Anybody have a model and set your own lines?

[chilidog]

Something that always intrigued me was how do books set the opening lines (not the ones that just mirror pinny or bet365). When I used to work at a book, we subscribed to betradar which set the lines for us, then we'd adjust based on bet activity. We were rarely balanced with action and usually laid off really lopsided action with other books or betfair.

Anyway, after fucking around with various systems/ideas I've had over the years, I started learning how to set my own lines based on each sport. I've been able to come pretty damn close to the NFL/CFB openers, but still haven't cracked NBA/CBB. Haven't even touched MLB yet.

Anybody else play with setting your own lines based on stats?

 

[BMR Genie]

Nice work, Chili! Looks like it takes a lot of experience and time to learn how to set your own lines. But why did you stop working in sportsbooks?

 

[RudyRuder]

i usually just randomly go down and guess what the lines are


if my line is wrong by ~3 points...i bet the opposite team

 

[Tanko]

I have a model that I use for CFB. I've been using it for several years and make minor adjustments as trends change. I try to take advantage of the game lines that disagree with my model. Injuries are incorporated into the model but, how you do that that is a very subjective process and can lead to mistakes.

How well does it do? One measure of performance is how well you beat the closing lines since that is statistically the best indicator of the market.

This year it is generating selections that are on average ~0.92 pts better than the closing lines. This is within the typical range I've seen over the years (0.80-0.99). In 2021, I'm beating the closing line on ~ 75% of selected games but only yielding a 58% win rate. I'm still evaluating the data to understand why there is a large delta.

Still lots of improvements to incorporate into the model as time allows.

 

[Roy Bacon]

i usually just randomly go down and guess what the lines are


if my line is wrong by ~3 points...i bet the opposite team

I would do something similar. Make my own lines in the blind and every odd that was off by 2 points I would investigate as to why. Usually there was a reason like a key injury.

You can also do a in the blind slot line which is very interesting. You make two lines for each game. 1st is what number would you lay to bet the favorite. 2nd line, on the same game, how many points would you need to bet the dog.

 

[chilidog]

I have a model that I use for CFB. I've been using it for several years and make minor adjustments as trends change. I try to take advantage of the game lines that disagree with my model. Injuries are incorporated into the model but, how you do that that is a very subjective process and can lead to mistakes.

How well does it do? One measure of performance is how well you beat the closing lines since that is statistically the best indicator of the market.

This year it is generating selections that are on average ~0.92 pts better than the closing lines. This is within the typical range I've seen over the years (0.80-0.99). In 2021, I'm beating the closing line on ~ 75% of selected games but only yielding a 58% win rate. I'm still evaluating the data to understand why there is a large delta.

Still lots of improvements to incorporate into the model as time allows.

That's one thing I'm not tracking that I need to add to my spreadsheet. Are you tracking CLV as the no-vig Pinny closer, or just the Pinny closer? And then, how're you handling ATS bets where the line you bet isn't a -110 (for example, -7 -115 and the line closes at -7.5 -110). Would it be better to convert the spread & line to ML decimal odds and track CLV that way?

 

[chilidog]

Nice work, Chili! Looks like it takes a lot of experience and time to learn how to set your own lines. But why did you stop working in sportsbooks?

It took a lot of research and reading sites that discuss a lot of advanced stats. Excel and Python become your best tools.

I'd say start with these 3 and go down the rabbit hole.

1. https://medium.com/@lloyddanzig/quantitative-sports-betting-6976e1ceaf0f

 

[Roy Bacon]

I wonder how many guys actually cap a game vs bet their hunch?

 

[Pimike]

It took a lot of research and reading sites that discuss a lot of advanced stats. Excel and Python become your best tools.

I'd say start with these 3 and go down the rabbit hole.

1. https://medium.com/@lloyddanzig/quantitative-sports-betting-6976e1ceaf0f

Saw thus

yes good stuff

 

[chilidog]

I wonder how many guys actually cap a game vs bet their hunch?

Based on friends that also bet, I'd say most people just bet their hunch. I don't know anybody on a personal level that caps the games.

 

[Roy Bacon]

Based on friends that also bet, I'd say most people just bet their hunch. I don't know anybody on a personal level that caps the games.

Yea, I don't either actually. Seems suicidal. Oddstrader has made researching box scores a lot easier.

 

[chilidog]

Yea, I don't either actually. Seems suicidal. Oddstrader has made researching box scores a lot easier.

i don't cap games in the old-school sense. i have models based on stats that I run with python and excel and they kick out the bets to look at, then i make my decisions. doesn't take much time at all.