NBA Stituational Bet, SDQL

Originally Posted by hyahya

Nash, can you detail how you calculate your own rating? I know you've broken it down earlier in the thread but could you provide an example?

I prefer to have a sports related rating system. Which means I treat every sport by its own queries and standards, the limits in the NBA does not apply to other sports and vice versa.
If i bet on Totals or ATS, i use z-score, percentage of years in profit since 2006, overall profit, last years profit, ROI%, last 3 years profit.
If i use SU, i leave z-score out.
So there are 5 or 6 main areas of evaluation. In each category for each sport i divide the queries in three groups. Best 33%, Middle 33% and Lowest 33%. Based on that they get 1 to 3 Points. In the end I calculate the Avg points for each query and in the end I take the best 33% Avg score wise. So let's say we have 270 NBA queries. This will leave us approx. 90 queries with 3 stars. The more queries we have the better in avg our results may be.

NBA SDQL Best Bet

1-6-15 7:58a CT

Play: PHX -3.5 -110

A and line<=-2 and p:HW and pp:HW and WP>50 and o:WP>50 and 2006<=season

Play on a road fav of 2 or more off b2b home wins a winning team, playing a winning team.

a few pointing to the Suns tonight

Originally Posted by nash13

I prefer to have a sports related rating system. Which means I treat every sport by its own queries and standards, the limits in the NBA does not apply to other sports and vice versa.
If i bet on Totals or ATS, i use z-score, percentage of years in profit since 2006, overall profit, last years profit, ROI%, last 3 years profit.
If i use SU, i leave z-score out.
So there are 5 or 6 main areas of evaluation. In each category for each sport i divide the queries in three groups. Best 33%, Middle 33% and Lowest 33%. Based on that they get 1 to 3 Points. In the end I calculate the Avg points for each query and in the end I take the best 33% Avg score wise. So let's say we have 270 NBA queries. This will leave us approx. 90 queries with 3 stars. The more queries we have the better in avg our results may be.

Thanks for the clarification on your ratings. I don't suppose you would be able to provide ratings for each query? I can't imagine that it would be feasible to do that but just thought I would ask as your results in NBA on your 3 stars are staggering.

Also, correct me if my knowledge about z-scores are incorrect, but I assume z-scores above 2 is significant and z-scores above 3 are "very significant"...so that may be a way to filter the trends accordingly

Is there anybody kind enough to tell me, of the 254 queries currently in the database, how many of them are active for today's two games? So far I have only picked up on one (NBA193)....

Originally Posted by lion043090

Also, correct me if my knowledge about z-scores are incorrect, but I assume z-scores above 2 is significant and z-scores above 3 are "very significant"...so that may be a way to filter the trends accordingly

In traditional stats, like you said, a z-score of 1.96 and higher is signficant with a .05 chance the results happened by chance; and a z-score of 2.58 or higher is signficant with a .01 chance the results happened by chance. Of course that only applies to the data if it's normally distributed. Normally, if you had the standard deviation and the mean, you could figure it out.

But in this context with this data I'm not sure if the z-score is calculated in that same traditional way, and therefore, the range of values and the significance level of values depending on sample size, will probably differ.. .might want to find out out how exactly z-scores are calculated in this context.

Nash thanks for the thoughts on ratings. Thats really helpful

I know some have a strategy of flat betting and that has been mine but the analysis is quite clear in my view that playing the pick by number of trends active for it is more profitable. Nearly 3x of you just straight played all ATS queries in the sheet this year with 1 unit per trend. This is also a much higher variability method so good bankroll management is key. Surprisingly there is very little extra value in playing trends where there is any contradiction. Like if it was 3 for team 1 and 1 for team 2 betting on team 1 is not much extra return. It also makes things simple in picking the plays. I've been workimg on excel to create seasonal models for nba ats using all the trends to optimize return and minimize risk and can probably share this within a week or two.

For those interested I have an excel took that automatically flags the trends and match ups. I originally thought maybe people would pay a small fee for it but there was no interest. I may just give away the basic one that just flags trends. My personal version pulls all historical data for the trends and presents it as well as the current season records. It's been quite helpful in miniMizing bad trends.

Originally Posted by Consigliere

Nash thanks for the thoughts on ratings. Thats really helpful

I know some have a strategy of flat betting and that has been mine but the analysis is quite clear in my view that playing the pick by number of trends active for it is more profitable. Nearly 3x of you just straight played all ATS queries in the sheet this year with 1 unit per trend. This is also a much higher variability method so good bankroll management is key. Surprisingly there is very little extra value in playing trends where there is any contradiction. Like if it was 3 for team 1 and 1 for team 2 betting on team 1 is not much extra return. It also makes things simple in picking the plays. I've been workimg on excel to create seasonal models for nba ats using all the trends to optimize return and minimize risk and can probably share this within a week or two.

