Algorithm Picks - Day 1 of 10

Hey, so I've developed an algorithm that predicts NBA scores and will consistently beat the closing line, I'm comfortably up so far this season, playing about 70% of games. We all know b*llsh*t is a dime a dozen on sites like this, so just for fun I'll post 10 days worth of lines and keep a running tally of wins/CLV. I'll post lines for every game before the Vegas lines come out, and just to make things simple, we'll say that any game that opens at 3 pt difference (for spread and O/U) will be bet at the opening line, and I'll use that to tally wins/CLV.

Obviously, with sports betting, anybody who ever "guarantees" winning picks is either a liar or a fool, and there will be down days. But I will say that I'm confident at the end of 10 days, I'll be ahead.

If you decide to bet using these, first, that's on your own risk. Being right on the games isn't always a guarantee of long term success if you don't bet using proper risk management. The lines are entirely data driven, using both public stats and a few I created using different combinations of public stats I don't have any inside knowledge and I don't even really look at the matchups. That's all done in the algorithm, I just punch in the teams and it spits out the numbers. It does account for home court advantage in both the spread and the O/U so don't factor it in. The only two factors outside the algoirthm I look at before placing a bet are injuries and back-to-back scheduling, as the algorithm doesn't account for those (although if the injured player is out for maybe 6-7 games the added/lost value will start to be reflected). It's meant to be a pure numbers analysis, asking the question "what would we expect the score to be if the expected version of Team A played the expected version of Team B?". 

For tomorrow's lines, here's what I've got, listed by favorite:

Bkn @ Orl: Bkn -1.5/210

Wsh @ Cle: Wash -2/215

Det @ Hou: Hou -8/215

Tor @ Bos: Bos -2.5/208

Mil @ NYK: NYK -4/213

Dal @ Chi: -4/209

OKC @ NOP: NOP -1/210

Cha @ Mem: Cha -3/223

Sac @ SAS: SAS -1/217

Ind @ Den: Den -2.5/212

Por @ Phx:  PHX -2/220

Min @ GS: GS -11/218

Mia @ LAL: Mia -1.5/217

sorry, that was meant to be Mem -3, not Cha

So the algorithm has Orlando as a better team the Nets?

@asmth14

 No, I'd take GS. 

The algorithm says GS is an expected 11 pt fav, Vegas opens with GS at a 7 pt fav, so I see value at GS -7, as I'm taking the opening line at undervaluing GS. Is it undervaluing it by 4 pts? Don't know, and it's irrelevant. Even if it's undervalued by 1 pt, I want that bet.

The best way to think of the algorithm is as an inefficiency finder, not necessarily a score predictor. Of course when I built it, it was designed to be a score predictor. But that's not what we're betting on. I'm not betting on an exact score, I'm simply betting on if Vegas line is efficient or not, which it often isn't, at least at the open. As you say, sometimes there's a "reason" a line opens where it does. But my assumption is always that it's because the line itself is inefficient and it will eventually move towards my line by the close. It doesn't always of course, but more often than not, it does.

This system isn't about trying to find "a lock" or a sure thing; Those don't exist. It's about playing the numbers and exploiting inefficiencies. If you do that consistently, you'll win consistently over the long term. It's about odds, really. A casino knows it only has a slight edge in blackjack (I think 51-52%?). It knows it can't predict of it will make money over, say, 10 hands. But over enough hands, it always will, because the odds are on its side. Same thing here.

To account for noise, usually I'll only take a difference of 3 pts as a bet signal. So in the Minn/GS game, if the line had opened with GS as a fav between 8.5-13.5  pts, I don't bet b/c I can't be sure I have an advantage. Anything GS -8 and below, I'm betting GS because I thinj Vegas is undervaluing GS. Anything GS -14 and above, I'm taking Minn, because I think Vegas is overvaluing GS.

So for example, the algo says Hou should an 8 pt fav. Vegas opened at Hou -1.5, so it's an easy Hou bet. Does that mean my line was "right" and Vegas is off by 6.5 pts? No, not necessarily. Maybe it is maybe it isn't, it's irrelevant. Because even if the line is only off by 2 pts and Hou wants by 2, I still win with that bet.

And so far today, Vegas has moved to Hou -3.5, so even if it stays there, I've locked in value and that's a solid bet, regardless of what happens in the game.

@Freddy66

 No,  it says Brooklyn is the better team, but in this game, they're only a 1.5 pt favorite, and not a 9 pt favorite. So no they're not the better team, but yes the bet is on Orlando.

Also remember this factors in home court advantage. If I plug in Orl @ Bkn, it gives me a line of Bkn -7

Also, for a bit of background, Ive got a degree in statistics and even though I'm a basketball fan, I approached this first and foremost as a mathematician. The algorithm was developed by running tens of thousands of simulations of historical data to find lesser known correlations between stats and scoring, developing custom stats to capture those, and running regression analysis on the current seasons data using the least squares method. Obviously I know how the algorithm works, but I don't know exactly why it spits out the numbers it does on any given game, other than that I trust it as generally efficient.

davidss is this you?

@TheBetGuy

 Ok thanks

This sounds great, thanks for sharing @ThBetGuy

I have a stats degree myself from years ago, just never utilized it in an official model. Very curious to see how this works out and if it gets more efficient as the season goes on.

Good luck to us! 

Post the picks

One question. Do you bet on every gam3,  Or only certain games?

80% of your algorithm picks basically agree with the current lines. i do not need an algorithm to tell me what the current lines are.

and some of your lines are absurdly off. Nets line s/b 1.5? rockets line s/b 8?

This seems very interesting

I'll be checking in on your model for the next 10 days

So based on those numbers you listed, what are your picks today? thanks!

@TheBetGuy

 Based on your listed lines, these are the lines on my book that are 3 pts or more off.

ORL +10

OKC +5

GS -7.5

LA +5

ORL OVER 210.5

DAL UNDER 215

CHAR UNDER 229

SAC UNDER 223

MIN UNDER 222.5