At this point... I have back tested 1800 games... a small sample size... but one game
tonight ... the system is telling me to bet. I concentrate on individual bets... and develop algorithms for the over bet... the under bet... Away -... Home -... Away +... Home +. I spend on average 3 -4 hours a day collecting data... processing it then analyzing Vegas for weaknesses.
Anyway... the system is currently only picking 1% of the 1800 games on Away - the points. My criteria is pretty stringent but one team barely makes the cut... but makes the cut nonetheless.
Fairfield -4
Who knows it might bomb. But all my data is pure ... my analysis is pure... but it is only good if it works moving forward.
Well, it is nice to see my Away Minus the points algorithm work. Back tested it sits at 91%. Again, it is only betting around .8% of the games. However, Fairfield fit the profile, so obviously I was curious whether the algorithm would hold up.
Fairfield -4...
They won by 24 so I am pretty sure it did. I know one predicted score doesn't prove a back tested algorithm but that is all I have to go on for now. If another Away Minus pops up... I will reply to this thread with the new game so those interested can track.
Feeling pretty good... on my algorithms but will continue to fine tune.
These are back tested results... I started a second 10 day trial using these algorithms 4 days ago.
Overs 15-2 88%
Unders 32-12 72.7%
Away - 11-1 91% not including FAIRFIELD today
Home - 46-24 66%
Away+ 21-9 70%
Home+ 18-8 69%
Total 143-56 71.86%
These algorithm are back tested over the season starting approximately around Thanksgiving. By back tested... I compare my predicted outcome versus Vegas predicted outcome. I have to wait for each team to play a certain amount of games before looking at the statistics. I settled on 5 games. I look for patterns and similarities of attributes that my model predicts better or worse. I then apply algorithms adding or eliminating games to produce a desired outcome. Now, it is time to test to see if those algorithms work moving forward. If they don't, which has happened multiple times... I go back to the drawing board. However, I believe these algorithms are solid enough that they deserve a test. Again, these are algorithms I built looking for weaknesses in Vegas odds. It bets approximately 10% of the games. Currently 6-3, as of Fairfield game, this ten day test. Tomorrow will be interesting... and I will continue to upgrade.
Top Three Algorithms
Away -
Overs
Away+
Bottom Three
Unders
Home +
Home - (By Far my Weakest)
Lots of Games tomorrow. I will publish my bets. Enjoy. I told all my friends to bet FAIRFIELD... no idea if they listened to me.
Just saw the lines are published... It takes me approximately 3 minutes to download all the data and then run my program. Unfortunately, none of my best algorithms have any published bets today. I am not sure how to take that. Is that a good thing. The other algorithms aren't bad but I need to work on them especially Home -... even though the results are bad. Here they are... I do like the under bets. FOR MODELING purposes.... whatever the end of regulation scores are... that is how they are recorded in my model. OT is disregarded. I don't care what happens in OT.