September 9, 2014

Baseball Therapy

Poisoned by Losing?

by Russell A. Carleton

Printer-friendly

Contact Author

Well, the Astros made news again last week by firing manager Bo Porter. According to media reports, which is the adult way of saying “I heard it from Larry in algebra class,” it was due to disagreements between Porter and general manager Jeff Luhnow. There seems to be a lot of that around nowadays. The Astros will break their streak of consecutive 100 loss seasons this year, but there’s no doubt it’s been rough for the past few years on the Gulf Coast of Texas. Luhnow is now the wampeter of the baseball karass (a billion points for that reference) and, strangely for such a mild-mannered guy, might be the most polarizing figure in the game today. Whether it’s fair or not, a narrative has developed around the Astros that their ruthless rebuild, while it may eventually prove effective, is an exercise in masochism. Sure, they needed to be rebuilt and rebuilds involve pain, but aren’t they playing Breaking Madden with a real live baseball team?

On an intellectual level, the Astros' rebuild makes perfect sense. Luhnow inherited a team with a barren farm system and a bunch of veterans in decline on the MLB side. Trading away veterans for prospects happens all the time in these situations. Many teams will at least sign some veteran free agent cover to say they’re making a run at .500. Luhnow didn’t—and specifically made a point of not doing it—and the Astros finished with the worst record in the majors three years in a row. (In fairness, the Astros did sign Scott Feldman and traded for Dexter Fowler last offseason.) They collected the draft and international bonus pools that went with that record. As a reward, the farm system now looks pretty good. And in an ecosystem where it makes sense to build from within, why spend money on fading free agents who will just push you lower in the draft order and bring you less pool money to swim in? Why throw money at a season you’ve already figured is lost?

But the side effect was more than 300 losses in the space of three years, not to mention an avalanche of criticism that the Astros were somehow breaking one of the unwritten rules of baseball and probably several Ferengi Rules of Acquisition. It’s one thing to be bad, but to not at least feint like you’re trying? Former Astros pitcher Bud Norris has had some pointed criticisms of the rebuild, and some in the media have asked whether management is ignoring the human element. The players aren’t robots, and at some point, the day-after-day-after-day losing has to wear on them. Right?

Still, if the goal is to build for the future, the losing shouldn’t matter. Plate appearances are plate appearances and while not everyone on the current roster will be on the 2017 World Series roster, there will be some who make it, and they can use this time to develop their skills. Call it live dress-rehearsal. The point is to amass as much talent as possible, and when the time is right, everything will be okay.

Or will it? This is a question I’ve asked before, once even concerning the Astros themselves. Does the experience of losing actually harm a player’s development? Teams will often talk about trying to create a “culture of winning,” especially on teams laden with young players. But is that just GM cliché-speak or is it an actual phenomenon? In the past, I’ve found little evidence to suggest that it’s real, but there is always a different way to look at a question.

Warning! Gory Mathematical Details Ahead!

To do this, I’m going to use a database of Baseball Reference WAR numbers from 1984 to 2013 for position players. The nice thing about a longitudinal database like this is that you can see how players change over time. What I did was to look at how long it took until a player stopped either collecting 100 PA in a season or stopped producing value that was above replacement level. He might have stuck around and been there in body, but he wasn’t producing on-field value while he was doing it. The last year in which he produced above-replacement value was considered his year of “death.”

We know that two major factors go into when a player stops being good. One is age and the other is natural talent. Good players might be in decline, but if they start off at 5 WAR, they can decline for a much longer time than players who start off worth half a win. I played around with a few models and found that a simple average of the last three years of WAR for each player ((Y1 + Y2 + Y3) / 3) had the best predictive power. I fitted a Cox Regression, which naturally controls for age (they are most commonly used in studies of mortality) and added the three-year running average of WAR as a covariate. Now, in theory, that should give us a decent idea of what the chances are, given someone’s age and recent performance, that he will wash out. But what happens if we add in the winning percentage from the teams a player was on? Will that have any predictive power?

