NBA Algorithm Picks - Day 6 of 10 (25-16 so far)

Well, it was bound to happen, our first non-winning day, taking a push with a 5-5 yesterday, bringing our total over 4 days to an enviable 25-16.

recap:

Overall: 25-16 (12-8 spreads, 13-8 O/U)

Day 1: 8-5 (4-3 spreads, 4-2 O/U)

Day 2: 3-1 (1-1 spreads, 2-0 O/U)

Day 3: 9-5 (4-3 spreads, 5-2 O/U)

Day 4: 5-5 (3-1 spreads, 2-4 O/U)

Day 5: TBP

For confirmation of the results so far, click on my name to see my post history and click on the thread for each day. Make note of the timestamps to confirm that all picks were made well before each game and make sure the tally on each day's thread matches the recap above. I can assure you they do, I'm just doing it this way so you don't need to take my word for it.

Here are the lines for tomorrow, as always once the opening lines come out later tonight, I'll post the picks based on that here on this thread:

BOS @CLE --------> Bos -1.5/206

SAC @ DET-------->pick em/219

NOP @ WAS -------> WAS -7/211

ORL @ ATL-------> pick em/211

IND @NYK-------> Ind -2/212

MIA @ OKC-------> OKC -6/211

DEN @ DAL-------> Dal -4/210

HOU @ MEM-------> Mem -1/215

PHX @ MIN------->  Phx -5.5/218

TOR @ POR------->  Por -3/214

CHI @ LAL --------> LAL -2.5/217

This is good stuff. Winning picks and complete transparency. Keep up the good work!

Based on the opening lines, here are the following picks for today's games. Looking for a rebound after a down day yesterday. In case you're new to these threads and wondering where these bets come from, our independent lines are posted the day before games (and hours before Vegas lines come out). When we spot an edge in the Vegas lines, we play it. An "edge" is defined as 3pts or more in either direction. So if our line says the total for a game is 210, than anything that Vegas comes out with between 207.5-212.5 is considered a dead zone and theres no play. Anything 207 and below and we'd take the Over. Anything 213 and above and we'd take the Under. If you look at the lines we posted above, and compare to the Vegas opening lines you can find at vegas-insider dot com, using the "rule of 3 pts" I just explained, you'll see where these picks came from and why some games are played and some aren't.

If you wish to use these to bet, and the lines here are different than the lines available at the time you wish to make your bet, go back and check the original lines we posted above to see if there's still value left in the current lines. 

Spreads:

DET +4.5

WAS -2

ORL +10.5

OKC +9.5* 

HOU +11

PHX -3

POR +1

*quick note about the MIA-OKC line, I made typo above in posting the alorithm lines, it was actually Mia -6, not OKC. Doesn't change anything as relates to the bets, either way the play is still OKC +9.5, just correcting the record)

Over/Under

BOS @CLE - Over 200

SAC @ DET- Over 213

NOP @ WAS - Under 215

ORL @ ATL - Under 214

IND @NYK - Under 217

HOU @ MEM - Under 221

PHX @ MIN - Under 222

Thanks for the info. Are you going to update the results as well?

@BBGunzz

 Results are updated on the next days thread, where I keep the running tallys. The results for today's games (Day 6) can be found on the thread for tomorrow's games (Day 7) which was just posted, thanks for following :)

So how do you account for injuries that we do not know until game time and the line swings?

@Ilost

 Ok, I am trying the over in boston game. Going to lock in now at over 199.5.  Looks like a good advantage there. Thanks

@TheBetGuy

 Also just wondering why totals do not have any half points? Do you round up or down?

@Ilost

 I don't. The algorithm is 100% stats-based, the idea being to take out all the opinions and "gut feels" out of it, which are always subject to our own biases and is what makes the average better lose money.

But of course, even though the algorithm doesn't technically factor in injuries, doesn't mean we shouldn't, as they obviously can be probability-changing events, here's how I do it.

First, when I say it "technically" doesn't factor in injuries, what I mean is, it kind of does and kind of doesn't. Officially, it doesn't, as it's 100% stats based. But I also use a weighting that puts roughly 70% of the weight on the previous 10 games, so once an injured player has been out for around 7-8 games, the algorithm starts to reflect the loss of the injured players production, as majority of the algorithms calculations will be done using stats the team has put up without them. On the other hand, if a player got injured in last night's game, and they've been a big part of the teams overall production up until today, then obviously the numbers the algorithm will use still assumes that the injured player is in the lineup and the probabilities won't be reliable for that team for at least 4-5 games.

For example, if Steph Curry were hurt in last nights game, than I would just not bet any GSW games for at least 4-5 games until his absence starts to get reflected in the algo. But if he's already been out for 10 games, than I know that the algo is using numbers the Warriors put up without him and I'll use it as normal.

And in the example you gave of if a player gets hurt, say, today after the bets are made, I don't bother with it. That would be, by definition, a random event (at least random to us since we have no way of knowing who might get placed on the IR list on any given day) and since it's random, it's both impossible to try to account for, and pointless, as if/when it occurs it's just as likely to help us as hurt us.

