rIP*Lev/sIP

If Joba’s expected leverage is 1.5 in the 8th, and you expect him to throw twice as many innings as a starter, then his relief value should be 75% of his starting value. If Joba is a 3.8 win starter, then he should be a 2.9 win reliever, not 2.4.

Here are a couple of issues I have with the formula you use to determine value.

1. Why aren’t pitchers done like hitters? It seems inconsistant to compare LF/CF one way and SP/RP another.

2. Why do you use .470 as the winning% for replacement relievers? It seems to me that a replacement reliever should have a winning% of .380, compared to an average reliever of .500.

3. What is the true exchange rate between the two positions? 1.16 runs seems awfully high. The only other time I’ve ever seen anybody mention the exchange rate between SP and RP it was stated at like 0.20 or something. That’s too low for me to swallow. There is a big enough sample of guys that have done both in the same season that we should know just how much harder it is to start than relieve.

4. How do you calculate leverage? My guess is you do it when the pitcher enters the game, which seems inherently incorrect. It should be done on a batter by batter basis. Just because a guy created his own high leverage situation doesn’t mean the game is any less on the line. At the very least, I’d like to see the results. My guess is a starter will have an average leverage of something like 0.90…certainly under 1.0.

5. Say you pitch to two batters, one is a 0.50 leverage (half) and one in a 2 leverage (double). Is your average leverage 1 or 1.25?

With Joba, you estimate 150 IP, but don’t tell us how he’s going to get those 150 IP. There are three ways, and all have different value and risk;

1. He could start 32 games and throw 150 IP, an average of 4.7 per start. The average major league starter throws ~5.8 per start. So you want to replace those innings with 1.1 of replacement level relief. The issue with that is, there is a bullpen tax. Guys who can only go 4.7 IP/start end up in the pen. If they didn’t, the replacement level for starter would be much better.

2. He could start 23-25 games with a normal workload, getting to 150 IP. This makes a lot of sense, although it does push us one step closer to a six man rotation. The real issue here is, if the plan is for 150 IP, he’s still going to get injured some % of the time, so his true EV would be less than 150 IP.

3. He could start 32 games with a normal workload getting him to 200 IP, and he loses about 25% of his innings to injury. If it were a one year thing, this would be the route I’d go. When you have to account for five years, loss of future effectiveness is a big concern.

Then there’s the Joe Torre factor. Starters are given well regulated workloads, whereas relievers are left to the whims of the manager. Very few players can survive the workload Torre puts on his favorite relievers.

Say Hughes is +2 wins next year, you get +2 from him in the rotation and +2.3 or so from Joba in the pen, if that… meaning you are ranging from 3.9-4.3 wins.

With Joba in the rotation you get +3.8 wins, and even if you have a solid 8th inning guy without Joba’s dominance, he could still produce +1 win… netting you +4.8 wins. And if that other 8th inning guy is worth +1.5 wins, then this scenario produced an entirely new win.

a. he’s A LOT better as a RP than SP … see Mo, Hoffman, Gagne
b. he’s too fragil , see Kuo
c. you happen to have 5 SP better than him healthy in the lineup.

a and c are pretty much proven to be untrue, b might be, but it’s hard to see him as possibly benig worse than Kuo, and even in Kuo’s case they tried numerous time to use him as a SP.