Starting New Trial NBA today... Found Huge Computer Glitch...

[usma1992]

Here goes... all my systems are back tested algorithms. Of course, it doesn't mean anything if it doesn't work moving forward. Many have checked out my college thread. It has been successful so far.

Dave

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
10:40 PM	BROOKLYN	111	LA LAKERS	122		233	-7	227.5		Too Close	Over

[usma1992]

So starting out based on back tested algorithms... here is where the model sits. I have been running it for weeks trying to fine tune my picks. I am ready for a second tests.

I produce my proprietary point totals then check to see where my model is weak or strong and compare it to where Vegas is weak and strong based on a few months of actual results.

Here is the data at beginning of this trial
Overs: 20-5 5% of games 80%
Unders: 37-15 10% of games 71%
Away-:14-6 4% of games 70%
Home-: 23-5 5% of games 82%
Away+: 2-0 <1% of games 100%
Home+:3-0 <1% of games 100%

Again, these games are back tested using an algorithm for each. I have conducted one trial and I was hitting right around 50%. I continue to upgrade. Starts today. Ideally, tonight's game has a 80% chance of hitting as an Over.

Enjoy
Dave

[usma1992]

That was an easy win... hopefully they come that easy later on. Regardless, do your thing.

Dave
The Teach

[usma1992]

1-0 So Far... here are my NBA bets... check out NCAAB... killing it.

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
3:10 PM	MILWAUKEE	139	DETROIT	106		246	13.5	244.0		Away Minus	Too Close
7:10 PM	PHILADELPHIA	131	CHARLOTTE	100		232	11	229.0		Away Minus	Too Close

[usma1992]

0-2 Yesterday on NBA... I need to look at Away - bets. That is my first priority next week. However, I went 19-1 on college... NBA on this trial is 1-2. Here are todays bets... Be careful... 2 Away- bets...

The Teach

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET		Actual Away	Actual Home
6:10 PM	DENVER	142	WASHINGTON	99		241	13.5	238.0		Away Minus	Over		0	0
7:10 PM	BOSTON	128	HOUSTON	104		232	11	227.0		Away Minus	Too Close		0	0
10:10 PM	PORTLAND	104	LA LAKERS	123		227	-10.5	227.5		Too Close	Under		0	0

[usma1992]

Based on todays..today's... information... I'm going to make some minor adjustments after the games are played.

The Teach

[usma1992]

Made one major adjustment... I try and use Vegas spread against themselves. I adjust all the current statistics based on the level of competition they play. I adjust the favorites stats up a little and underdogs down a little.

However, I think there is so much parity in pro sports that I was making too large of an adjustment. I changed my weighting which obviously changed all my past results. So I did a complete revamp with adjusted algorithms. I went 0-4 yesterday so I had to make some adjustments.

I knew I jumped the gun Friday when I relaunched. Starting another trial today... Two bets... I also went back to NFL to make a similar adjustment.

Dave
The Teach

[usma1992]

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
7:10 PM	SAN ANTONIO	102	PHILADELPHIA	132		234	-14.5	237.0		Too Close	Under
8:40 PM	BOSTON	125	DALLAS	114		240	3.5	238.5		Too Close	Over

[usma1992]

Getting pissed... I think the changes I made were more than justified and the model is stronger for doing so. But then I go 0-2 right out of the gate. Not only did I go 0-2... I wasn't close. Both of those bets were so far off. For as great as my college program is doing, NBA is killing me.

I will keep trying ... I don't mind being wrong... but not being close. Then again, I did choose the night to bet the under when Joel Embiid scores 70. What are the chances?

Dave
The Teach

[usma1992]

So I examined both games that the system picked... and this is what I came up with. First, the Dallas vs. Boston game didn't come close to hitting the over.

What I noticed is that both teams do not attempt 2 points shots very often. They are both all about the 3 pt shot. Therefore, if they go cold on the 3 point shot like they did... there is no real chance to hit the over. Both of their adjusted zscores for 2 point attempts were -1 or greater so I created a filter to eliminate those scenarios. It didn't really affect any previous games. Second, filter I added was never to bet an over greater than 241. I think that filter is too high so I lowered it to 237. Therefore, the model will no longer pick an over that is higher than 237. This filter did eliminate some past games but the percentages went up from 73% to 78%.

The Philly under bet was that also way wrong. I examined that also and found that the def away matchup PPG versus the home PPG zscores... were out of whack. Def for San Antonio had adjusted zscore of well of 2... Off for Philadelphia had a zscore of over 4. I added this filter to my under bet criteria. Interestingly enough, it cut the amount of games bet from 10% to 5% of the past bets. But, the percentage probability went from 71% to 82%.
I don't know if this is falling on deaf ears... but this is how I adjust my system daily if I look at games... that I was way incorrect on. I haven't made any adjustments recently to college because it is working. I am hoping this is the last adjustment I need to make for a while.

Give me some feedback... if you get a chance.

