Early Projections: Hitting Counting & Rate Stats

In my last article, I focused on the early-season projection accuracy of hitter playing time. Today, the rest of the standard 5×5 Roto hitting stats finally take center stage. Besides the counting stats, I turn each of them into rate stats to help determine projection accuracy. After completing the analysis, three options stick out.

As a reminder, here are the projections I used and some background on the analysis.

Projections

  • Steamer (FanGraphs)
  • ZIPS
  • DepthCharts (FanGraphs)
  • The Bat
  • The Bat X
  • Davenport
  • ATC (FanGraphs)
  • Pod (Mike Podhorzer)
  • Masterball (Todd Zola)
  • PECOTA (Baseball Prospectus)
  • RotoWire
  • CBS
  • Razzball (Steamer)
  • Paywall #1

To create a list of players to compare for accuracy, I took the TGFBI ADP (players in demand at that time) and selected the hitter in the top-450 drafted players (30-man roster, 15 teams in TGFBI). Generally, all the players were projected with the following exceptions. Rotowire and Pods didn’t have a Josh Rojas projection while Pods also didn’t have projections for Jazz Chisholm Jr. and Mike Brosseau. Additionally, CBS was missing several projections. I am blamed for part of it because I forgot to pull designated hitters and they didn’t project as many outfielders. Finally, I just removed the projection for Yasiel Puig.

To determine accuracy, I calculated the Root Mean Square Error (RMSE) for three 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.

  • All hitters with the three Pod and Rotowire missed.
  • Removed the three along with several who missed quite a bit of time (Lewis, Mondesi, Trout, Ozuna, Hicks, Rendon).
  • Every hitter that I pulled from CBS.

Finally, I created two additional projections. One is the average of the above projections. For the second one, I averaged the results from the best two projections. With all the theatrics out of the way, here are the results for the four counting stats (CBS didn’t include stolen bases).

2021 Early Projections: Runs
Minus missing 3 RMSE Minus missing 3 & Injuries RMSE CBS List RMSE
Average 26.2 Average 26.4 Average 24.6
Razzball & ATC 26.5 ATC & Steamer 26.7 ATC & Steamer 24.9
ATC 26.7 ZiPS 26.8 ATC 25.1
Razzball 26.7 ATC 26.9 Steamer 25.2
Steamer 26.8 Steamer 27.0 Pods 25.4
Pods 26.9 Razzball 27.1 Razzball 25.4
ZiPS 26.9 Pods 27.1 ZiPS 25.6
DepthCharts 27.4 DepthCharts 27.6 DepthCharts 25.8
Davenport 28.0 PECOTA 28.3 PECOTA 26.4
PECOTA 28.2 Davenport 28.4 Mastersball 26.7
Mastersball 28.5 Mastersball 28.7 Davenport 27.0
Bat 28.6 BatX 28.9 Bat 27.0
BatX 28.6 Bat 28.9 BatX 27.1
Rotowire 29.0 Rotowire 29.0 Rotowire 27.3
Paywall #1 31.2 Paywall #1 31.9 CBS` 29.5
Paywall #1 30.8

 

2021 Early Projections: Home Runs
Minus missing 3 RMSE Minus missing 3 & Injuries RMSE CBS List RMSE
ATC & PECOTA 9.65 ATC & PECOTA 9.70 ATC & PECOTA 9.17
PECOTA 9.74 PECOTA 9.78 ATC 9.27
ATC 9.78 ATC 9.85 PECOTA 9.31
Average 9.80 Average 9.91 Average 9.35
Razzball 9.91 Davenport 9.98 Razzball 9.52
Davenport 9.93 Razzball 9.99 Davenport 9.55
ZiPS 10.01 ZiPS 10.00 ZiPS 9.59
Steamer 10.02 Steamer 10.14 BatX 9.61
BatX 10.08 Pods 10.19 Steamer 9.62
Pods 10.10 BatX 10.23 Pods 9.64
Mastersball 10.12 Mastersball 10.25 Mastersball 9.68
DepthCharts 10.25 DepthCharts 10.37 Bat 9.76
Bat 10.26 Bat 10.43 DepthCharts 9.82
Paywall #1 10.39 Paywall #1 10.46 Paywall #1 9.82
Rotowire 10.53 Rotowire 10.55 Rotowire 9.97
CBS 11.61

 

