The Marginal Value Of A Win
This post is inspired by a question Sky Kalkman raised on twitter. Specifically, he wondered about the wisdom of the Rays paying $7 million for Rafael Soriano, when the franchise has proven time and again that they are capable of finding bargains for below-market rates, eschewing the going rate for wins and building a contender on a shoestring payroll.
I like the move and think it was a good one for Tampa Bay, even though it’s not an efficiency maximizing move from the perspective of the price of wins. They’re going to pay Soriano about $4 to $4.5 million per win for 2009, which is certainly not a discount, and is actually above the going rate that we have seen for most players signed this year. Additionally, the $7 million he will earn would represent nearly 11 percent of the total team payroll from 2009. That is a significant allocation of resources to give to a closer.
However, the Rays are not in a league average situation. Their position in expected outcomes is quite a bit different than most teams. They have the talent of a contender, but share a division with New York and Boston, which drives down their chances of making the playoffs. As such, they have to protect themselves from variance more than most clubs will, as a bullpen implosion (like they had in 2009) can essentially derail their chances of playing in October.
Soriano adds a win or two to the roster, which may not sound like much, but the win that he’s bringing is extremely valuable, given the precarious nature of the Rays playoff odds. By adding a premium relief ace, they’ve insured, to an extent, against a disaster. The security that he brings doesn’t offer the same reward for the dollar that you may find by taking a flyer on a young, unproven, power arm, but the Rays don’t need more upside as much as they need less downside.
We talked a bit about this a few weeks ago, but the composition of a team’s talent and their relation to their division opponents can have a pretty significant effect on their internal marginal value of a win. A win to the Rays is significantly more valuable than a win to the Astros because of the respective effect of that win on the odds of either team making the playoffs.
Because of where they stand, it makes sense for the Rays to pay the market rate for wins, because that price is lower than the value they’re getting from that win. It does not make sense for the Astros to pay the market price, because the return they will get on those additional wins is below the going rate. If Tampa Bay and Houston pay the same price for the same player, it will be a good deal for the Rays and a bad deal for the Astros.
We cannot use the price of wins in the market as the guide for what every team should be willing to pay for a win. It will never deviate too tremendously from the average (no one, not even the Yankees, should pay $10 million per win for a player), but teams do have their own internal marginal values of a win, and they won’t equal the going rate of wins in the market for all 30 teams.
Tampa Bay should be insuring against bad outcomes at the expense of the best one, which is what they did with Soriano. It was the right move, even if it doesn’t offer the opportunity to be a huge bargain.
24
This is great stuff, David. Well done.
I agree completely, Dave. Teams around contention can afford to pay close to the market rate for additional wins, but teams like the Astros need to develop ways to pick up below-market wins because they happen to be so far off of contention.
They could have had Horacio Ramirez for a lot less money, and he’s just as good as—if not better than—Soriano.
What I don’t understand is why Soriano, or any relief ace for that matter, is only expected to contribute 1-2 WAR. Why are all runs treated equally when the situations they are saved in are clearly not.
If your relief ace is 20 RAR shouldn’t that be worth significantly more than two wins vs. a 6th inning relief pitcher who is 0 RAR?
Or put another way if you flip their roles would not the 6th inning reliever cost your team significantly more wins than the relief ace would provide?
For instance Jonathan Papelbon pitched 68 innings for Boston and contributed 19.8 RAR for 1.9 WAR. Manny Delcarmen pitched 59 innings with 2.5 RAR for .2 WAR.
If their roles had been switched I find it hard to believe that Boston would have only lost 2 extra games.
Given the typical usage for a relief ace, doesn’t it make sense to provide additional value to the RAR they save since a run saved is considerably more valuable in the 8-9th innings of a 1-3 run game vs. a run saved in the 6th inning when down by 4?
I may be wrong, but isn’t this where leverage index comes in to play? I might be completely off-base here.
The value of a closer is so overrated.
They only pitch in leads. What are the chances of a game being decided strictly in the 9th inning? 1/9, if you don’t count extra innings. What about relievers who pitch out of jams in the 6,7, and 8th innings? Isn’t run prevention more valuable in a bases loaded tie game than in the 9th inning with a 1-3 run lead? In fact, I think it is more valuable to pitch your best relievers in tie games than when given leads.
