Announcement

**Horfin** · 02-01-2013, 04:24 PM

Now that I think I have ironed out most of the wrinkles I will explain all of the systems.

System 1: This is a computer model based upon the likelihood of a dog/favorite to win. I utilize the opening line, the line in the afternoon (2 to 4 pm), the percent of bets placed on the dog and the movement of the line. From these number I extrapolate 6 different numbers - called factors (i.e., the closing line divided by the % on the dog, the change in the line divided by the % on the dog, etc). I have tracked this data since 2009 and have a database that tells me which numbers within a particular factor are more than 80% likely to pick a winner. If more than 2 factors point to a team, then it is a play. This system works very well, however, if I do not continuously update the results it begins to fail. Also, from time to time it will hit a losing streak. The system started out hot this year and then went even.

System 2: System 1 was not producing enough plays, so I expanded out on System 1. System 2 tries to account for the record of the dog and the record of the favorite and how that relates to the line movement and the percentage of bets on the favorite and the dog. This system has done very poorly; however, I believe that this was my error. The system spits out a result which is either a play on the road team, home team or no play. On one part of the system, it was spitting out road team when it was supposed to spit out home team, due to a formula that I wrote incorrectly. Hopefully this explains the 19-28 record meaning it would be roughly 28-19. Now that I have fixed it, we will see. The probable downfall of this system will be the fact that it picks too many games and I can't find a way to narrow down the plays.

System 3a and b: This is the most comprehensive system and quite frankly the system that I think will replace all of the above, once all the bugs are tweaked out of it. The idea behind this system is that I use all of the numbers above (line, open, close, percentages, records, home and away, change in line, difference between the winning percentages of both teams) and based upon prior results, the system creates what the line should be, within a 80% historical accuracy. Stated otherwise, if I put in any game from the past into this system with all of the variables it would pick the past games at an 80% accuracy rate (against the spread). However, there are a TON of variables. At the end of the program it spits out a line, compares that line to the real line and makes a play of either the Dog, Favorite or No Play. My recent adjustment was to weed out 30% of plays, leaving the best plays. The difference of 3a and 3b is home dog or road dog (3b).

Right now, I feel like System 1 is the safest followed by 3 and then by 2. However, my mistake on 2 may redeem that system. I am quietly tracking a System 4 which is what happens when these systems agree on all 3 or 2 of the three. I had to throw that data out due to my mistake on System 2.

Sorry for the long explanation but people have asked and now I can point to this post in the future for the great (hopefully) results.

In case you can not tell, I am fascinated by numbers and the relations they have with one another. (No I'm not a Dork.)

Announcement

Horfin 02 01 2013 NBA

Horfin 02 01 2013 NBA

Comment