZEILE (Fantasy Pros)*
• Paywall #1
• Average of the above projections

To create a list of players to compare for accuracy, I took the NFBC Main Event ADP (players in demand at that time) and selected the hitters in the top-450 drafted players (30-man roster, 15 teams in the Main Event). To determine accuracy, I calculated the Root Mean Square Error (RMSE) for two different sets of values. RMSE is a “measure of how far from the regression line data points are” and the smaller a value the better. Additionally, I included the actual and league average rates for reference.

Final thoughts on hitters

• While a few projections sneak into the top-six RMSE values (e.g. Razzball with HR, Pods with Runs), the BATs and any of the “averages” (Average, Depth Charts, ATC, ZEILE) reign supreme. The six end up occupying 90% of those top-six spots. For me, I’m thinking of using some select-crowd strategy (i.e “The select-crowd strategy is thus accurate, robust, and appealing as a mechanism for helping individuals tap collective wisdom.“). Trying to go with the better performing, non-correlated sources (e.g. Pods, Razzball, and the Bat X). Some dude has a spreadsheet that helps to average several different projections.
• The BAT’s distinguished themselves as stand-alone projections, but they need a reliable playing time estimate to go with them (e.g. ATC).
• One interesting point is with stolen bases. Rotowire does their projections by hand and they moved up in the pack. There might be something to adding a personal touch to the stolen base estimates.

After the first couple of articles, there was a trend that just continued with each successive analysis, use the Wisdom of the Crowd. Not every projection can catch every nuance, but a combination stands out. I guess the same results will materialize for pitchers, but that analysis will have to wait. I am going to take a break in order to catch up on Mining the News.

* ATC, Clay Davenport, Depth Charts, Harper Wallbanger, Luke Gloeckner, Mike Podhorzer, CBS Sports, ESPN, RotoChamp, Razzball, numberFire, Steamer, THE BAT, and ZiPS.