Do you wonder whether you're getting maximum production from your roster, given the value you have to spend?

Early in the season, most successful teams concentrate on building roster value, with current point production only a secondary consideration. But as the season enters the second half, available trades become more scarce, and you should be aiming to produce as many points as you can.

Using my preferred productivity measure of SWP per eligible game (SWP/G), I've developed a table which shows the maximum SWP/G average you can currently get as a function of franchise value. As you may recall, an eligible game for a player is defined as any game for which that player is active on a major league roster, and therefore eligible to play. Games during a stint on the disabled list don't count - because as a franchise manager, you can trade away a player on the DL to avoid the lack of production. But games missed when a player is active are counted as eligible games. This seems to be the best across-the-board measurement standard to evaluate all players - starters, relievers, and hitters - on an even playing field.

Here's how I approached the project. First, I excluded any player who was currently inactive (in the minors, released, on the disabled list, etc.). I also excluded from consideration any player with less than 25 eligible games. Some of these players look attractive, but their SWP/G averages are less likely to have stabilized. For everyone else, I calculated a player's SWP/G average as the quotient of his total SWP divided by his eligible games, using stats as of the All-Star break. I also used the prices that were in effect during the All-Star break.

Next, I used a standard spreadsheet optimization routine to develop the maximum total SWP/G for a roster, given a maximum value to spend. All of the typical position constraints were also satisfied. While it is possible that the optimization failed to find the best solution on occasion, the results seem reasonable. Here is the resulting table which shows the total SWP/G as a function of available franchise value, in $5 million increments:

The best team money can buy (on this basis, at least) costs $133,740,000. As the chart indicates, that roster produces an average of 289 SWP/G. For a detailed list of each supporting roster, click here.

How should you use this information? I'd suggest that you look at the numbers in the vicinity of your team's franchise value, and determine whether you think your roster is likely to produce something close to the listed SWP/G level. It's unlikely that you'll have the precise roster that produced the table values, nor is that necessary to be competitive. A quick analysis of the current top 50 worldwide teams indicates that these teams are about 15 SWP/G under this "efficient frontier" on average, so anything within 15 SWP/G should be quite competitive.

There are a number of caveats I should mention:

• SWP/G averages are measured over the entire season, and a player's current pace of production will likely be higher or lower than the season average. You might rather have someone with a lower historical average if you think the player is going to produce at a higher level going forward.
• In fact, the optimized roster is probably going to emphasize players who have recently been on a hot streak. So, there might be a tendency for these SWP/G numbers to be unrealistically high as a long term expectation.
• The optimized rosters do not consider trading limitations. As your roster value increases, you can't freely trade to continually maintain the optimal roster. So it's unrealistic to expect that many teams will have the optimal roster at any point in time, and it's also unrealistic to assume that the optimal roster today will continue to be optimal next week.

Given these caveats, I am not suggesting that you should use this precise method to assemble your roster. There is significant lattitude for judgment to play a role. In fact, one of the best (and most difficult) strategies is to anticipate players whose future production will be better than their past production. These players will not only add point production, but will usually produce capital appreciation as well. And of course, as we all know, averages change as the season progresses. So the best strategy is to pick the players who will perform the best in the future; these aren't necessarily the same players who performed the best in the past.

Still, this type of analysis can provide a useful guide to help you assess how efficiently your roster value is deployed. Small differences in SWP/G can mount up over the course of a season. Most players have about 75-80 games left, so a sustained improvement of 5 SWP/G will generate an extra 375-400 SWP by the end of the season.

I'll dig a little deeper into this analysis next week, to see what other information might help you assemble the best bunch for your buck.