Gentlemen. I have been fine tuning my MLB system for the regular season and am finally ready to post my picks. Basically, every morning I pull a ton of stats for each day's matchup, broken down into Total Season, Home/Away, Last 30 games, Last 20 games, Last 10 games; over 50 individual statistics in all. Every once in a while I'll run the stats to determine which ones are actually relevant to those wins.
The current set of filters consists of 8 variables that each team must have an advantage in, in order to be a play. Over the last 22 days, it is 31-12, 72.09%, with no plays over -175. This extremely high percentage is not necessarily an indicator of future success. What's important is the ability to constantly find which stats are more relevant as the season goes.
GL everyone.
Thursday, June 6th, 2013: PHI +127, 1 units to win 1.27 ATL -107, 1.07 units to win 1.00
Yeah it does sound a little too good but even though it is a small sample size, it has some pretty serious implications.. Unfortunately I really like the opposite sides of these games. Well both probably go 1-1
Good points made by you guys. Past performance is not necessarily an indicator of future success. The key is to keep working with the data. There's no magic set of statistics that will guarantee wins year in and year out.
Record:
1-1, +0.20 units
Friday, June 7th: PHI -140, 1.40 units to win 1.00 ARI -132, 1.52 units to win 1.10
[QUOTE=pm4e;18868596]Good points made by you guys. Past performance is not necessarily an indicator of future success. The key is to keep working with the data. There's no magic set of statistics that will guarantee wins year in and year out.
I'd be willing to bet, it's just a ton of work to nail down what those parameters would be.
Right now, my most complete set of data extends from 4/21 to yesterday, with only 3 misc days that are unavailable. I filter out all games that have a starting pitcher with less than 10 innings pitched this season. I'm also filtering out any game that has a team with odds of higher than -175. This leaves about 400 games in the data pool. If I take my 12 most relevant statistics, I get a record of 29 wins, 19 loss, 60.41%. That breaks down to a record of 10 wins, 13 losses from 4/21 to 5/12, and a record of 19 wins 6 losses from 5/13 to yesterday.
I can actually go one step further and get a total record of 18-5, 78.26% over the full 4/21-Yesterday set if I use my 15 most relevant statistics.