Summation of 2021 Projection Accuracy

Earlier this offseason I examined the accuracy of several projections. I’m going to give my recommendations today on the best way to balance those findings between getting the best results and keeping the process simple. I’m going to focus on playing time and stats for pitchers and hitters. While there could a way to weigh every single stat of every projection, it’s just a waste of time in my opinion. The best answer is to aggregate the best options.

I know some people will want a more in-depth answer while the following will be too much for others. Some projections, like ATC, are already trying to perfect the mix and still fall short of a straight average. The cause for the disconnect is that some of the stand-alone projections are constantly improving. What may be the best projection mix in one season is suboptimal in the next. I’m willing to have 95% of the projection accuracy and instead spend my time looking for information that the projections might have missed.

First, here are the original articles.

• Projection Accuracy: Early March Plate Appearances
• Early Projections: Hitting Counting & Rate Stats
• Early March Hitter Counting & Rate Stats
• Late March Hitter Counting Stats
• Late March Hitter Rate Stats
• Early March Pitcher Counting Stats
• Early March Pitcher Rate Stats
• Late March Pitcher Counting Stats
• Late March Pitcher Rate Stats

In those articles, the following projection systems were used. Some of the projections are a single source while others are an aggregate of several projections.

Single source projections

Steamer (FanGraphs)

• ZIPS
• THE BAT
• THE BAT X
• Davenport
• Pod (Mike Podhorzer)
• Masterball (Todd Zola)
• PECOTA (Baseball Prospectus)
• RotoWire
• Razzball (Steamer)
• Paywall #1

Projection aggregators

• DepthCharts (FanGraphs: Steamer and ZIPS)
• ATC (FanGraphs)
• ZIELE (Fantasy Pros)*
• Average of all the above projections

And the following is how I interpret and would utilize the results.

Pitcher Playing Time

For the playing time stats, Razzball owned the competition with Pods as the only stand-alone consistently near the top. Not surprisingly, all four aggregators stayed near the top but were probably being dragged down by some of the off projections.

Pitcher Rate/Counting Stats

The Average owns this category with a clean sweep once the rates are adjusted to the league scoring environment. The other aggregators made a decent showing. Of the stand-alone values, both of the Steamers (Razzball and FanGraphs) stood.

The one exception I would include is to consume as much information on Saves as possible. Who gets Saves is entirely manager driven so there is no way to project them. I would look at all available options and dive into any disagreements.

Overall Pitching

The simple answer is Razzball or any of the aggregators (ATC, ZEILE, or Depth Charts). Their results held up in all categories.

If I was to aggregate my own set, I’d go with an even weighting Razzball, Pods, and ATC (possibly ZiPS – Depth Chart).

Hitter Playing Time

For playing time, three of the aggregators, Average, ZEILE, and ATC shoved in this category (Depth Charts takes a hit because it only uses one playing time input). It’s an easy win for the Wisdom of the Crowds.

Hitter Rate/Counting Stats

For the stand-alone projections, THE BAT’s stand out (I’d use THE BAT X). Just below THE BAT are all four aggregators.

Overall hitting

THE BAT X since it’s handling its playing time weakness by “stealing” playing time estimators from ATC. Otherwise, I would go with ATC or ZEILE.

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