Sports Model Edge

**djiddish98** · 07-20-10, 05:28 PM

I believe this comes down to personal preference. I'm willing to bet edges of 1% or higher (depending on how short the odds are), because I like to gamble. Others might like to wait for much larger edges (again dependent on the odds), but I would imagine they are betting less frequently.

I would also add that it might be dependent on what edges your model is showing, and the distribution of the actual results with regards to the closing line.

**Justin7** · 07-20-10, 06:29 PM

A major problem with modeling is how to correctly assess your edge. How sure are you that you have a 1% edge?

**Maverick22** · 07-20-10, 06:35 PM

Justin, What are ways to determine your edge? I had a model a while ago before an epic computer crash

And i wasnt exactly sure how to determine edge and est win %. How do you do it? Or is determining that a typical "secret" people dont give out?

**Sportslover** · 07-20-10, 07:36 PM

I generally bet a line if I estimate my edge to be 2% or better.

**TomG** · 07-20-10, 07:57 PM

Originally posted by Maverick22

Justin, What are ways to determine your edge? I had a model a while ago before an epic computer crash

And i wasnt exactly sure how to determine edge and est win %. How do you do it? Or is determining that a typical "secret" people dont give out?

If you can't use your model to determine the est win % and thus your edge, then you don't have a working model.

**Justin7** · 07-20-10, 08:08 PM

Originally posted by Maverick22

Justin, What are ways to determine your edge? I had a model a while ago before an epic computer crash

And i wasnt exactly sure how to determine edge and est win %. How do you do it? Or is determining that a typical "secret" people dont give out?

There are lots of ways to test it.
1. How predictive is your model of line movement? If the opener moves in the direction of your model 60-70%, you have a good model.
2. What is the win-loss record in the past based on backtesting?
3. How do you do on action point grading of your plays?

**Optional** · 07-21-10, 12:55 PM

Originally posted by Justin7

3. How do you do on action point grading of your plays?

Sorry for an off topic question, but what does this mean? (I don't bet baseball)

**roasthawg** · 07-21-10, 01:58 PM

Originally posted by neila

I'm building my model and I wondered at what percentage will you take a bet. Since this is as much art as science, do you wait until your model shows your desired edge or do you take the bet when it shows any positive return? If 52.38% at -110 is a benchmark, do you take a bet at or over that or do you insist on + 5% over that or what? Essentially, where do you jump in? Thank you for your consideration.

If you have an edge you bet it... use kelly to determine optimal bet size.

**Justin7** · 07-21-10, 02:00 PM

Originally posted by Optional

Sorry for an off topic question, but what does this mean? (I don't bet baseball)

It's not as useful in MLB... But in NFL, add up the total points in each game that you covered by, and subtract however many more points you needed to cover for each game.

**Indecent** · 07-31-10, 04:23 PM

Originally posted by Justin7

But in NFL, add up the total points in each game that you covered by, and subtract however many more points you needed to cover for each game.

I know people don't like to talk about their models too much, but what kind of results should you be looking for on action points. Do you cap the max action points per game (10 or -10 for win/loss) as most books would do when evaluating the network, or leave it as is?