Ready to build some new sports forcasting systems

[RJJM94]

I want to develop some statistical approaches to forecasting sports.

I want to start with NBA and finish the season. Then follow with MLB and football later on.

I am good with excel. I have a major in finance/economics so I understand how to price securities and would like to convert that into a betting price.

I'm willing to put a lot of time/adjustments into this to try to make it forecast the scores of various sporting events.

Where is a good place to start?

 

[RJJM94]

I guess I could start looking at previous stats for teams and see how they correlate with the final score. Do something like 1 week stats, 1 month stats, YTD stats. And see how each correlates and weight them accordingly?

 

[Santo]

Lots of work has been done on this. My advise would be to adopt KISS. Try and isolate variables that are either under or over-represented in market prices.

 

[RJJM94]

what is Kiss?

 

[Santo]

Keep it simple

 

[RJJM94]

What is a good website to pull stats off of to excel?

 

[RJJM94]

just found www.82games.com it has some great statistics for NBA, just need to figure out how to make all the stats = a final score.

 

[fhmesq44]

Contact Brewers7, he has spreadsheets for NBA going back years with a lot of data in them.

I know guys that put together models often use pace of the game to calculate their line for o/u. I haven't done it and don't know how they calculate it, but they have had success. Lines also get A LOT sharper as the season progresses and become much harder to beat.

 

[heatherRays]

nba is hard to handicap

 

[tatehill2000]

bro get in line, You are in a long long of NOT SO ILLUSTRIOUS names. You will fail, What you are suggesting has be attempted over a mil times,

good luck,

and whatever you do, dont attempt useing a progression.

tater

 

[Iceman]

I am not sure he will fail. Sure if you expect the computer to spit out a number and you can just go bet it right then and there, complete faiure no doubt but here is the what I think is the good news:

If you can get this thing to 51-52% and then you can spend time on 10+ outs, working off and on thorughout the entire day, hitting openers, playing back at steam at close, etc... Then maybe you raise that 51-52% your program spit out and get it to 53-54% and you can make quite a bit of money off of 54% but the whole thing is work, don't ever doubt that. Trying to beat widely avaibile numbers more then likely isn't going to happen over time.

Santo is always very correct on his statement. Don't get to technical or try to out think yourself. If you can find numbers the market doesn't seem to value as much as they should then you are on to something.

 

[RJJM94]

I'm willing to put decent time into developing something simple. Like something that looked at 5 differnt stat matchups between two teams.

What do you guys think are the most significant stats to look at in NBA? Which ones are best used for forecasting the future.

Also I want to try and throw a couple stats in there that are not looked at very often and use those to make my system unique. Any input would be greatly appreciated. If anything, you guys will be able to reap the benefits of this when I post the results.

 

[trytrytry]

ive heard there is a correlation to how late the visotor flight landed and how they do that day on the court... not sure ho you get that data however!