What value firemen? (July 6, 2003)
I posted the following thought at fanhome:
Let's say the top reliever is about 2 runs / 9 IP better than average (or 3 runs / 9 IP better than replacement). That's worth about +27 runs per 81 IP over replacement.
A top reliever will have a Runs-per-win converter of about 5 or 6 (instead of the standard 10 or 11). So, that works out to being about +5 wins over replacement.
The going rate is about 1.85 million$ / marginal win, which works out to about 9 million $ / year.
More likely, the top reliever is about 2 runs / 9 IP better than replacement (please, don't just look at 1 year performance... you have to look at a player's true talent level). And the way they are suboptimally used, the RPW converter is over 6.
Working it all out again, and we see that works out to +3 wins over repl, or about 5.5 million $ / year. THAT's what their market value is. If you can use them optimally, they'll be worth over 6 million $.
A MLB regular hitter is worth about 4 million $ / year.
--posted by TangoTiger at 07:14 PM EDT
Posted 10:37 p.m.,
July 6, 2003
(#1) -
Rob H
The going rate is about 1.85 million$ / marginal win, which works out to about 9 million $ / year.
Can you elaborate on this point? I would think for a team in contention the marginal value of a win is greater than average. Ugh, that means Jason Isringhausen might be worth close to what the Cardinals are paying him. That will teach me what happens when I pay attention to anything Chris Kahrl says about the Cardinals.
Posted 2:03 a.m.,
July 7, 2003
(#2) -
MAH
Neat analysis. It also suggests a solution to the "motivational" problem of relief specialists asking (demanding) to be used suboptimally in order to rack up "saves" that they can convert to cash in the free agent market: offer to pay them more on the express condition that they will be used in a way that maximizes their IP*LI, and not their "saves".
Posted 7:14 a.m.,
July 7, 2003
(#3) -
tangotiger
According to the Voros link I put up last week, the "playoff factor" will increase the marginal $/win to 2 or 2.15 or somethign / win. It's certainly possible that his study would be improved greatly by increasing sample size, and that maybe we'll find a 1.5 million$/win for the average team, and 2.5 million$/win for teams in the hunt.
Posted 8:41 a.m.,
July 7, 2003
(#4) -
David Smyth
Slightly off-topic, but is it true that Bill James has now "admitted" that he was wrong about closers, that you need 1 guy to handle the final inning, etc. He is implying that the considerations of psychology and having well-defined roles are more important than he (we?) thought...
Posted 9:29 a.m.,
July 7, 2003
(#5) -
tangotiger
Since the media has mangled his original definition of how to use a bullpen, I wouldn't be surprised if they've mangled anything else he has said on the subject.
All I can say is that Rollie Fingers, Bruce Sutter, and Goose Gossage were fine and did not suffer any psychological depression or anxiety attacks.
Posted 12:28 p.m.,
July 7, 2003
(#6) -
Rally Monkey
"He is implying that the considerations of psychology and having well-defined roles are more important than he (we?) thought..."
It may be that pitchers in well defined roles will pitch better than those who have no idea when they are going to be called in. Some teams (like the Angels) just seem to come up with good relievers whenever they need them. It could be a lot of things, from talent evaluation, defensive support, and knowing how much to have them throw in the bullpen.
The whole idea of a closer coming in to pitch the 8th inning when its close, and letting a someone else pitch the 9th with a 3 run lead has been overdone. On average, the highest leverage situations happen in the 9th inning. If this wasn't the case Troy Percival would not have the highest leverage index in baseball. In some rare cases Percival might be better used in the 8th or earlier, but I don't think the benefit outweighs the benefits of having your players used to their roles.
If Percival is not being used optimally, in his case I don't think its a question of when but how much. Fingers threw 100+ innings every year. Gossage was usually around 80-100. Percival is more like 50-60. Scioscia doesn't use him in tie games. It hurt yesterday, as he went with Schoeneweis the Unhappy instead of Percival, who was warming up just in case the Angels took a lead in the top of the 9th.
Posted 12:39 p.m.,
July 7, 2003
(#7) -
tangotiger
I agree that the biggest difference is quantity. As I showed in an earlier article, the leverage situation of Gossage and Sutter in the 9th was pretty much what Percy and Hoffman have been getting. The difference is that Goss/Sutter have received equally high-leveraged situations in the 8th and even 7th, innings that Percy and Hoff don't see.