Dave

[usma1992]

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
7:10 PM	DENVER	124	INDIANA	113		238	5	240.5		Too Close	Too Close
7:30 PM	NEW YORK	119	BROOKLYN	112		231	4.5	223.0		Too Close	Too Close
8:10 PM	UTAH	110	NEW ORLEANS	124		234	-6.5	238.5		Too Close	Too Close
8:10 PM	PORTLAND	100	OKLAHOMA CITY	126		226	-14	234.5		Too Close	Too Close
10:00 PM	LA LAKERS	106	LA CLIPPERS	124		230	-8.5	230.5		Too Close	Too Close

[usma1992]

Made more upgrades... It picked Portland Under last night. So it is 1-2... but it isn't picking very many games now. Almost none. I'll ride with it for a week or two and see what happens. No picks again tonight.

If it continues to give no picks... I will have to expand the parameters and take a lower percentage.

Dave

[jamesrav]

wouldn't 'no picks' simply indicate your model and the sportsbook model are coming to the same conclusion, and so there's nothing to exploit? Or as you are apparently leaning towards, time to take more risks. I will re-watch that Billy Walters documentary, I seemed to recall he'd take extreme measures just to move the line 1/2 point in the 'wrong' direction in order to then make a huge bet on what he felt was a valid bet. So he was cutting it pretty thin, where 1/2 point was a big deal.

I have serious doubts that the NBA, NFL, or MLB can be 'beat' even with the best machine learning ; maybe NCAA since its still less analyzed than the big pro sports. Seems like the most fertile ground would be in the seemingly 'minor' sports, like women's volleyball or women's tennis.

[usma1992]

No picks again tonight... College last night were 3-0... plus a 2 game ML parlay. Didn't lose a bet last night.

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
7:10 PM	UTAH	125	WASHINGTON	115		240	7.5	245.5		Too Close	Too Close
7:10 PM	PHILADELPHIA	120	INDIANA	113		233	5	236.0		Too Close	Too Close
7:30 PM	BOSTON	121	MIAMI	107		228	6.5	224.0		Too Close	Too Close
7:40 PM	DENVER	116	NEW YORK	111		227	2	223.0		Too Close	Too Close
7:40 PM	MINNESOTA	115	BROOKLYN	110		225	4.5	218.0		Too Close	Too Close
10:00 PM	SACRAMENTO	121	GOLDEN STATE	116		237	2	242.0		Too Close	Too Close
10:40 PM	CHICAGO	113	LA LAKERS	114		226	-4.5	228.0		Too Close	Too Close

[BradytheK9]

You're 100% right about Celtics. I been saying same thing since last season. Me & my dad watch all their games together & early in 2nd qtr we have good idea how game will finish. Im not exactly sure what Joe MaZZZulla does cause doesn't seem like much coaching is going on ever since Smart got traded.

[usma1992]

Final a game... tonight... 2 games in college.

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
7:10 PM	PHOENIX	118	INDIANA	117		235	4.5	243.5		Home Plus	Too Close

[usma1992]

Won last night which was nice... maybe... I just need to patient and wait for system to pick games... or I have to open up the parameters. Which lowers effectiveness but allows you to bet...

No games tonight... Plenty of college to bet on. Check my thread.

Dave

[usma1992]

Another bet today...

Time

AWAY TEAM

PROJ AWAY SCORE

HOME TEAM

PROJ HOME SCORE

TOTAL

VEGAS SPREAD

VEGAS O/U

SPREAD BET

O/U BET

Actual Away

Actual Home

3:40 PM

MEMPHIS

110

INDIANA

123

232

-9

229.0

Home Minus

Too Close

0

0

[usma1992]

Trying to loosen up the parameters in NBA... any advice is appreciated. I am simply looking at the month of January at this point and seeing if I can generate enough winning bets. Right now, the month of January is only generating 17 bets. It is 12-5 ... so that is a nice hit rate... but there isn't a lot of action.

I am comfortable with NCAAB... I feel like it is generating enough picks to make it worthwhile. NBA isn't. It is generating a couple of picks a week.

I will keep trying. Any suggestions... I am all ears. Should I just keep the parameters super tight or run two different systems.

Dave

[prolexus]

Don't chase Action....quality not quantity.....

[usma1992]

Prolexus... I hear you but if it isn't producing any action then... it is worthless. I expanded the parameters and I was able to increase game count in the month of January from 17 to 40. In addition, I actually increased the percentage win rate to up 2 %. So I got more games at a higher percentage.

So I plan to take it for a spin on new trial.

Dave

[prolexus]

Thx Dave. Your efforts and attention to detail, noted and appreciated!

[usma1992]

I was able to upgrade and tighten at the same time... I was just looking at Jan games... I took it from 17 games to 27... modest upgrade. But win percentage went to 78%.

On the looser version it has Philly... GS over... I'm not betting it... 27 games... that's almost 1 a night.

I can handle that... let's see how it runs.