2021 Early Projections: RBI
Minus missing 3 RMSE Minus missing 3 & Injuries RMSE CBS List RMSE
ATC 25.9 ATC 26.3 ATC 24.4
ATC & Pods 26.2 Pods 26.7 ATC & Pods 24.6
Average 26.3 ATC & Pods 26.7 Average 24.8
Pods 26.4 Average 26.7 Pods 24.8
Razzball 26.5 Razzball 26.8 Razzball 25.1
Paywall #1 26.6 Paywall #1 27.1 Paywall #1 25.3
Steamer 27.0 PECOTA 27.2 PECOTA 25.6
PECOTA 27.0 Steamer 27.4 Steamer 25.7
Mastersball 27.0 Mastersball 27.6 Mastersball 25.7
BatX 27.4 Rotowire 27.8 Davenport 25.9
Bat 27.4 BatX 27.9 Rotowire 26.0
Davenport 27.7 Bat 27.9 Bat 26.2
Rotowire 27.7 Davenport 28.1 BatX 26.2
DepthCharts 29.2 DepthCharts 29.6 DepthCharts 27.6
ZiPS 29.6 ZiPS 29.7 ZiPS 28.1
CBS 29.7

 

2021 Early Projections: Stolen Bases
Minus missing 3 RMSE Minus missing 3 & Injuries RMSE
Razzball & Bat 5.57 BatX & Razzball 5.12
Razzball 5.58 BatX 5.14
Average 5.65 Bat 5.16
Bat 5.70 Average 5.16
BatX 5.70 Razzball 5.23
Pods 5.75 Steamer 5.24
ATC 5.81 Pods 5.29
Steamer 5.81 ATC 5.34
PECOTA 5.85 DepthCharts 5.46
Davenport 6.00 Davenport 5.55
DepthCharts 6.03 PECOTA 5.58
Paywall #1 6.12 Paywall #1 5.65
Rotowire 6.23 Rotowire 5.73
ZiPS 6.26 ZiPS 5.80
Mastersball 6.72 Mastersball 6.18

 

The quick takeaway from here is that the systems that dominated the playing time analysis (ATC, Razzball, and the Average), are the top finishers. This finding should not be a surprise with playing time, along with its close cousin, injury projection, being the holy grails of fantasy analysis.

I have a few hot takes, but first, here is how the systems did at projection batting average and the counting stats as rate stats. I included CBS with the AVG but they didn’t have plate appearances in their projections. With their previous bad showings, I might just move past them.

 

2021 Early Projections: Batting Average
Minus missing 3 RMSE Minus missing 3 & Injuries RMSE CBS List RMSE
Bat & PECOTA 0.0334 Bat & PECOTA 0.0336 Bat & PECOTA 0.0332
Bat 0.0337 PECOTA 0.0340 PECOTA 0.0336
BatX 0.0340 Bat 0.0341 Bat 0.0338
PECOTA 0.0341 DepthCharts 0.0344 DepthCharts 0.0340
DepthCharts 0.0343 BatX 0.0344 Razzball 0.0340
Razzball 0.0343 Razzball 0.0344 BatX 0.0341
Steamer 0.0346 Steamer 0.0346 Steamer 0.0343
Average 0.0346 Average 0.0347 Average 0.0343
ZiPS 0.0348 ZiPS 0.0348 ZiPS 0.0344
ATC 0.0349 ATC 0.0350 ATC 0.0346
Pods 0.0358 Pods 0.0359 Pods 0.0355
Davenport 0.0370 Davenport 0.0369 Davenport 0.0367
Mastersball 0.0371 Mastersball 0.0374 Mastersball 0.0370
Paywall #1 0.0374 Paywall #1 0.0374 Paywall #1 0.0371
Rotowire 0.0395 Rotowire 0.0390 Rotowire 0.0386
CBS 0.0388

 

2021 Early Projections: Runs per PA
Minus missing 3 RMSE Minus missing 3 & Injuries RMSE
BatX & DC 0.0218 BatX & DC 0.0216
BatX 0.0219 DepthCharts 0.0219
Bat 0.0220 BatX 0.0219
DepthCharts 0.0222 Steamer 0.0221
Steamer 0.0225 Bat 0.0221
Pods 0.0226 Average 0.0224
Average 0.0226 Pods 0.0225
ATC 0.0229 ATC 0.0228
ZiPS 0.0232 ZiPS 0.0229
Razzball 0.0239 Razzball 0.0235
PECOTA 0.0249 PECOTA 0.0246
Mastersball 0.0250 Mastersball 0.0249
Paywall #1 0.0264 Paywall #1 0.0260
Davenport 0.0271 Davenport 0.0273
Rotowire 0.0285 Rotowire 0.0279

 

2021 Early Projections: Home Runs per PA
Minus missing 3 RMSE Minus missing 3 & Injuries RMSE
BatX 0.0124 BatX 0.0123
BatX & ATC 0.0125 BatX & ATC 0.0124
Bat 0.0128 Bat 0.0128
Average 0.0129 Average 0.0128
ATC 0.0129 ATC 0.0128
Steamer 0.0130 Steamer 0.0129
DepthCharts 0.0131 DepthCharts 0.0130
Razzball 0.0131 PECOTA 0.0130
PECOTA 0.0132 Razzball 0.0130
ZiPS 0.0134 ZiPS 0.0133
Mastersball 0.0136 Mastersball 0.0135
Pods 0.0136 Pods 0.0135
Davenport 0.0138 Davenport 0.0136
Rotowire 0.0140 Rotowire 0.0139
Paywall #1 0.0145 Paywall #1 0.0144

