This week, let's consider another trading strategy - one which I think is superior to pitcher rotation, at least during the first half of the season. The strategy is to use your trades to generate price gains. Since a player's price change is directly related to the number of times he is bought or sold, successful execution of this strategy relies on your ability to forecast what other managers will be doing. Sometimes it is fairly easy to anticipate a buying surge. For example, immediately following Kerry Wood's 20 strikeout game, he was a cinch to appreciate significantly. In general, hot streaks are a good predictor of price increases, although early season hot streaks are easier to detect, since they show up in year-to-date rankings as well. Another good tactic is to identify players who have recently suffered big price declines which don't reflect their future potential. An example is Kenny Lofton, who dropped earlier this season on fears that he had seriously injured his knee. As it turns out, the injury was not severe - but since it occurred shortly before a repricing period, many managers bailed early, pushing his price down. This made him a clear bargain the following week.

I don't want to dwell on ways to identify likely price gainers; that's your job as a manager. Instead, I want to evaluate how the benefits of this strategy compare to the expected point gains from rotating starting pitchers.

Since the ultimate objective of any manager is to produce as many SWPs as possible, we need a way to relate price gains to SWP production. To do this, we need to first measure the relationship between price and performance. I like to express player production in terms of average points per eligible game (SWP/G). An eligible game is defined as any game for which the player is on his team's active roster. Using eligible games, rather than actual games played, puts players who don't play every day - like pitchers and platoon players - on a consistent basis with players who do play almost every game. It doesn't matter whether a player produces no points by having a weak game or by sitting on the bench - "no points is no points". And while a historical SWP/G average doesn't necessarily reflect the consensus expectation for future performance, it does provide an objective measure against which to evaluate relative price. Further, through the first two months of the season, player prices are 80% correlated with SWP/G, so the relationship is statistically strong.

I think a graph most clearly shows this relationship. Shown here is a scatter plot of price (shown in $millions on the x-axis) vs SWP/G (on the y-axis). (Click on the graph for a larger image.) For this graph, I included only players who were active on June 1st and who had been active for at least 60% of their team's games. Slightly over 600 players satisfied this definition, and these players clearly comprise the bulk of the relevant universe of players for active managers. The players represented by the points near the top edge of the cluster are generally the best values. It is probable that some of these players are really "early overachievers" who the market does not believe can sustain their productivity. And those points near the bottom edge are generally the poorer values, although again, some of these are players who started slowly, but who will probably produce at a higher rate in the future - like Randy Johnson, for instance. In general, though, there is a definite slope to this cluster, and this slope represents the average expected pick-up in SWP/G as price increases. (In fact, you might say that the ultimate job of any manager is to identify the best relative bargains within this point cluster of opportunities.)

For this universe, that "slope" is currently 1.8 SWP per game per $million. Restated another way, if you gain $1 million in roster value, you should (on average) be able to upgrade your point per game production by 1.8 SWP. With roughly 100 games still remaining this season, a gain of 1.8 SWP/G should result in an increase of 180 SWP over the rest of the season. Obviously, every trade and gain situation will be somewhat different, but over the course of many trades, it isn't unreasonable to expect to achieve average increased production in this general ballpark. Using this standard, an 80 SWP pickup for a pitcher rotation trade has roughly the same expected value as a trade which produces a roster value gain of only $450,000 in early June. A $450,000 per trade price gain isn't an unreasonable target to beat. Of course, as the season progresses, the total expected point gain for a constant SWP/G increase is reduced. (An increase of 1.8 SWP/G for 50 remaining games only works out to a total of 90 SWP.) So, as the season wears down, pitcher rotation may become a more effective trading strategy. But, there is no reason to believe that your gain from pitcher rotation will be any less in August or September than it will in June, so I'd think the best strategy for now is to look for gains opportunities. Later in the season, you'll have a better idea of how many pitcher rotations you can afford.

RotoGuru is produced by Dave Hall (a.k.a. the Guru), an avid fantasy sports player. He is not employed by any of the fantasy sports games discussed within this site, and all opinions expressed are solely his own. Questions or comments are welcome, and should be emailed to Guru<davehall@rotoguru2.com>.