The theory here seems to be that if a player is always losing and knows that he is going to be on a losing team, it slowly drains the life out of him. In psychology, we call this learned helplessness. When you believe that there is nothing that you can do to affect a bad situation, you stop trying. It’s one of the main theories about why people develop depression. It’s not one major event that gets you. It’s the day-after-day-after-day stress. Maybe in the same way, the constant losing makes players stop trying. It won’t reduce them to impotence, but maybe it makes a detectable difference at the edges. Players who were already teetering lose the will to get better and fall into the void.

I entered several variations on what winning percentage a player experienced into the Cox Regression. I started with winning percentage in the present year, and then combined winning percentage in the past two years and three years. Players who changed teams took the “old team’s” winning percentage with them. One year’s winning percentage had little effect on whether a player would stop being productive, but when I put in the last two years' winning percentage, the significance level decreased. At three years, it got smaller still, but even at its lowest, the significance value (p-value) was .148. The arrow pointed in the direction that might be expected. More losing over time meant a greater likelihood of “death.”

The way that a Cox Regression works is that it models the chances that going forward a person will die (or some other event will happen that removes them from the sample). That risk stays with you as you get older, because for all of us, eventually the risk of death is 100 percent. In this case, if a player has been playing for teams which have had a .400 winning percentage over the last three years, the model suggests that he will have an extra 7 or 8 percent chance of washing out (based on a 1 WAR player) than a guy who played on a bunch of .500 teams. Putting that into roster construction terms, on a roster with 14 position players, the constant losing is introducing the risk that one of them will stop being productive in the very near future, and the players carry that scar for the rest of their professional careers. It’s possible that the player for whom the Grim Reaper comes isn’t one of the ones you had counted on to be there for the future. The problem is that it might be. A team might have the goal of starting a youth movement with live auditions at the MLB level, but it might very well lose some of those young kids while they do that.

One thing to note about this particular regression is that the lack of significance was due to a high standard error (to go with a big regression coefficient). In behavioral research like this, that’s usually a sign that the effect varies widely from person to person. There are some people who can live with the losing. There are some who take it too deeply to heart. Maybe there’s a way to tell who’s who, but right now, I don’t know that it exists.

Still, let me be the first to say that I shouldn’t over-sell these findings. I’d feel so much better if that p-value were below .05, but it’s not. But then I don’t want to under-sell these findings either. That losing might cause short-term harm to team morale is one thing. That it might affect talent in the long term is another. But let’s assume for a moment that these are real findings. Should the Astros really worry about it? In theory, if Houston wanted to buy their way to a shot at a .500 record, it would probably take adding another 10 wins or so to their current team. Shopping for that on the free agent market, even if we assume that they would find some bargains, would take something on the order of $50 million dollars. Probably more. Plus, finishing .500 would mean less money in the pool next year for draft picks and international signings. Yes, they might lose the contributions of one player in a couple years as a result of all the losing, but they can paper over that crack with a couple million. We come to the rather bleak conclusion that even if we grant that the effects of losing are real and are as big as the Cox Regression believes them to be, the Astros should actually ignore them. It’s a cost of doing business.

The Great Experiment

If there’s something to be said about the Astros' Great Experiment, it’s that this is a rebuild in its purest form. There have been critiques that the Astros' front office has been treating players “like robots,” ignoring the human element and assuming the players won’t notice. The commentariat have also lately taken to wondering whether we can put numbers to how much losing affects the actual guys in the clubhouse. These findings suggest that there really are negative effects from living in an environment of constant losing. However, while there are negative effects, speaking from a talent accumulation point of view, fixing them isn’t worth it. Make whatever aesthetic case you want.

The Astros are betting that when the new talent shows up and the team is competitive again, all will be forgiven, both among the fans and the potential free agents they might wish to sign to augment their roster. It’s a big bet, but one that I think history shows they will win ... assuming the talent on the farm comes through. If the goal is to win a World Series, the Astros are following a plan that makes logical sense. I don’t know whether they calculated in advance what the “human cost” might be, or whether they had made contingency plans for when the grumblings of discontent arose. If they did those calculations, they clearly agreed that this would not be a big issue; if they didn’t, then they guessed right.

Russell A. Carleton is an author of Baseball Prospectus. Follow @pizzacutter4
Click here to see Russell's other articles. You can contact Russell by clicking here