Posted 5:04 p.m.,
July 7, 2003
(#8) -
Vinay Kumar
Another reason that a relief ace would be more valuable to a contending team is that he can pitch proportionally more innings during the post-season. For instance, Rivera averages about 70 innnings/season in the regular season, but he has pitched 80 innings in the Yankees' 87 post-season games over his career.
I don't know what the marginal $ is for post-season wins (whether you measure by games won or series won), though. That would be interesting.
Posted 6:09 p.m.,
July 7, 2003
(#9) -
jto
Tango,
Can you elaborate on where you got some of the numbers in this discussion? Is +2runs/9 innings just a reflection of era compared to league era? How do you know the run converter is 5 to 6 instead of 10-11...is that just logic or is there some numbers to back it up?
Posted 8:56 p.m.,
July 7, 2003
(#10) -
tangotiger
Vinay, that 80/87 thing is very interesting! What's the breakdown for say Pettite?
jto: An "optimal" LI would be somewhere above 2.0. So, if the average RPW converter is 10 to 11, then a leveraged RPW converter would be about 5 for a top optimally used reliever. However, a topreliever gets an LI of about 1.7 for the most part, giving him a RPW of over 6.
Posted 10:53 a.m.,
July 9, 2003
(#11) -
Derek Zaba(e-mail)
The going rate is about 1.85 million$ / marginal win, which works out to about 9 million $ / year.
Tango, $1.85MM is better described as an upper bound rather than a best estimate of the value of a marginal win. You are assuming that the only relationship between wins and revenue is a direct causal relationship- if win, then revenue. While it is certainly true that incremental wins produce incremental revenue, it is possible that there is also a reverse causal relationship (a team with a higher revenue base due to non-winning variables will spend more money and win more games) or some other phenomenon that creates a higher correlation coefficient among wins and revenue which than is due to direct causation.
Derek
Posted 11:12 a.m.,
July 9, 2003
(#12) -
tangotiger
The causal relationships would be the following
1 - more talent+experience leads to more payroll
2 - more talent leads to more wins
3 - more wins leads to more revenue
So, if we take say the Yanks or Redsox, they do #1 (get good players), and they pay for it. At the same time, those players will do #2, and generate wins. #2 leads into #3, causing more revenue.
The causal relationships you can combine #2 and #3 to form:
1 - more talent+experience leads to more payroll
2 - more talent leads to more wins which leads to more revenue
With #2, we know that when the talent adds 1 win, it adds 2.65 MM in revenue (of which say 1.85 will be redirected to the players).
If a team spends more initially, you are talking about #1, which is somewhat related (though not in direct causal effect) to #2.
I don't see how a team having a higher base of revenue, say like the Redsox, would be able to spend their disposable income any wiser than the Blue Jays.
Posted 3:00 p.m.,
July 9, 2003
(#13) -
Derek Zaba(e-mail)
Tango,
My point is simply a correlation versus causation point. I agree that there is a causal relationship between wins and revenue (wins produce revenue, either in a single season or over the course of many seasons). I am proposing that this is the only relationship that may exist. In the example I proposed and you clarified, the Red Sox (insert other team that would generate relatively high revenue with replacement players) do not have to spend their disposable income more efficiently than the Jays (low revenue with replacement players), on average and over time, because they can just simply outspend them and win more games with the same profitability as the Jays. More generally, however, there may be other unidentified relationships that cause wins and revenue to be correlated than a simple wins produces revenue relationship that increase the correlation coefficient beyond what it would be if we separated out the causal effect of wins producing revenue.
Derek
Posted 3:12 p.m.,
July 9, 2003
(#14) -
tangotiger
Ah, gotcha.
Sure, I can buy that there may be other variables. I for one would have expected the increase to not be linear, but tied to the "fan base". That is, I would have expected that with more people to draw from that the Redsox increase in wins would be proportional to their fan base, rather than be as linear as every team. It seems that the only non-linearity (or at least the pronounced non-linearity) is tied into the playoff possibilities.
Until other variables are identified, the marginal wins x 2MM rule-of-thumb seems to be pretty good.