 

2021 Early Projections: RBI per PA
Minus missing 3 RMSE Minus missing 3 & Injuries RMSE
BatX 0.025 BatX 0.025
BatX & ATC 0.026 BatX & ATC 0.025
Bat 0.026 Bat 0.025
ATC 0.026 ATC 0.026
Average 0.027 Average 0.026
PECOTA 0.027 Pods 0.027
Pods 0.027 Steamer 0.027
Steamer 0.027 PECOTA 0.027
Razzball 0.029 Razzball 0.028
Mastersball 0.029 Paywall #1 0.028
DepthCharts 0.029 DepthCharts 0.028
Paywall #1 0.029 Mastersball 0.028
Rotowire 0.031 Rotowire 0.030
Davenport 0.032 Davenport 0.032
ZiPS 0.033 ZiPS 0.033

 

2021 Early Projections: SB per PA
Minus missing 3 RMSE Minus missing 3 & Injuries RMSE
BatX & ATC 0.0083 BatX 0.0082
BatX 0.0085 BatX & DC 0.0082
Bat 0.0085 Bat 0.0083
Average 0.0086 Average 0.0084
ATC 0.0087 DepthCharts 0.0086
DepthCharts 0.0088 ATC 0.0087
Steamer 0.0089 Steamer 0.0087
Razzball 0.0089 Razzball 0.0087
Pods 0.0090 Pods 0.0088
ZiPS 0.0092 ZiPS 0.0090
Rotowire 0.0092 Rotowire 0.0090
Paywall #1 0.0096 Davenport 0.0091
Davenport 0.0097 Paywall #1 0.0093
Mastersball 0.0105 Mastersball 0.0099
PECOTA 0.0107 PECOTA 0.0099

While I plan on analyzing pitchers and hitter projections from projections pulled right before the season starts, I have three possible recommendations with these results.

  1. Use ATC for all analysis except batting average component (bottom half finish) and stolen bases. Since it combines and weights the playing time from other projections, its counting stats are consistently near the top.
  2. Use the rates from The BatX with the average or ATC playing times projections. The BatX just dominates the rate stats and just needs help with playing time.
  3. Use the average of several of the “better” projections. It’s tough to beat the Wisdom of the Crowds.

An important point is that these projections are from early March. I pulled the same projections right before the season started and I will see if there are any changes to the analysis with them after I analyze early pitcher projections.

 





Jeff, one of the authors of the fantasy baseball guide,The Process, writes for RotoGraphs, The Hardball Times, Rotowire, Baseball America, and BaseballHQ. He has been nominated for two SABR Analytics Research Award for Contemporary Analysis and won it in 2013 in tandem with Bill Petti. He has won four FSWA Awards including on for his Mining the News series. He's won Tout Wars three times, LABR once, and got his first NFBC Main Event win in 2021. Follow him on Twitter @jeffwzimmerman.

newest oldest most voted
Joe Wilkey
Member
Member
Joe Wilkey

This is all based on last season only, right? Not to be a Debbie Downer, but I would have a hard time making any definitive decisions based on one season.

Also, I’d be interested to see not just RMSE for the players but a global analysis as well for each system. A lot of offensive categories were down last year as compared to the last full season of 2019, so if one system consistently undershot one or more categories for every player, it’s not necessarily a bad thing if they’re all being missed by roughly the same amount, since it’s the environment that’s being underestimated, rather than particular players being over/underestimated.

I would think you’d want to not just do a raw RMSE, but maybe do some adjustment based on population total. For example, I just pulled the top 261 hitters (approximately the # of hitters in the top 450 according to the ADP in the projections), and ZiPS overestimated RBI/PA by 0.009 for the league as a whole, which is on the order of 5.6 RBI/600 PA. I don’t know how you’d necessarily go about this, but I would think you’d want to reward the ones that are best relative to their assumed statistical environment, at least from a fantasy standpoint.

Mays Copeland
Member

To use your example for ZiPS and RBI/PA:

The hitters in this sample averaged 0.120 RBI/PA in 2021. (That doesn’t weight by PA, which it should, but it’s a start.)

ZiPS projected an average of 0.132 RBI/PA for those same players.

To correct it, you need to switch from RBI/PA to RBI/PA above the average. So Jose Abreu got 0.178 RBI/PA, which is 0.58 above the average (0.120). ZiPS projected him for 0.190 RBI/PA, which is also 0.58 above its average. You base your RMSE off of that 0.58 and 0.58.

You should double-check my math, but I think that’s the basic idea of it.