While discussing , Joe over at River Ave Blues said something very interesting that I believe is very important for the Yankees:
While I’ve soured on WAR for a number of reasons, one big reason is that it understates the value of high-WAR players. That is, players worth seven wins over replacement per season are exceedingly rare. They are worth far, far more than double a 3.5-win player, because 3.5-win players are far more common. To take this further, a team with one 7-win player and one 0-win player is in a far better position than a team with two 3.5-win players. The first team can replace the 0-win player, but the second team will have trouble finding reasonable, and reasonably priced, upgrades over the 3.5-win players.
Now, for the sake of the discussion let’s assume that WAR is a perfect number – this isn’t fWAR or bWAR or anything, just an abstract concept of wins over replacement player. Which situation would you rather be in? A 7 WAR player and a replacement player, or two 3.5 WAR players? Or, more concretely, would you rather have 2011 and (0.2 WAR) or (3.4 WAR) and (3.6 WAR) over the course of one season? It shouldn’t matter really. In the abstract, both should be worth exactly the same in terms of winning. 7 wins is 7 wins.
Now, the real world is a little more complex than that. If you offered me the choice between players that on average are worth that much at the beginning of the season, I’d actually pick having the 3.5 WAR players on my team every time, because you’re hedging against risk. If each of the four players have equal chance of being injured, then you’re in a much riskier position by taking the 7 WAR player – because chances are the team has a replacement player at Triple-A waiting for a call up. That said, there might be an advantage to the 7 WAR player, especially if they are a pitcher, with what I call the “short bench effect” – the ability to allocate more playing time to better players in higher leverage games or situations, like the playoffs. Either way, the difference is tiny.
That’s all well and good, but Joe alludes to a larger point: teams can go out and get more players. If Carl Crawford is playing at replacement level, the Yankees can go swing a trade for an upgrade without trouble. That’s not so easy with the 3.5 WAR player. And since they are the New York Yankees, they can go out and buy someone on the free agent market if they want to. Or take a bad contract back. Or trade away from player they signed for $3 million out of Venezuela. In fact, the Yankees have the best capacity in baseball (over a long period of time at least) to seek upgrades to inferior players. That means that they benefit more from the 7 WAR player than any team in baseball.
There are huge real-world implications for that statement. It means the Yankees should be looking for a lot more than Cito Culvers on draft day than other teams. It means that they should be be patient on the free agent market for the really, really good players out there, and that they should be able to offer those free agents much more money. It makes those 3.5 WAR type players expendable, and good options to be traded. I’ve been advocating for trading , in the abstract at least, for some time now on that principle. The Yankees could trade Swisher, a 3.8 WAR player last year, and then replace his production on the free agent markets while benefiting from the trade.
And at the same time, it means the Yankees could theoretically make a Win-Win trade with a team like the Mariners for a player like Felix Hernandez. Seattle has less ability to go out and buy upgrades for their players. That means that they could benefit from trading away Felix Hernandez, a 7 WAR player, and acquiring, say, three players worth 3 WAR. The Yankees would replace two of those players with free agents at roughly equal production, while Seattle just improved by 2 wins. Both teams benefit from the Quality-for-Quantity trade. This is pretty close to the paradigm that brought Curtis Granderson to the Yankees.
If you enjoyed this article, please consider sharing it!
EJ, I have to disagree strongly. There may be some circumstances where 7 wins is 7 wins. Most of the time, playing time means you want to concentrate your WAR in as few players as possible. The extreme of this logic would be, would you rather have nine 7 WAR players or eighteen 3.5 WAR players? Since eighteen people can’t all play regularly, I’d take the nine 7 WAR players every time. If your goal is to win one World Series, you’re better off maximizing upside and hoping for overall good health in a given year. I suppose if you held a gun to my head and said I had to win the World Series three out of three years, I’d at least consider hedging some of my bets and praying some the 3.5 WAR players have career years.
If you read the whole post, I’m not sure we disagree.
EJ,
This is a great post. Those experienced with Fantasy Football are aware of this concept.
