Last time out:

Cincinnati -110. Won a one-run game. I was very impressed by Brad Penny, who got a lot of ground balls. It was one start, so you can't get too crazy.

Kansas City -118. They had Dontrelle Willis on the ropes early and let him off them. Man, this is a bad baseball team.

Didn't like anything Friday, and took the weekend off for unrelated reasons.

I was thinking this weekend about edges and the proper way of approaching this project. For any large set of bets, your profit should simply be edge times volume. If you have a .1|PERCENT| edge every time you make a play, you make more money betting a thousand times than a hundred assuming you have the bankroll to handle variance. This is, more or less, the entire principle behind owning a casino. Offer games of chance that are positive expectation for the house and keep the doors open 24/7/365. You get into the long run pretty quickly and you make money, even when your edge is at its smallest, such as the pass line on a craps table (0.4|PERCENT|).

So if I think a performance analyst has an edge on the bookmakers, I should be playing more games, applying a small edge many times. This isn't a craps table, though, where the odds are constant with each roll of the dice; this is more like playing blackjack, and specifically, playing blackjack while counting cards. The edge isn't constant. You're meandering through breakeven and negative-expectation situations while waiting for the positive expectation one, and when it comes along, you have to be hitting it hard. A card-counter will bet $25, $25, $25 and then, when he calculates sufficient edge, bet $100 or $200, taking advantage of the situation by getting as much money as possible onto the table. Variance is still an issue, of course--having an edge doesn't mean you'll win a given hand or set of hands--which makes having a substantial bankroll even more important. But the key is that you're varying your investment based on the size of your edge.

Now, it's much easier to calculate the number of aces and tens left in a deck or shoe than it is to know if the Tigers are really a 2-1 or 2.2-to-1 favorite. The principle, though, I believe to be sound. So to that end, I'm going to make two adjustments beginning this week. I'm going to offer more picks--the equivalent of betting one unit playing basic strategy--and vary my unit sizes upwards with some frequency based on my perception of the edge. I think this is the proper way to maximize theoretical ROI.

Texas/Cleveland over 8, one unit. Perhaps in 2007, a RIch Harden/Fausto Carmona matchup would have warranted this number. Not this time. The Indians' strength is showing power and patience against right-handed starters.

Houston +195, one unit. My love for the underrated Wandy Rodriguez has been established, and as bad as the Astros are, they're simply not a 2-1 dog in any game he starts.

Florida -137, two units. Ricky Nolasco at home against a team, the Reds, that the Marlins are on par with. They have the much better starting pitcher.