Predicting % prob of a team covering the line

[antonyp22]

I've had good success with models that generate a "fair" line for a match-up and betting with varied amounts based on the differential between my "fair" line and the line offered by various books.

I now want to be able to quantify this in terms of a % probability of a team covering the line i.e. 57%, 60% e.t.c. and use this to bet with a Kelly system.

What are the various approaches to doing this? Fitting the predicted line to a distribution? Monte-Carlo simulation?

Any tips or pointers would be much appreciated.

[flsaders85]

Run your model through Monte Carlo for sure...but I wouldn't commit to Kelly until you're model is proven to beat Pinny closers assuming efficient market....that's how you determine your edge which is approproate for Kelly.

[Baeoz]

Run your model through Monte Carlo for sure...but I wouldn't commit to Kelly until you're model is proven to beat Pinny closers assuming efficient market....that's how you determine your edge which is approproate for Kelly.

I've read Justin 7 (I think) and seen a YouTube vid of his saying that. What about straight head to head betting? I'm not interested in beating the spread, just bet on a winner. How do you determine edge? Sorry to jump in antonyp22, just thought I'd ask too.

I'm trying a monte carlo. I know my winning percentage. I grab all the matches for which I have odds data since last April and then run that period of months, assuming I'll win at that win percentage and record the wins/loses for each odd datum (e.g $1.50). Repeat that 10,000 times, then calculate z-Score and EV for each odd datum and divide EV by 10,000. This gives a list of EV, z-score, percentage for each odds and I see if it's (EV) positive. Am I on the right track?

[flsaders85]

I've read Justin 7 (I think) and seen a YouTube vid of his saying that. What about straight head to head betting? I'm not interested in beating the spread, just bet on a winner. How do you determine edge? Sorry to jump in antonyp22, just thought I'd ask too.

I'm trying a monte carlo. I know my winning percentage. I grab all the matches for which I have odds data since last April and then run that period of months, assuming I'll win at that win percentage and record the wins/loses for each odd datum (e.g $1.50). Repeat that 10,000 times, then calculate z-Score and EV for each odd datum and divide EV by 10,000. This gives a list of EV, z-score, percentage for each odds and I see if it's (EV) positive. Am I on the right track?

Do you have pinny closers for your market? Convert them to their no-vig implied "win percentage." Your edge would be the no-vig pinny implied probality divided by your win percentage minus 1:

(pinny %/your % ) - 1

[Baeoz]

Do you have pinny closers for your market? Convert them to their no-vig implied "win percentage." Your edge would be the no-vig pinny implied probality divided by your win percentage minus 1:

(pinny %/your % ) - 1

Where would I get them from? If there's a historical site I can work it out for high-level, but Pinnacle is scratchy on ITF matches. Is there a mathematical way?

[gloriesam]

use this calculator

http://bettingresource.com/converting-moneylines-odds-into-win-probablity.html