Look at Brad Lidge: He posted one of the worst seasons for a closer in baseball history and yet, he was still able to save 31 games.
“If their roles had been switched I find it hard to believe that Boston would have only lost 2 extra games.”
Well, I don’t. Even Joe Borowski managed to save 45 games in 2007 for the Indians.
Besides the cases of Rivera and Hoffman, I don’t see why closers need to be given the extra hype.
“Look at Brad Lidge: He posted one of the worst seasons for a closer in baseball history and yet, he was still able to save 31 games.”
But he also blew nine or ten. They might have won 100 games with a solid closer. And he had a negative WAR as well.
In any case, the original commenter was making a point about WAR for closers not reflecting their true value, and I am inclined to think that he was right.
A player in a “reverse Lidge” situation is Ryan Franklin …
In 08, Franklin was a big part of the BP that led the league in bullpen losses and blown saves … and it cost StL dearly.
In 09, Frnklin was much better, and was one of the main reasons why StL won the division.
’08 — 6 L — 17 SV — 8 BS — -0.4 WAR
’09 — 3L — 38 SV — 5 BS — 0.9 WAR
It’s kinda difficult for me to wrap my mind around the idea that Franklin’s difference in performance was only worth 1.3 WAR.
Franklin reduced his WHIP by .2 points (BIG improvement) and lowered his HR/IP numbers from 7.8 to 31 (IP per HR), and that made all the difference.
Another value one has as an effective closer is that the team has ONE guy that it can rely on, versus using the “committee” approach. Now, I don;t know how that translates in terms of actually “wins”, just that it’s a “comfort/reliability” type thing, and factors heavily in how the BP is used in games and throughout the season.
I think moneyball and BP have done a decent job educating the masses as to how closers *can* be over-rated and over-valued, but we also have cases of guys that were good as a setup man, but were unable to duplicate the performance once given the “closer’s” job … and we also see how effective reliable closers are in comparison to teams that suffle closers around, so there has to be *some* type of increased value for pitchers that do actually produce in the role of a closer.
This is so dramatic given the leverage of the situation and/or how we perceive it. A reliver in the 7th could lose the game for his team by giving up the go ahead run, but it doesn;t leave the immediate impact of a guy giving up a walk-off game-losing run, as a blown save is most probably a “loss”, when it was a “very probable win”.
Closers just seem to be a really “funny” situation. One thing I am more certain of is that pitchers don;t always perform the same in the role of closer as they do in the role of “reliever”, even though statistically we make that assumption (or at least I think we do).
Yes, but…
http://www.beyondtheboxscore.com/2009/4/29/856308/bullpen-chaining-and-reliever-war
You are correct about leverage index. Multiply the WAR by the LI and you get a better picture of value.
LI is included in WAR.
It does? I’ve never heard that before. I thought WAR was designed to be context-neutral.
Thanks for clearing that up, Dave.
I agree about Tampa – they must pay up in order to get to the top of the division again, but I’m not sure about Houston. Sure, they could do almost as well for much less money, but so what?
I’m not sure I like the train of thought that goes – “no realistic chance of making the playoffs, so why bother with (often overpriced) veterans?”. It seems like a recipe for 10-15 teams to cut bait in the offseason and provide essentially hapless opposition in the hopes that a few youngsters each year turn out to be good. There are not huge waves of untapped minor league potential that can be good major leaguers out there (possibly relief pitching aside). If some teams aren’t putting up a good fight and essentially planning to win 70-80 games, they league becomes pretty much like upper level european football (4-5 teams dominate a league, others just try to not get relegated).
For example, Toronto has been fighting the “good fight” for most of the last decade, putting together good/mediocre teams that were doomed to miss the playoffs unless things went incredibly well, and frankly, that’s preferable to the decade the Rays had, at least to me.
If every mediocre/bad team rebuilds at once, you get a lot of terrible baseball and occasional lightning in a bottle. I guess I’m saying the someone has to go after those “useless” wins between great draft position and marginal playoff contention. There’s value to baseball as a whole to poor teams making marginal improvements (I’m not saying Houston has managed that, but rather that they are intending to do so).
Dave is not saying teams shouldn’t be chasing those “useless” wins, he’s saying they shouldn’t be overpaying in pursuit of them. It’s only worth overpaying when you’re likely to get a non-linear result from doing so, ie making/not-making the postseason. Otherwise, you should be paying market rate (or less, which is quite possible when you’re starting with a very bad team so even low-value players may be an upgrade).
