THE BOOK--Playing The Percentages In Baseball

Tuesday, July 24, 2007

Forecasting Pitchers

Building on the work I previously released, here is some more data:

This is all pitchers
- born between 1957 and 1974
- aged 25 to 28

I put each pitcher in one of seven experience classes.

Experience n	 25-28 	 29-32 	 rate  	 Wins 
1	36	 3,745 	 2,832 	76%	 0.582 
2	72	 2,988 	 2,137 	72%	 0.521 
3	108	 2,105 	 1,553 	74%	 0.502 
4	144	 1,403 	 903 	64%	 0.474 
5	180	 912 	 628 	69%	 0.474 
6	216	 499 	 474 	95%	 0.462 
7	611	 123 	 131 	106%	 0.398 

The first line read like this: I take the 36 pitchers who faced the most batters at age 25-28.  They average 3745 batters faced per pitcher.  At age 29-32, they averaged 2832 batters each.  That age 29-32 number is 76% of the age 25-28 number.  They had a .582 W/L record.

The second line are the next 72 pitchers with the most batters faced at age 25-28.

For the first 5 lines, which comprises of the 540 most experienced pitchers (which is at least 695 batters faced over 4 years), they are all at the 70-75% rate. 

As you can see by the wins column, the quality tracks the experience, though this is hardly a surprise. 

(For you replacement-level fans, note that the win % for the last group of pitchers, the scrubs, is .398, which can be compared to the W/L record of their teammates of .484.  On a neutral team, that .398 adjusts as .408.  This is further evidence that the replacement-level for pitching is indeed .410.  This would breakdown as .380 for starters and .470 for relievers.)

Overall, the number of batters faced at age 29-32 is 75% of the total faced at age 25-28, give or take 100 batters faced, for each experience level.

So, forget about how many batters they did face.  Forget how good a pitcher he is.  The short-rule to forecasting a pitcher’s future IP (for these age classes) is simply to reduce his actual batters faced by 25% over that time period.

An annual 11% drop achieves this.  So if you have a pitcher with 3000 batters faced at ages 25-28, his next 4 years will be roughly: 668, 594, 529, 471 for a total of 2261 batters faced at ages 29-32.

***

By the way, the Marcel rule is to use 50% of the previous year, and 10% of two years ago, plus 200 batters faced.  This also works quite well.

***

Also note that if I looked at all pitchers born since 1884, their age 29-32 batters faced is 27% fewer than their age 25-28 batters faced.  In short, historically speaking, we’re not seeing much change.