In my league, a person offered to trade me two players who each average 12 points per week for somebody of mine who averages 20 points per week. They were curious why I rejected their “generous” offer.
In this scenario, if I accepted, I would have to play both of those guys each week. Why is that bad? I would have no room for locating an up-and-comer or a hot hand. I’d be stuck with 24 points no matter what.
However, if I keep my 20 points per week guy, I know that I can easily find either a free agent or somebody on my bench who is worth at least 7 points per week (as 7 points seems to be what the average replacement generates). Therefore, by rejecting this offer, I keep myself from losing 3 points, because the two positions currently on my roster are worth at least 27 points per week.
The same concept applies to a high payroll team such as the New York Yankees. Unlike say, the Mariners, the average replacement for the Yankees is *not* worth 0.0 WAR. The Yankees can afford to overpay for a free agent or gamble on several players and see who performs the best. Furthermore, their strong farm system means that, depending on the position, their actual minor league replacement is going to be worth more than 0.0 WAR.
WAR is great for comparing the value of two players and answering the questions, who’s better. However, WAR alone is a poor at measuring a player’s worth to a given team.
In the hypothetical in which you gave a choice of 1 7.0 WAR player and a replacement player or 2 3.5 WAR players for a year, I’d take the two 3.5 WAR players because you have less injury risk by spreading the WAR and less ability of your star being pitched around assuming it’s a position player or having his value concentrated in few games if he’s a pitcher. If you don’t put the limitation that I have to keep the players for a year, I’ll take the 7.0 WAR player and the replacement player so that I can more easily make an upgrade and improve the team.
I think the point Joe was making if we read between the lines is that the value of marginal increases in WAR shouldn’t be calculated in a linear way. It’s less a problem with WAR itself and more a problem with the simplistic way in which people calculate surplus value and trade value.
I agree with you though that Seattle would be better off not having so much WAR concentrated in one player, especially not in a starting pitcher who can only make so many starts in a year. You add to that the fact that Seattle finished 29 games out of first place and isn’t likely to make the playoffs anytime soon and the value of an elite starter to a playoff team and I conclude that King Felix is wasted on the Mariners and that they should trade him because they would make the team a lot better by doing so. I don’t think they view it that way but they should.
Agreed, I was just thinking about the linearity concept myself.
Agree with everything you said. That was the point I was trying to make – and Ari (comment above) sums it up perfectly: Replacement level for the Yankees is not 0.0 WAR. They have the resources to shift replacement level higher than the average, sometimes dramatically higher.
David Cameron makes an excellent point:
http://www.fangraphs.com/blogs/index.php/linear-dollars-per-win-again/
It’s an interesting discussion. Cameron brings up the topic of risk avoidance vs. the exponential value of marginal war and states that a lot of teams have a strong inclination towards risk avoidance. It’s not just an issue for some teams, it’s true for all teams.
To a large extent, a salary structure that values WAR in a linear fashion is already acknowledging that marginal WAR has exponential value because if you are paying a salary based on a 7.0 WAR, the range of possible outcomes isn’t a bell-shaped curve. You are far more likely to have a player age poorly and significantly underperform expectations or get hurt and provide nothing than you are to have a guy who is already a superstar elevate into a 10.0 WAR player (especially over a long term contract) and that risk has to be built into a player’s compensation. Conceptually, you might be paying a premium for incremental WAR but then applying a large discount for risk. Thus, I think surplus value calculations based on linear value still underestimate the values that superstar performers provide because they ignore the pricing of risk into those contracts.
I’m not sure if I explained that well or not. Does that make sense?
Conceptually, another way to think of this is that Maximizing total team war is likely very highly correlated with Regular Season Success. However, maximizing WAR from the 8 regular position players, the DH, the top 3 starting pitchers, and the top 3 relievers should be more highly correlated with post season success. Thus, the trade of superstar with high WAR from a poor team for 2 or 3 players with more total WAR from a playoff contender could be mutually beneficial.