from Amazon
Friday, March 19, 2010
Will Mariano Rivera save only 22 games this year, and with a 3.53 ERA?
I’m quoting Rob quoting Allen Barra using PECOTA:
Scariest of all, Rivera from 44 saves to 22, and an ERA that moves from 1.76 to 3.53.
Neyer’s response:
I don’t believe Mariano Rivera will save only 22 games. I will say, too, that if your system says those things, it’s probably worth checking under the hood just in case one of the belts is running a little loose.
Just last week, I wrote a pretty good piece about what it means to forecast 31 HR for Pujols in 2009 and I concluded:
As you can see, it runs the gamut from Delgado’s 4 to Pujols’ league-leading 47. These 13 hitters were forecasted to hit a combined 401 HR in 2009. And how many HR did they actually hit in 2009? 400. That’s right, Marcel nailed it.
So, the forecasting systems work… if you know how to properly interpret what it is they are trying to tell you.
Let’s look at who Marcel has forecasted with the highest saves:
mSV ERA playerID nameLast nameFirst
33 3.55 rodrifr03 Rodriguez Francisco
30 3.07 nathajo01 Nathan Joe
29 3.20 papeljo01 Papelbon Jonathan
28 3.18 riverma01 Rivera Mariano
27 3.60 cordefr01 Cordero Francisco
27 3.69 wilsobr01 Wilson Brian
27 4.04 fuentbr01 Fuentes Brian
26 3.39 hoffmtr01 Hoffman Trevor
25 3.58 valvejo01 Valverde Jose
23 3.26 soriajo01 Soria Joakim
23 3.80 jenksbo01 Jenks Bobby
23 4.50 lidgebr01 Lidge Brad
321 3.57
Let me first tell you what this list is NOT:
1. It is NOT telling you that these 12 relievers are going to finish in that order
2. It is NOT telling you that these 12 relievers are going to finish in some order in the top 12
3. It is NOT telling you that these 12 relievers are going to end up with those exact saves
4. It is NOT telling you anything that specific about any single one of those relievers
What it IS telling you is:
1. That those group of 12 relievers will finish with around 321 saves (*)
2. That those group of 12 relievers will finish with an ERA around 3.57
3. That probably half of those 12 relievers will finish in the top 12 in saves
(*) Though with Joe Nathan already knowingly out for the season, let’s say that the remaining 11 will finish with around 291 saves and a 3.62 ERA.
Mariano Rivera has about a 50/50 chance of finishing with more than 28 saves as he has of finishing with less than 28 saves. What happened to Joe Nathan could happen to anyone. It seems to me that we should be reporting the 75th percentile forecasts for all above-average players and the 25th percentile forecasts for all below-average players. This will match the expectations of fans that don’t appreciate regression toward the mean.
But, regression toward the mean is real, and that’s why we report the 50th percentile forecasts for all players.
That said, 22/3.53 does seem low. Not absurdly low. I’ll guess that would be Marcel’s 35th percentile, if Marcel would do that. But, why not report the even more absurd Bill James forecast: 44 saves, 2.12 ERA. Rivera has 3 times exceeded that figure, and 9 times not. Presuming Mariano does not age, a 44-save forecast seems like a 75th percentile forecast. The Fangraphs FANS, who are optimistic across the board to begin with, have him with 40 saves, which probably traslates to 30-35 saves once you deflate the optimism for all players.
This is just like when I asked for the chance of Lincecum posting a below 2.50 ERA: it’s the same chance of him posting a 3.90+ ERA. This is proven by empirical evidence of similar superstar pitchers. And, the Fans have a good sense of this to also say that 2.50 for Lincecum was as likely as 3.90. And if 2.50 and 3.90 are just as likely, then you have to forecast Lincecum for close to 3.20. If you were an oddsmaker and you were taking bets and you were putting your money on the line then you’d have to forecast Lincecum’s 50/50 line to be somewhere close to that.
So, exactly where would Rob Neyer and Allan Barra put Mariano Rivera’s 50/50 line, enough that they’d take bets on either side of that line? It sure as heck is not going to be 44 saves. They’d get at least 3 times as much under action as they’d get over action. I agree that 22 saves is pretty low. Somewhere in the high 20, low 30s is right.
But, why are these articles only written on downside forecasts for great players? Why not on upside forecasts of bad players? Because, it’s the same system doing it. Bill James’ forecasts are getting a pass because he’s got optimistic forecasts for the above-average players.
And one thing I’ve learned from the Fangraphs fans voting in their forecasts, they like to forecast at the 75th percentile alot more than than 25th percentile. And so when they see a 44 save forecast for Rivera, they think it sounds about right.
But, they’re wrong.