In any case, it happens so rarely that it's still far better to simply bet the lines at the open. For every one game where that hurts you b/c a star player gets hurt and the lines moved (which is overall a relatively rare event) you'll probably have dozens of games where the early bets have helped you just from the regular movement of the lines throughout the day.

Take a look at the O/U bets we made today at the open compared to where the lines are now.

Sac/Det - 1.5 pts of value movements

Ind/Phi - 4 pts of value

Hou/Mem - 3 pts

NOP/WAS - 7.5 (!)

Now, to be clear, this much value is an unusual amount, and of course, sometimes the games go against us, but in general, I usually expect to see 3-4 pts in value by betting the opening line with my own numbers than I would by waiting until the close to see for injuries. If a star player gets injured between now and the games, first of all I have a 50/50 chance that it helps me (If it's Harden, for example, that can only help my  OKC + 9.5) but even if it doesn't, and it completely kills one of my spreads, it was still far better to bet the games at the open and get to capture the moves in all the other games.

 Regarding the totals, I do get decimal numbers, I just round up/down, same as spreads, but with spreads I do use half numbers, just because a half point can represent up to 50% of the total when you're talking spreads, but not really a factor either way when talking about totals. In the case of Bos/Cle, the actual line is Bos -1.67 and the total is 205.84

@TheBetGuy

 FYI  It is Towns birthday today and no one from Phoenix can stop him !  Both him and Edwards both had over 40 points last time they played !

@middleman820

 So I'm gonna pick on you a little bit here, not to be a dick, but just because this is a perfect example of the type of bad "gut feel" analysis I mention above, which accounts for why most bettors lose money. It also presents a good opportunity to show how proper analysis should be done.

Even at first glance, this seems like a poor factor to include in your analysis. What's the mechanism of a player performing better on their birthday? It isn't obvious to me. Is it that they try a little harder on their birthday? Their opponents go a little easier on them? Both of those seem like a stretch. In addition, it could also be a negative. How do you know the player wasn't out all night partying with strippers BECAUSE it's his birthday, and he's hungover af for the game today? I certainly don't know, the point is, none of us do, which makes this sort of thing more or less impossible to analyze.

The other part of the analysis, how Towns and Edwards both had 40 points is a little closer to the mark, but still not really useful. First of all, they haven't played Phoenix all season, the game you're referring to was last season. So the team he's going to play tonight isn't the team he played then, he's playing the 2021 Suns team today, and he put up those numbers against the 2020 Suns team. Second, both Edwards and Towns average PPG are in the low 20's, which means them scoring 40 points were extreme outliers towards their typical production. Of course outliers can and do happen, but no real analysis can ever be based on the potential of a player (or team) to hit their outliers, you always have to assume some sort of mean reversion. Of course it's possible for Towns to have a 40 pt game, it's also just as likely that he has an outlier game on the other end, more like he did last game against the Clippers where he scored 8 pts. The only plausible assumption that can be made for any player or team is that, in general, their performance on any given game will look most like their typical performance

As an example of how proper analysis would be done, let's go back to that birthday stat. Let's assume that there actually is some validity to the idea that a player plays better on their birthday. It's not enough to then just bet every team who has a birthday that day, you need to know how much of a boost that gives him in order to place a good bet. So you'd start by analyzing as much data as you can about how players performed on their birthday vs their average performance to get a baseline number. Let's say you do that and, lo and behold, the average player scores 5.2% more points over their average performance on their birthday. That's a useful stat. So you then go back to the Phx-Min game and you take Towns average PPG, and increase it by 5.2%. Of course, that only accounts for about 1 pt overall, so probably not a factor.

I've simplified it a lot in this example, but that's fundamentally how handicapping works. You start with a hypothesis about a stat/factor, you then sift through the historical data to confirm that the hypothesis does in fact correlate to scoring points, then try to quantify how much exactly it improves/decreases a teams ability to score, and then apply those factors to the teams average scoring production. This gives you an overall "expected score" for that team, and then you need to try to predict what happens when a team with that expected production plays a team with teh defensive skill of whoever their opponent is for a given game. Because offensive production is only half the battle, the opponent matters too. An average offense playing against an average defense would be expected to score around their average. But if they played a below average defense, we would expect them to score more than they typically do. And quantifying how much above/below their average they should score based on their opponent is basically the crux of handicapping.

Good work.

Good stuff. Def agree and disagree with a few of these. Even though I’m all for data. I believe the best way to cap is watch games and players BOL 

@PGAromes82

 I think there's a place for that (factoring in one's personal opinion in their capping), perhaps to make a judgement call when the numbers are right on the edge of the 'bet/don't bet' spectrum, and assuming that you're genuinely knowledgable about the game. On the overall though, I think the numbers always have to lead, and once you start overruling your statistical analysis with your gut I think you're going to get into trouble. The problem is humans are notoriously bad at assessing our own expertise and we tend to overestimate our ability to make future predictions about these things.  I'm sure there are cappers out there who are successful doing just that, of course, and if that strategy works for someone, obviously they should keep doing it. Overall though, I think most bettors rely far too heavily on their gut and not nearly enough on the numbers.