My old post about A-Rod  colluding with the opposition to massage his stats was drawn from a New York Times article.  It suggested that in blow-out games when a few extra runs here or there would not affect the outcome, A Rod  would collude with friends on the other side to raise improve their stats and his.

Now a follow-on article in the NYT says a data analysis suggests this did not occur.  I personally think A Rod is getting a bad rap and did not cheat.  But the statistical analysis does not make a convincing proof of this hunch.

The data say that in low stakes situations A Rod had many fewer runs than in high stakes situations.  This leads the NYT to conclude that either there was “no tipping going on or it was pathetically ineffective.”

This when a little game theory is useful.  When the game is close, you have the most incentive to exert effort to win.  When it is not close, even the likely winner slacks off as there is no reason to work hard.  This is the game theory analysis of R and D races for example (though the details are quite intricate and can sometimes be surprising).

This logic meant A Rod would naturally, given the dynamics of the game, have played hard when the game is close.  The collusion has a countervailing effect when the game is not close.  But this effect may not be large enogh to counteract the natural incentives of competition.  So the data do not prove anything.  But I still think A Rod did not cheat.

(Another Hat Tip: Pablo Montagnes)