danr
Veteran
Posts: 1,871 
|
Post by danr on Apr 30, 2014 1:19:41 GMT -5
During the past two off seasons we've had some lengthy debates about roster construction and the best way to build a winning team. Nate Silver's new fivethirtyeight.com on ESPN took on this question and did some complex analysis. You can see it here: fivethirtyeight.com/features/whats-the-best-way-to-build-a-major-league-baseball-team/ and you don't have to be an ESPN subscriber. The site also is terrific for political analysis, particularly of polls. This piece examines whether it is better to have a well balanced team of moderately good players, somewhat above replacement level, or a team with a few superstars and otherwise replacement level players. The conclusion is that either method works. The advantage of a balanced team is that injuries to one or two players do not devastate the team. The advantage of the superstar-oriented team is that a key trade replacing one of the replacement level players can significantly improve the team. The Sox are cited as a team that has taken the balanced approach while the Yankees are cited as a team that has taken the superstar approach. |
|
|
|
Post by fenwaythehardway on Apr 30, 2014 7:56:44 GMT -5
Conclusion: it doesn't matter, just build a good team however you can.
|
|
|
|
Post by rjp313jr on Apr 30, 2014 8:09:09 GMT -5
The best way is to not be closed minded and think there is only one way. You need to have a philosophy but not such a strict one that you have no flexibility. For example, people fear the Sox will never spend big on a free agent again - that would be wrong for the team to dig their heels in like that. I don't believe it to be the case but some do. Likewise, I think it's wrong to just spend big because it's the best available guy (choo/Elsbury)
|
|
|
|
Post by brianthetaoist on Apr 30, 2014 8:42:26 GMT -5
There seems something slightly off about their methodology to me, although I can't quite put my finger on it. It seems to me that the question is more "should you concentrate your money in a few players or spread it around more evenly?" rather than just a question of the ex post facto distribution of production. I mean, one of the benefits of having more good players is that you maximize the chances that a few of them have particularly good years ...
Plus, WAR, by its very definition, is supposed to measure the magnitude of players' contribution to team wins, and if a study didn't come out like this, it'd be a fundamental indictment of WAR as a statistic, wouldn't it? It's supposed to be cumulative, no? If a team accumulates more WAR than another, it's by definition supposed to have more wins no matter how that WAR is achieved.
I guess, in the end, its flaw can be seen in the way it identifies the advantage of the "balanced" approach in overcoming injuries. This isn't just an incidental benefit of having a balanced team; it's nearly the entire rationale behind it. Retrospectively counting up WAR completely ignores this fact, it seems to me (a stars-and-scrubs team that lost a couple stars to long-term injury wouldn't show up as a stars-and-scrubs team in this methodology). And month-to-month WAR totals seems like such a noisy statistic that I can't even wrap my head around why anyone would expect a correlation to anything from it. By that logic, one of the biggest stars in the Yankees stars-and-scrubs approach is Yangeris Solarte (second on WAR for April).
I like Jonah Keri, but this seems off to me. Maybe if he started with the distribution of career WAR/season of the players on the 40 man roster on Opening Day, or expected WAR from projection systems or something.
|
|
ericmvan
Veteran
Supposed to be working on something more important
Posts: 8,932
|
Post by ericmvan on Apr 30, 2014 14:40:26 GMT -5
I like Jonah Keri, but this seems off to me. Maybe if he started with the distribution of career WAR/season of the players on the 40 man roster on Opening Day, or expected WAR from projection systems or something. Yes. A simple Marcel of WAR on the 40-man roster, and then look at actual WAR as a function of projected total WAR and the distribution of WAR. That's your study. Also, look at the next year projected and actual WAR to see if there are some WAR distributions that are easier to improve than others. |
|
