THE BOOK--Playing The Percentages In Baseball

Tuesday, December 28, 2010

Do low-talent teams overspend on free agents?

Matt had a pretty good Q&A from the saberfreak article.  He also said this from Question 36:

Matt: Teams don’t appear to PAY less for wins when they have smaller fan bases or are further behind on the projected win curve, but that’s because they are outbidding teams who are in the sweet spot of the win-revenue curve and/or from big markets. Small market teams and below average teams certainly pay for a lower quantity of these players, but they pay what it costs when they do buy those guys’ services. Sometimes they are doing it because of a splash effect, sometimes they are doing it because they think the rest of the league is undervaluing the player, sometimes they do it because they just overestimated the player’s talent, sometimes the player has a unique value to the town or is willing to take a discount. But the market is set by above average teams from big cities, and the prices are driven by that part of the demand curve even when they are signed by other teams.

So, what he’s saying is that rather than teams pay for their specific marginal $ per marginal win, all teams are forced to pay up to the “trend setters”, and that is, the teams in the sweetspot of the playoffs, those teams with say 85-90 wins of talent.  Whatever they pay, they set the market.  And, if you want to play with the big boys, you’ve got to pay like the big boys.

This would mean that a team with a low talent level, say 75 wins and under, if they pay like the big boys (say Gil Meche with the Royals), they are, overall, going to get burnt, as they will have overpaid by say 25% to 50%.  In a rational market, Meche would never be bid on by the Royals, because he can never have the same impact (monetarily) with the Royals as he could with the Redsox.

Theoretically, this makes sense.  If we had the knowledge required to make those deals.

But, what are we really talking about here in terms of impact? It’s not like God tells us what a team’s true talent level is.  We know that one SD = .060 wins in talent level among the teams, but our uncertainty level is going to be high.  Taking a guess, I’ll say that it’s one SD = .040 wins of uncertainty (this includes the chance that the player won’t play as often as he should because of injuries). 

Furthermore, with only 162 games to play, it will happen on a more than rare occasion that a team that we think is a true talent of 75 wins (meaning that we think are 75 wins +/- 6.5 wins) is going to win 85-90 games.  Or on the flip side,a team that is a true 87 win team (i.e., true 87 +/- 6.5) is going to win 72-77 games.

So, I’m ok with Matt’s point as a point of general principle.  But I think you are going to be severely constrained to apply that as a more general rule within the context of what we actually have to work with.