Dave

[usma1992]

I had to open it up... and so these are the picks today... I'll give it another try. I like my upgrades. Both of these picks were on the verge of a No Bet. I am betting them both. College crushed it again. I back tested the model and ran it for the month of January... starting at 1-2... lets see what happens.

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
7:10 PM	LA CLIPPERS	128	WASHINGTON	108		235	12	236.5		Too Close	Over
10:10 PM	MILWAUKEE	125	PORTLAND	111		236	10.5	236.5		Too Close	Over

[Sund]

Close, but just couldn’t quite get there

[usma1992]

My experiment failed... I decided to try and loosen up NBA a little and Overs was one of the areas. I have my parameters set so that it should not bet an over ... greater than 236.5. Of course these games both game in at 236.5 so it didn't hit that parameter.

The second thing I did was artificially adjust the betting so that if my total was within two points of the Vegas total... it would trigger an over bet. This of course got me on both games. My totals were spot on and I shouldn't have adjusted the betting to add 2 points. A mistake that I have already corrected.

I am simply trying to find a way to open up the application without losing too much winning percentage.

I have lowered my over thread hold to 235 and I have taken away the two point add on.

Of course, tonight... no bets. Trying to get better.

Dave

[usma1992]

Time

AWAY TEAM

PROJ AWAY SCORE

HOME TEAM

PROJ HOME SCORE

TOTAL

VEGAS SPREAD

VEGAS O/U

SPREAD BET

O/U BET

Actual Away

Actual Home

7:40 PM

PHOENIX

121

ATLANTA

117

238

3.5

249.0

Too Close

Under

0

0

[usma1992]

Somehow... I earned a push... I watched the game... and both teams were on fire... I thought I was done... my NBA filters are very tight. If it picks games... it reached reasonably criteria. Away - hasn't missed. Two games tonight... surprisingly.

Time

AWAY TEAM

PROJ AWAY SCORE

HOME TEAM

PROJ HOME SCORE

TOTAL

VEGAS SPREAD

VEGAS O/U

SPREAD BET

O/U BET

Actual Away

Actual Home

6:10 PM

INDIANA

131

CHARLOTTE

112

243

9.5

237.0

Away Minus

Too Close

0

0

7:10 PM

HOUSTON

106

MINNESOTA

111

217

-7

221.5

Too Close

Under

0

0

[usma1992]

Maybe the system... is finally tight enough but working. Both games looked great to me... 3-2 now... with new parameters. I will take it... I am going to probably throw out all Philadelphia bets for a month because of Embiid injury. All my stats are based on him playing. I do the same thing with college football when QB goes down. Not worth betting.

No games tonight.

[prolexus]

Agreed, thank you

[usma1992]

Time

AWAY TEAM

PROJ AWAY SCORE

HOME TEAM

PROJ HOME SCORE

TOTAL

VEGAS SPREAD

VEGAS O/U

SPREAD BET

O/U BET

Actual Away

Actual Home

7:10 PM

CLEVELAND

127

WASHINGTON

107

234

11.5

233.5

Too Close

Over

0

0

10:10 PM

NEW ORLEANS

110

LA CLIPPERS

118

228

-7

229.5

Too Close

Under

0

0

Two Bets Tonight... Rare

[bubba99]

Time

AWAY TEAM

PROJ AWAY SCORE

HOME TEAM

PROJ HOME SCORE

TOTAL

VEGAS SPREAD

VEGAS O/U

SPREAD BET

O/U BET

Actual Away

Actual Home

7:10 PM

CLEVELAND

127

WASHINGTON

107

234

11.5

233.5

Too Close

Over

0

0

10:10 PM

NEW ORLEANS

110

LA CLIPPERS

118

228

-7

229.5

Too Close

Under

0

0

Two Bets Tonight... Rare

Just trying to understand your system, how is Cleveland projected to win by 20 against an 11.5 spread and it's considered too close
But the total projected at 234 against 233.5 and it's a play?
Just curious, great job so far.

[Sund]

Your over and unders look too close ? I’m not sure how your system works! ? Thanks for all the work you put into this!

[usma1992]

So... again... I know it gets confusing. Maybe, I should eliminate my projections and just produce the bet without the numbers. Understandably, the numbers can get confusing.

First, the projected numbers are just the first step in producing a pick. After generating numbers, my system looks at match ups and examines past history to see what type of results were generated with similar matchups.

So even though, it looks like a solid bet... the match up eliminates the bet.

As far as the over/under... yes, they look very close to Vegas odds, I noticed that. However, based on past history... these bets triggered and so I guess you have a choice to run with it or not. I bet both games... I will continue to tweak the system. Let's see how it does tonight.

I most curious on the FGCU ... E Kentucky in college.

Good Luck
Dave

[Sund]

Thank you ! I’m sure it’s frustrating, getting these same dumb questions over & over again ! I’m playing the college games and will watch the outcome of the NBA ! Good luck