You’re missing the next phase of the analysis here.
Decreased demand for free agents (when rebuilding teams don’t need them anymore) will decrease prices of free agents, which will make more free agents “bargains,” which will cause more teams to sign them.
Ultimately all the same guys will be playing baseball, except for a handful of uber-fringy Bobby Crosby type free agents who will be unable to get jobs at all.
If you want to reform the system so that players still get paid just as much but rebuilding teams will be willing to pay for free agents and playoff contenders won’t snap up every good dude on the market, you could subsidize and/or penalize signings based on team W/L records, so that, say, a team that won 60 last season only has to pay $2M per win while a team that won 100 has to pay $6M per win.
A great break down, Dave. This is why I’m on Fangraphs multiple times per day.
I agree witht the concept, but does Soriano, given his injury history, really represent much insurance against bad outcomes? Granted he’s an elite closer when he’s on the mound, but how much confidence can you have that he’ll be on the mound throughout the season?
I was thinking the exact same thing – this also must have been how they convinced ownership.
I think it would be very very interesting to track the price payed per WAR over time. Perhaps creating an index or looking at year over year changes. I’m interested in inflation/deflation in win value. I wonder how that tracks with other forms of inflation – cpi, wage inflation etc etc. This would be a valuable thing to keep track off over time. No?
Here you go: http://www.fangraphs.com/blogs/index.php/win-values-explained-part-six
Dave, I don’t even know why you needed to waste time writing this.
I mean, it’s just obvious.
It’s unfortunate that you need to write an entry explaining to Ed Wade why it’s ridiculous for a team to overpay a player when he won’t make a difference (a playoff difference at least).
I can’t believe there are fans out there that defend the Lyon signing. Even worse is that Wade is still a GM…
This isn’t about the Astros. Even if Dave used them as a foil. I’m sure he was looking for “the obvious opposite to Tampa in competitive position and acquisition philosophy” and the Astros, given recent events, were easily the first team that came to mind, but let’s not derail things to far in Ed Wade’s direction — (there’s a whole other thread for that).
But anyway, this (excellent) article is about how the context of team changes its calculations, and how that context itself changes as the team climbs the curve out of the cellar and through respectability to contention. One of the hardest thing for GMs (and fans) to do is realistically assessing where a team actually is — as we saw with (to pick a team further back the curve) the Mariners in 2007, when Bavasi ignored the M’s pythagorean results and gave away most of the Pacific Northwest to Baltimore to get Bedard in the belief that the team was on the cusp of contention (rather than just lucky). That would’ve been a bad trade under any circumstances, but the idea that you have to pay more to get wins when you’re ready to go to the postseason is sound. You just have to be clear-eyed about whether you’re really ready. And what you should be paying.
And unfortunately, this is not obvious to a lot of fans. Fans live on hope, which — while both necessary and admirable — doesn’t always lead to sensible expectations. The good news for Astros fans is that it’s easy to get much better quickly when you’re starting from a low point (just look at the 2009 Mariners), but you have to know what you’re doing. And know where you are on that curve. And that’s not necessarily obvious.
It’s hard enough to convince a lot of people that there’s a fairly consistent price for a marginal win in the market that teams pay, and that it actually applies to everyone, pitcher or a position player, utility guy or star — we see people pushing back at this idea almost every day in the comments section here at Fangraphs. To then suggest that it makes sense for teams to pay over that amount under certain circumstances is going to strike some people as hand-waving or a sort of have-it-both-ways relativism. It’s human nature to prefer simple, “Golden” rules that never change and apply to every circumstance, and our inner child — or inner Joe Morgan — tends to try to simplify things to get there. Which is why easy-to-calculate stats like ERA and RBI remain popular, and why the “conventional wisdom” about so many things retains such a hold. Part of growing up is discovering that the world is more complicated than we thought and that there are subtler rules governing the simple rules that we ignore at our peril. But not everybody has got there yet.
So while it may be preaching to choir here, articles like this are both good and necessary. And I know I’m going to be linking to it in discussions I have elsewhere.
