Point Diiferentials!!!!!!!!

**tto827** · 02-07-13, 04:00 AM

Interesting concept, but I doubt it hits higher than 50% over the long run. Any large discrepancies from the spread that you get would be due to different strength of schedules, and who they played at home vs. on the road.

**yak merchant** · 02-07-13, 04:09 AM

Originally posted by Leatherass

Handicappers,
Have a new theory that might deserve a slight look into (have only tried this with NCAA basketball). Throughout the season, teams of the same conference play eachother. Let's look at the Western Kentucky @ North Texas game tomorrow (02/07/13).

What I had been doing is comparing the teams both opponents had played. Below are two lists compiled for the game, a + denotes a win by the following number and a - denotes a loss by the following number. I then average all +'s and -'s to get an average of how each team is doing via conference. Then after all is concluded, incorporate the spread and try to obtain as much value as possible.... :

North Texas when playing:
TROY +17
MTU -18
USA +10
UALR -5
ULL -4
FAU +2
ARST -9
FIU -6
Average: NT is losing by -1.625 points

Western Kentucky
TROY +4
MTU -19
USA -8
UALR -5
ULL +23
FAU -3
ARST -14
FIU +13
Average: WKU is losing by -4.375

From the above data I conclude that the "theorhetical score" should be North Texas winning by 2.75 points. Incorporate the spread of -2.5 and you get NT winning by .25 points. This to me would be a game to pass on. Although I don't have the time tonight, there are some games in which you can get 4, 5, 6, or more points in value. UNC for example on Tuesday night. Now of course you must take into account where the game is played and injuries among other things. This is the method I use and it had worked for last season. I can't seem to program Excel to spit out an answer for each game but pen and paper will do the trick, although it is lengthy. Any opinions or thoughts would be awesome.

Best of luck to all,
Devin

PS sorry for the misspelled words. I have class in the morning and need to get to bed ASAP.

Well I'm not going to tell you that you "Can't" make money with such a system, (hutennis will be along any minute to do so and explain to you how absolutely nothing can be gleened from past data, everything is completely random, and that model building is absolutely incapable of producing profits), but using something that simple and only looking at final scores you are fighting an uphill battle. I started with a very similar "model" trying to predict college football in I believe 1994. And while there are always glimpses of hope in the numbers, and hot streaks, that is a failure waiting to happen. Now don't get me wrong, comparing strength of opponents can be a good idea, but just think of how the following scenarios would affect a basketball model

1. Team A get's up by 20 at half and then coasts to a 8 point win in many games.
2. Team B played in a bunch of really close games, but most of them ended up in a foul fest in the last minutes making the eventual Margin of victory 7 points.
3. Team B had a star player hurt in all but one of the games.
4. For common opponents Team A played 6 of 8 games at home, Team B played 5 of 8 on road.

Obviously I'm not going into the gory details of how to solve those problems (nor do I have a satisfactory answer to many of the underlying issues with the game stats), but I will give you a couple of pieces of advice that would have saved me a decade of screwing around.

Averages are good for some statistics, but especially in basketball look into using medians.
In no sport is Home/Away performances more important than in college basketball. I would find a way to incorporate that into your model.

And if you can't get excel to calculate something that simple, you probably have a long way to go if you are going to be a "Stats" handicapper. Don't get me wrong there are guys who make money watching tapes all week and couldn't tell you anything that has to do with a team's stats, but if you are going to do the fundamental handicapper thing, you need to start reading around here about scraping and databases.

Good luck.