For those interested I have an excel took that automatically flags the trends and match ups. I originally thought maybe people would pay a small fee for it but there was no interest. I may just give away the basic one that just flags trends. My personal version pulls all historical data for the trends and presents it as well as the current season records. It's been quite helpful in miniMizing bad trends.

Great explanation, i would like to send you my analysis sheet, but i have no automatic play generating tool so far, only opening the sites in excel via web automatization.

secondly you only cover NBA right? i think there is a lot of value in NHL and MLB too.

Originally Posted by nash13

Great explanation, i would like to send you my analysis sheet, but i have no automatic play generating tool so far, only opening the sites in excel via web automatization.

secondly you only cover NBA right? i think there is a lot of value in NHL and MLB too.

Nash, I'm developing it for NBA right now but I can port it over to cover anything in the future. Mods would need to be made because of the data structures but pretty simple once the base code is layed down. The seasonal models etc would be more difficult. The big body of work is the initial query setup for historical and current season. Once that is done the year to year updates are not too much. At the rate queries are being added though keeping up with the data is difficult. Currently I see no way of doing an NHL version for this season but I will definitely do an MLB version since the NBA tool should be completed by that time.

Originally Posted by pip2

Is there anybody kind enough to tell me, of the 254 queries currently in the database, how many of them are active for today's two games? So far I have only picked up on one (NBA193)....

Sorry pip was busy today, but my filtered version of the ATS queries had 2 active for Phx and 1 for Det. Hit on both fronts...good day to start laying out 1 unit per trend!

Originally Posted by nash13

Great explanation, i would like to send you my analysis sheet, but i have no automatic play generating tool so far, only opening the sites in excel via web automatization.

secondly you only cover NBA right? i think there is a lot of value in NHL and MLB too.

I've been spending a little more time on NHL, and I think there is definitely good value to be had there. I'm waiting to see how your demo works. If it can handle NHL well, hopefully we can play with the numbers and figure out the best way to handle those trends.
Thanks for the backtesting results for the NBA! Great info to have, and definitely makes me think it's time to reassess. I played for lots of volume early in the season and got stung, but it looks like it was just a matter of optimizing and sticking with it.

Originally Posted by Consigliere

Sorry pip was busy today, but my filtered version of the ATS queries had 2 active for Phx and 1 for Det. Hit on both fronts...good day to start laying out 1 unit per trend!

Thanks consigliere. I found 2 queries, and I was worried I might be missing a bunch. I guess I missed one of the Phoenix queries...

Originally Posted by nash13

The problem here is, that it is linked to a Excel extension i recentley bought. But if the development of the extern program works, there will be no need for that.

@all i updated the recet seasons NBA Results.

Overview:
All Picks with double plays and contradictions:
ALL: 973-762 (P:123,43 Y:7,11% SR:56,08%)
ATS: 617-479 (P:82,47 Y:7,52% SR:56,3%)
TOT: 356-283 (P:40,96 Y:6,41% SR:55,71%)

Without Contradictions (Double Plays allowed)
ALL: 616-415 (P:145,56 Y:14,12% SR:59,75%)
ATS: 389-267 (P:86,99 Y:13,26% SR:59,3%)
TOT: 227-148 (P:58,57 Y:15,62% SR:60,53%)

Single Plays only, Contradictions and Double Plays eliminated
ALL: 317-229 (P:59,47 Y:10,89% SR:58,06%)
ATS: 167-121 (P:30,97 Y:10,75% SR:57,99%)
TOT: 150-108 (P:28,5 Y:11,05% SR:58,14%)

My Own System (Double Plays allowed, Contradictions disallowed, Only 3-Star Plays)
ALL: 318-190 (P:99,38 Y:19,56% SR:62,60%)
ATS: 220-138 (P:62,2 Y:17,37% SR:61,45%)
TOT: 98-52 (P:37,18 Y:24,79% SR:65,33%)

Profit by month:
Oct: 0,55
Nov: 84,77
Dec: 10,40
Jan: 3,66

So take what you want from it, but I think taking a road along trusted plays pays out big time.
I will reconsider investing money in the program, since the demo will arrive end of the week.
It could save time. Focus on playing the games and switching the analysis to other sports (esp NHL and MLB aswell)

Nash, thank you for this extremely valuable summary/report!

Please forgive my ignorance but what exactly does the excel extension you are demoing do?

NBA SDQL "Best Bet" 2014-15

Overall Group Record - 108-97-3 - 52.6% (+4.51) (max wagered -to win 1 unit)

Yesterday's recap 1-0 (+1.00)

Official Entries: Max 10 of 10: (no particular order)

1. JMon - 21-22-1 (-2.05)

2. pip2 - 22-19 (+1.85)

3. nash13 - 7-4 (+2.82)

4. Ronald S. - 10-17 (-7.98)

5. hyahya - 7-6-1 (+.35)

6. Mako-SBR - 15-16 (-2.60)

7. Heart - 10-1 (+8.92)

8. Consigliere - 13-8-1 (+4.50)

9. GolfAddict - 3-4 (-1.30)

10. Open Spot - let me know if any one wants it