This has turned out to me one of my more difficult articles. First, I prepared it while on a family vacation, so I didn't want to spend too much time researching and writing. Second, the early results didn't at all show what I expected, and I have been trying to get behind some of the numbers to find the underlying causes. I have some facts, and I have some hypotheses, but I'm not confident I know what is at work here. But time's a wastin', so let me think out loud here, and maybe something will occur to me!

I started by looking at how each of the 18 "efficient" rosters (ranging in value from $50 million to $135 million) would have actually performed over the first 15 days after the All-Star break. I chose 15 days because that should give all pitchers 3 starts, and should provide enough time for the hitters to trend toward their SWP/G averages.

Guess what? None of the teams performed up to expectations! In fact, none were very close. The worst effort was the $100 million team, which totaled only 71% of its expected SWP production, but the best performer only achieved 80% of its expectation (the $85 million team). Every roster had at least one player who outproduced his average, but no team had more than three overachievers. Hitters did a little better than pitchers (Nen and El Duque were on all rosters, and both had off periods - although El Duque's recent debacle was not included). Still, the best hitting team only came in at 86% of expected hitting point production.

To put these shortfalls into better perspective, over a 15-day period this amounts to a 900-1200 SWP deficiency vs. expectations. Pretty noticeable. What could be the cause? I considered these possibilities:

1. Maybe the market really is efficient, and these players looked cheap vs. their averages because the market realized that they were going to be underproducing. Frankly, with apologies to Smallworld, I doubt that the market is that efficient!
2. It could just be bad luck.
3. Maybe these were just bad weeks for the entire universe of players. To test this, I considered all players who were active throughout the entire 15-day period. In aggregate, hitters produced at 96% of expectation, while pitchers as a group were almost "right on" at 99%. These certainly don't sound like they explain the results I was seeing.

I did notice something interesting in the details of the player universe. Only about 40% of the hitters produced better than their season averages, and the median hitter only produced at a 90% clip. This probably results from the asymmetry of the hitting points. Hitters have limited opportunities to score negative points, so on any given day, it is more likely that a hitter will produce slightly less than his average, offset by the occasional big game. It looks like this daily relationship even holds somewhat for a 15-day period.

Pitchers showed a more balanced profile. They also showed much greater variability than the hitters. This is due to the wider range of possible game totals (positive and negative), and the reduced number of games actually played over a 15-day period.

So much for science. In the absence of fully understanding the reasons for this dismal performance (dumb luck - or maybe "smart luck" - is still my leading contender), let me leave this as an open issue in order to move on to a different question. How well could you have done over this 15-day period if you had 20/20 foresight? I reassembled the rosters using All-Star break prices, but this time maximizing actual point production over the next 15 days. Obviously, unless you are clairvoyant this is not a realistic way to assemble your roster, but it is still fun to consider.

The quick answer is that you could have really made a run up in the standings! These "best" rosters produced 1,500-2,000 SWP better than the expected production of the correspondingly valued "efficient" teams, and roughly double the actual point production of those teams. Interestingly, you only needed $92.5 million to achieve the maximum output of 5,463 SWP. But you could even have amassed a cool 5,000 SWP with only $55 million to spend. All you would have had to do was discover such blossoming players as Mark Portugal, Jon Lieber, Mark Clark, Lenny Webster, Todd Helton, Tony Womack, Royce Clayton.... you get the drift. I'm sure no one had this roster. Ah, but there' always next month, eh!