NBA System Back-Testing Available

I have been learning the NBA SDQL at this link:https://killersports.com/nba.py/query?sid=guest&text=+&=1&submit=query

And I could use some more practice.

If anyone has a stats-based system they're wondering about, let me know and I'll see if I can back-test it.

The data base here goes back at least to 1995 (from some of my test results) all the way to current, and the query is pretty powerful, able to look at previous-game(s) results and all, days rest, lots of stuff.

Yes!

Interesting .. however, at the moment that degree of complexity appears to be beyond my knowledge of the query .. and it may be beyond the query itself except in little bits at a time.  I'll take a look at it when I'm more adept at the query.

Here's an example of what the query can do in the NBA using this database that I've pretty much confirmed goes from 1995 thru today.

Supposing a team with a SU win percentage of from .650 thru .750, a really good team, plays a team with a winning percentage from .200 thru .400, a pretty bad team.  What would be the ATS result.

The query would be: WP>=65.00 and WP<=75.00 and o:WP>=20.00 and o:WP<=40.00 .. and the results are 704-713-29.  The Books know what they're doing.

However .. supposing the weak team's previous game was a blowout for them by 15 or more points, do the books adjust for that?

This query would be: WP>=65.00 and WP<=75.00 and o:WP>=20.00 and o:WP<=40.00 and o:p:margin>=15.0 .. and the results are 146-169-6, so the Books adjust for the weaker team's little rise in self-confidence, but not significantly, either that or the stronger team compensates, as the the weaker team only wins ATS 53.65 percent of these games, barely enough to break even at -110. 

But what if the poor weaker team themselves got blown out by 15 or more points in its previous game?  What affect does that have.

The query for that is: WP>=65.00 and WP<=75.00 and o:WP>=20.00 and o:WP<=40.00 and o:p:margin<=-15.0 .. and the result is 53-21-3, a winning percentage ATS for the stronger team of 71.62%!

Of course, that's over 17.5 seasons!

But still it appears to illustrate that a weak team blowing someone out doesn't have as strong an affect on that team as getting blown out .. and that's one of the hypotheses that can be made from using this query.

Interesting though, in 1995 this scenario that has a 71.62 winning percentage over 17.5 seasons was 1-5-1 .. so if someone had thought of it then and had only that season's worth of data to wok with, they might have discarded the hypothesis.

To continue by year:

1996 8-1-1, 1997 6-1-0, 1998 0-0-0, 1999 3-1-0, 2000 2-1-0, 2001 4-1-0, 2002 3-0-0, 2003 5-2-0, 2004 2-0-0, 2005 2-0-0, 2006 0-1-0, 2007 6-2-0, 2008 4-0-0, 2009 3-2-0, 2010 3-2-0, and 2011 (this season to-date) 1-2-1.

So, aside from 1995, this system was pretty good .. until 2009 when it's been 7-6-1 to-date.

It all depends on the makeup of the teams in a given year and the luck of the schedule.

Though this simply may reflect a bigger wave length in the quantum luck, it may also mean there's something to this.

With a little more analysis using this query, one might be able to determine characteristics that caused 1995 and the most recent three years to be so comparatively anomalous, exclude them, and thus maybe achieve a win percentage greater than 80.00 percent.

This is just one example.

Someone with a sharper knowledge of NBA psychology than I possess might be able to develop some more workable hypotheses that consistently hit around 58 - 63 percent with about a hundred plays/year.

That would be a great goal for consistent winning.

Indeed, those professional handicappers who present picks for sale touting a particular pick as being based on a past scenario with a 75% winning percentage may simply have presented a ton of queries to this database and found a few that are truly predictive.

By the way, this site has database info for the NFL and MLB too.  

Another interesting analysis is how teams do when a key player is missing.

Again, the site is: https://killersports.com/nba.py/query?sid=guest&text=+&=1&submit=query 

We'll start with Steve Nash.  Nash has been with the Suns since the 2004 season.  So how have the Suns done ATS at home since then when Nash was in the lineup?

The query is: season >= 2004 and site = home and team = Suns and Suns:Steve Nash:minutes>0.1 .. 157-158-7 with an avg line of -7.1 .

Changing the home parm to away .. 170-135-10 with avg line of -0.6, over 55%, with Nash being a minor money-making difference away when he's present.

But what if Steve's not in the lineup?

Simply change the minutes parm to "<" ..

.. So at home .. 5-6-0 with an avg line of -2.1, and away .. 5-14-0  with an avg line of 4.6.  When the Suns are away without Nash, bet against them ATS, as it's a 73.68% back-tested chance of a win!

How about the same regarding Superman.

Howards been with Orlando since 2004, and during that time: in lineup at home: 168-153-11 -6.0, in lineup away: 166-149-10 0.7, out of lineup at home: 1-1-0 1.8, out of lineup away: 2-2-0 1.6.

No wonder they call him Superman -- he hasn't missed many games.

How about God when he was with Cleveland.

Lebron was with Cleveland from 2003 thru 2009, and the results: in lineup at home: 165-139-3 -6.3, in lineup away: 159-147-5 0.7, out of lineup at home: 4-2-0 1.2, out of lineup away: 3-4-0 5.3.

You can also check for points scored, so if you wanted to know how the Lakers did ATS in their next game after Kobe laid 40 or more on someone, the query is: Lakers:Kobe Bryant:p:points>=40 and the result: 28-24-1 1.5.

Specs for constructing queries is found here: https://killersports.com/NBA/PDF/query_manual.pdf 

I'm interested to know if the rest of you come up with anything significant here. 

Here are some more interesting results.

Since 1995 when a team has a winning percentage of .600 or greater and they're playing at home against a team with a winning percentage of .500 or less and the day before the home team was blown out by 15.0 or more points on the road .. the home team roars back to go 30-1 SU but only 18-12-1 ATS.  That's only happened once this year so far, with CHI taking it out on PHO both SU and ATS.

Regarding teams with at least one win and one loss, it's been said that a good team (winning percentage of .600 or better) playing on the road has a tendency to look past a weaker team (winning percentage of .500 or less) if the very next day the good team has to play again and against another good team.  But since 1995 the good team covers the spread against the weaker team in that scenario 97-66-5, 59.51 percent of the time. .. But, if instead of 0 days rest the good team has a couple days of rest to think about their next good opponent, they look past the weak team 16-27-1 ATS!  Both are contrary to general expectations.  Part of that, though, could be the fault of a very sharp sheduler.

This year there will be 29 times when a team is playing its fifth game in six days.  So far 16 of those situations have been played, and the talk was to bet against those teams ATS as they'd be really tired and the Books wouldn't sufficiently compensate.  But just the opposite has occured so far, as the dead tired team has won ATS 12-4!  Apparently the Books over-compensated!  But who would have guess it?!

Anyway, read through this publication: https://killersports.com/nbalog.pdf

It contains the site's database analysis for this NBA season, which includes 10 perfect scenarios for each of the 30 teams.