I read the Hard Ball Times for interesting fantasy analysis throughout the baseball season, but a recent article discussing possible inefficiencies in the fantasy marketplace didn’t sound right to me.

One of the central arguments here is that fantasy owners place too heavy a value in current production, causing them to make decisions on limited sample sizes of data.  First, this argument only really applies if the league in question has owners who are chasing the new hot thing.   The league I’m in is hyper aware of past performances, and while they usually aren’t using HR/FB, UZR/150 (I’m not quite certain what fantasy value this statistic provides) and other non-mainstream statistics/metrics, they have a general understanding of the concept of regression to the mean.

I’m not in a good position to argue with the statistical merits of evaluating players based on a small sample size, but I have noticed some trends in my league.  I’ve offered Ryan Zimmerman in a number of trades this year because I have two solid thirdbasemen (Longoria is my other).  Generally, my league scoffs at the idea of Zimmerman being included in deals for pre-season first and second round draft picks.  I offered one multi-player trade including Zimmerman that would have brought David Wright to my team (I know, another 3b) because I wanted to increase my batting average and steals.  It’s obvious that David Wright has had a much more impressive fantasy career to date than Zimmerman based on the largest sample size of data available, their career statistics.  Sticking with Wright is playing it safe.  But this could be misleading.  The smaller sample size of this season reveal some realities that aren’t represented by the career numbers:

1. David Wright has protection issues in an injured Mets lineup.
2. Ryan Zimmerman has gained protection in a Nats lineup (Adam Dunn now hits behind him)
3. Current trends have Zimmerman reaching career numbers this year in HR and AVG.  David Wright is trending down on HR and RBI’s significantly.

My criticism would be that #3 would be exempted by looking at the larger sample size considering the regression towards the mean.   When taking other outside factors into consideration, there may be reason to believe that both players would continue their current trends this season despite the historical data.

This could just be a specific exception to a general rule, but if there is plausible data to point to a reason why a player is breaking out or slumping, assuming the players will gradually revert to their former selves for the remainder of the season is not necessarily a sound idea.

Note: Keeper leagues are a different deal all together.  I think the argument from the Hardball Times makes more sense if your in a keeper league because the outside factors listed above are so volitle.