Dave – I’m curious if the marginal value of a win is actually less for the Yankees if Cashman figures that they are a going to be most likely over 100 wins barring a disastrous string of injuries, and that the Red Sox and Rays need 2 or 3 premier players to get to 100 wins? For example, what is Holliday worth to them at this point? Sure you don’t want him to go for a bargain so that the Red Sox can sign him and Beltre and Lackey, but given the holes as they are now, the Red Sox tendency not to exceed the luxury tax threshold by much, and their contracts, how likely is it that the Red Sox can make 3 signings of that quality? Similarly, is the talent difference between the NYY and the Rays enough so that signing Holliday is a needed boost to push the NYY over the top?
Cashman is partially worried about making the playoffs, and also about how they will do once they are in the playoffs. So he also thinks about how the Yankees will do against Anaheim, Boston, Philly, St Louis, etc, in a playoff series. Some of the extra spending is probably aimed at that.
Great article – very thought provoking. I do have a couple of questions.
First, it looks like the Rays basically traded Iwamura, a 2-3 war player when healthy, for Soriano, a 1-2 war player when he stays on the field. Iwamura would have cost $4.85M this year, and Soriano will now make $7.25M. Doesn’t that seem like a very high premium to insure against downside?
Dave, I’m also reminded of your post on 11/20 regarding diversification. For a team in the Rays financial situation, diversifying would seem to be the smarter move: spreading the risk among a variety of capable players would keep costs low (unlike Soriano) while limiting downside (like Soriano).
Which all makes me wonder if this was more about upside than downside (sorry if this is simply becoming a semantic issue). In Soriano the Rays have a player they can plug in towards the back end of the pen and let work until he breaks down or becomes ineffective. They basically don’t need to play around for a month with two or three players to find an effective reliever; instead, they lock in a 1-2 war player from the outset of the season. I guess I’m saying Soriano is simply much more expedient.
Even with that said, I wonder if the premium was too high for the Rays.
The cost of downgrading from Iwamura to Soriano is the arbitrage cost involved in moving wins from one part of the roster to another.
The Rays had a billion infielders and no good relievers. Even though the number of potential WAR on the roster is now smaller, the number of ACTUAL WAR they receive should be larger, because Iwamura couldn’t really be squeezed into their lineup next season.
Well, that seems like a steep transaction cost. $2.5M for the salary alone, and between .5 and 1 WAR, or $2-4M. $5M for the privilege of getting one less win and paying a player more money? Is that the marginal value of a win for the Rays? At that cost, wouldn’t it make more sense to look to the market for a different reliever?
It seems more likely that the Rays made a misstep somewhere between trading Iwamura and trading for Soriano. This position seems to be reinforced by Dave’s previous article about diversifying, unless I’m missing something, which I may be.
I have a question about a closer’s value. In high leverage situations, the pitcher’s performance has a greater effect on a win or loss than in low leverage situations. If pitcher pitches well in high leverage situations, we makes the team win more games than if he pitched well in low leverage situations. But, should the player get credit for when he is put in? If there are two equal pitchers, but one is put in higher leverage situations, should the one put in higher leverage situations have a higher WAR? It seems to me that the player value should be leverage independent because he has no control over the leverage; the leverage depends on the coaching staff’s decision.
Dave you say that LI is included in WAR but yet it doesn’t show up in the numbers.
In 2009 Mariano Rivera is listed at 20.1 RAR and a WAR of 2 despite a pLI of 1.72. By comparison David Roberston is listed as 7.3 RAR a WAR of .7 and a pLI of .68.
I don’t see where LI is showing up in WAR at all. Could you help me understand what I am missing?
This is from the “Rivera and the Value of a Closer” on Fangraphs on Dec. 4th.
“This line by Rivera is projected to be worth 2.4 WAR. So then is 2.4 WAR truly as good as a reliever can be? Probably not. That’s mostly because this measurement doesn’t account for leverage index. This certainly can be done, and it is done in some places.”
Maybe Dave is saying it can be included in some WAR calculations, but I also always thought it was not included in Fangraphs’ WAR. Jamie’s numbers above sure make it look like it is not included.
I have been wondering the same thing as you Jamie in regards to leverage increasing WAR. In your above post, could the calculation for RAR already have Li calculated into it?
Following this logic, the Marlins really should spend the $30 million to keep Uggla and sign both Matt Holliday and Jose Valverde.
http://gameofinches.blogspot.com/2009/12/marlins-need-matt-holliday_28.html