THE BOOK--Playing The Percentages In Baseball

As is usually the case, you CANNOT combine FA and protected players in any kind of salary versus performance analysis…

Right, the good baseline would be to do the same with starting pitchers as well.  There are plenty of low-paid starters, not because of misvaluation, but because they only hit arbitration this year.  Take Felix for example.

The question therefore is what kind of correlation should you expect, given that the service class plays a huge role.

Tough to do in less than 300 words!

Is there a ranking out there some place that shows which relievers exceeded their value the most by how well they did in high leverage situations relative to what their normal performance would have predicted?

Yes, see the “Clutch” column in the Fangraphs link I posted in the thread.

Thanks

Thanks. That is for one year (I think). Anything long term?

You get to select “3 calendar years”, if you go to the top with the drop down.

Thanks again

At the top end, I count 6 guys who are +1.5 or more. So that means that only 6 relievers add even as much a half win per year due to their clutch performance as opposed to their normal performance

The volatility of relief pitchers has been well documented. All but the cream of the crop - the Nathans, Riveras, Hoffmans of the world - can be great one year and terrible the next.

My only problem is, when constructing a bullpen in the off season, how in the world do you know which scrap heap RP are going to be good, or which RP on the market are worth spending any kind of money on? I’d say Kevin Gregg, at 4.2 million, is at the high end of what you should be spending on a “serviceable” closer type. How do you tell the David Aardsma from the Aaron Heilman?

You project a relief pitcher the same way you project any other pitcher.  Relievers are only volatile, I think, because they only amass 60-90 IP per year as opposed to starters, at 180-220 per year 2-3 times as much.  I assume that the volatility of starters, if you looked at 60-90 IP increments, would be nearly identical to that of relievers.

Teams (the stupid ones at least) get fooled constantly by small sample size with relievers.  Plus they often look at the wrong stats in order to evaluate them.

Plus you have the tricky issue of pitchers who start and relieve.

Probably the untapped resource is simply mediocre starters who can take a run off their true talent ERA just by being relievers (short relievers that is).

But I don’t think the idea (the correct strategy) is to just use generic pitchers as relievers or never to pay a lot of money for a reliever.  The idea is to evaluate and value relievers the same way you evaluate and value any other players.  You do a projection on the player, you prorate that projection by projected playing time, and in the case of relievers, you multiply that by expected leverage.  One of the problems in valuing relievers, and a source of error even for relatively smart teams, is that they set the replacement level too low.  That will cause you to overvalue all relievers.  Most sabermetricians who have looked at this set the replacement level for relievers really high.  I think it should be.  You can take just about any of the best AAA starters, turn them into relievers and I think you have a pitcher who is only 1/2 run worse than average, per 9.  That is a VERY high replacement level.

Thanks, MGL, I think that’s a good way of looking at it. My only issue is how you evaluate those 60-90 IP bursts that are broken up by off seasons. Was there a dead arm period for a reliever the prior season, did he have a lingering health problem that wasn’t bad enough to have surgery or DL, but that 4 months of rest and rehab can fix? If a guy is a 4 RA/9 in 60 IP year 0, 3 RA/9 year 1, 5 RA/9 year 3, do you just expect him to be a 4 RA/9 going into the current off season?

Or do you look more at the particulars, instead of trying to neutralize a guy’s stats, you weigh factors like a guy being a flyball pitcher in a flyball park, groundball pitcher with porous defense due to injuries in the infield, etc. ?

I think the obvious conclusion is not to spend a ton of money on relievers, because of their fungibility, or as you put it, the high replacement level.

You look at particulars for all pitchers and all batters, if you can.  And again, I don’t think they are any more fungible than starters.  There are tons of fungible starters as well.  And, as I said, I don’t think you “just don’t pay a lot for relievers.” There are plenty of relievers who are worth a lot of money - in fact, almost every closer on every team is a 2-3 win player.  That’s worth a lot of money. If I offered Rivera, Nathan, K-Rod, Jenks, etc. for 9 mil a year, would you take them?  If your answer is yes, which it should be, then you can’t say, “Don’t ever pay a reliever a lot of money.”

Again, relievers are like any other players.  You value them and you pay them appropriately.  It is probably true that lots of teams mis-evaluate relievers for the reasons we mentioned, but that does not mean that you should make generalizations, like they are fungible or you shouldn’t pay a lot of money for them.

I find the 300 word maximum to be curious. You can’t really say much of anything in 300 words. I think it leaves the hardcore stats guys bored and the newbies devoid of important info.

For whatever reason, that’s how they want to operate.