Is it possible to create an accurate odds algorithm?

[abev]

I have written vb code to create odds for the 4 major US sports for moneylines, spreads, and over unders. I am using between 30 and 40 variables to create the lines. My goal was to duplicate what the 'real' lines were and I would estimate that about 50% are legit lines and the other 50% can be way off. Is it too much to expect to be able to create lines programatically without a human 'touching' the line?

[tachi]

...

[xyz]

Just to clarify, are you saying 50% of the lines outputted by your program are close to the lines put out by the books, and the other 50% are not close? Are you using the opening lines or closing lines from the books for comparison?

I have written vb code to create odds for the 4 major US sports for moneylines, spreads, and over unders. I am using between 30 and 40 variables to create the lines. My goal was to duplicate what the 'real' lines were and I would estimate that about 50% are legit lines and the other 50% can be way off. Is it too much to expect to be able to create lines programatically without a human 'touching' the line?

[Data]

I have written vb code to create odds for the 4 major US sports for moneylines, spreads, and over unders. I am using between 30 and 40 variables to create the lines. My goal was to duplicate what the 'real' lines were and I would estimate that about 50% are legit lines and the other 50% can be way off. Is it too much to expect to be able to create lines programatically without a human 'touching' the line?

It is possible. The problem is to find predictive variables vs. descriptive ones. Most likely, your variables are descriptive and will not help with creating future lines. On the other hand, if your variables are predictive then you are on a short track to prosperity.

[abev]

Just to clarify, are you saying 50% of the lines outputted by your program are close to the lines put out by the books, and the other 50% are not close? Are you using the opening lines or closing lines from the books for comparison?

Yes - 50% are solid opening lines and the other 50% are horrible opening lines.

It is possible. The problem is to find predictive variables vs. descriptive ones. Most likely, your variables are descriptive and will not help with creating future lines. On the other hand, if your variables are predictive then you are on a short track to prosperity.

Hmmm data...I would guess most of my variables are descriptive. But wouldn't past performance be indicative of future performance?

[Data]

I would guess most of my variables are descriptive. But wouldn't past performance be indicative of future performance?

More likely, not. Read these threads: http://forum.sbrforum.com/search.php?searchid=1116293

[abev]

More likely, not. Read these threads: http://forum.sbrforum.com/search.php?searchid=1116293

Thanks for the info, Data. I enjoyed some of your responses.

So is there an accepted method of how bookmakers come up with the line? My main concern is figuring out where the lines should open at, not who is going to win or by how much.

The last option is are there any sites that consistently come out with lines at least 24 hours prior to the event start?

[Data]

More likely, not. Read these threads: http://forum.sbrforum.com/search.php?searchid=1116293

It seems that the search links may expire. This was the following search query:
Search: Keyword(s): datamining ; Forum: Handicapper Think Tank

[Data]

So is there an accepted method of how bookmakers come up with the line?

This should give you an idea: http://www.google.com/search?hl=en&q=LVSC+linemaking