Anyone actually use betting models? worth learning?
ok so ive been betting on nhl and some nba for about a year now, mostly just gut feeling and reading injury news. been roughly breakeven which is honestly probably lucky lol. keep seeing people talk about building their own models in excel or python and im wondering if its actually worth the time to learn that stuff or if its just nerds overcomplicating things. like does it actually give you an edge?
13 answers
Top Answer
210
honestly yeah it changed how i bet completely but its not magic. i started with a really basic excel sheet for nhl, just expected goals stuff and adjusting for rest and back to backs. took me months to get something that wasnt garbage tbh. the real value isnt that the model is perfect, its that it forces you to actually think about WHY youre betting something instead of vibes. i line shop my model numbers against Boomerang Bet and a couple others and only bet when theres a gap. most days theres nothing worth betting and thats kinda the point. if you cant sit on your hands when the model says no edge then dont bother building one
83
can confirm. i did basically the same thing with hockey and the biggest change for me wasnt even profit at first, it was that i stopped making dumb impulse bets. once you have a number in your head you trust you just stop chasing. took me like a full season to trust it tho
115
wait when you say model do you mean like predicting the exact score or just probabilities? im confused what the output even is supposed to be
↳
88
probabilities, not exact scores. like my model spits out team A has a 58% chance to win. you convert that to odds (so ~1.72) and if the book is offering 1.90 on that team theres value cause theyre implying a lower probability. exact scores are basically impossible to predict consistently, dont even try
61
fwiw the math part isnt actually the hard part. converting probability to odds is just division. the hard part is getting clean data and figuring out which inputs actually matter vs which ones are noise. a lot of stuff that feels important (recent form, momentum etc) barely moves the needle
60
yeah this. spent way too long adding features that did nothing. simpler models with fewer good inputs almost always beat the kitchen sink approach. overfitting is real and it'll wreck you when the season actually starts
35
for someone who only knows excel is python actually necessary or can you get pretty far with spreadsheets? dont really wanna learn coding ngl
↳
36
you can get surprisingly far with excel honestly. lots of profitable bettors never touch code. python helps when you want to automate pulling data and backtest over thousands of games quickly but if youre doing one league manually excel is fine. dont let the python crowd gatekeep you
28
one thing nobody mentioned, your model is only as good as the closing line you compare it to. if you consistently beat the closing line (CLV) thats the real proof youre doing something right, not your short term profit. track that
13
agree on CLV. profit is noisy over small samples, you can be sharp and lose for months. beating the close is the signal that you actually have an edge even if variance is killing you short term
20
idk man if these models actually worked wouldnt everyone just be rich? the books arent stupid and they have way more data and smarter people than some dude in excel. feels like cope to me honestly
↳
11
i thought that too but heres the thing, you dont need to beat the books overall model, you just need to find the spots where their line is soft. they cant perfectly price every game in every league, especially smaller markets and props. nobodys getting rich quick but a small consistent edge on the right markets is very real. also yeah most people who try DO fail, thats not a contradiction
8
late to this but can vouch. built a dumb little nba totals model last year and it lost money the first month and i almost quit. stuck with it, beat closing on like 56% of my bets over the season and ended up green. the discipline it gives you is worth it even if you never go full quant. just be ready to suck for a while first