[hyahya]

I like to look at the specific spot, pick out 3 or 4 factors I deem most relevant to input into the SDQL search. From there I attemmpt to fine tune it. I try to avoid the pitfalls of data mining bias this way.

For the above query, I felt first game back from a 4+ game road trip with the last game being a loss against a AF coming off a home win were the most relevant factors. Could've also input conference!=o:conference as that accounts for a portion of the cross continent travel given it's the first game of a western road trip for the Wiz.

[cofaga]

p:HFL and tA(o: points)<100 and po: points>110 and op:W and opp:W and oppp:W and opppp:W

A decent defensive team after allowing more then 110 points at home in a loss as the favourite against a team on a 4 win streak

[tonywayne]

p:HFL and tA(o: points)<100 and po: points>110 and op:W and opp:W and oppp:W and opppp:W

A decent defensive team after allowing more then 110 points at home in a loss as the favourite against a team on a 4 win streak

There's a cleaner way to include that streak in this case:

o:streak = 4

Saves you the trouble of putting all those oppppppp entries.

[cofaga]

There's a cleaner way to include that streak in this case:

o:streak = 4

Saves you the trouble of putting all those oppppppp entries.

Cheers

[19th Hole]

Does anyone consider the the SDQL DAILY NBA TIPSHEET plays?

[JAnthony]

tA(points)>=100 and oA(o: points)>=101 and WP>=60 and tS(L, N=5)>=4 and game number>=30

Total oriented trend, when a playoff team has a chance of getting getting back on the track against a poor defensive team.

[JMon]

Does anyone consider the the SDQL DAILY NBA TIPSHEET plays?

I'm not much of a "team specific" user of sdql. However, I do review them and they also a great way to learn.

[Mako-SBR]

I'm not much of a "team specific" user of sdql. However, I do review them and they also a great way to learn.

Agree, not much long term value with them for the most part as they're just too volatile.

They're actually a big part of the reason that the few SDQL-based pay touts out there have rarely exceeded 50% for long before crashing back to Earth over the course of a long season.

For a wide variety of reasons the majority of team/player trends break and expire at a much faster rate than the more generic league-wide trends...assuming the league-wide trend was based on logic and has a large number of games to analyze.

Excellent practice though within the db for sure, like that angle of the "daily" broadcasts.

[Mako-SBR]

tA(points)>=100 and oA(o: points)>=101 and WP>=60 and tS(L, N=5)>=4 and game number>=30

Total oriented trend, when a playoff team has a chance of getting getting back on the track against a poor defensive team.

That's not bad at all, good foundation to tweak and experiment off of.

It's basically taking advantage of the books having to walk the total down with each loss as the playoff team runs through that negative streak, in order to keep the square bettor engaged and willing to keep throwing money at the over/fav.

Might possibly work the other way as well (to the Under) by reversing the scenario and playing the under whenever a strong playoff-headed team reels off 4 wins in 5 games and crushes the Over, will have to play with it.

Either way it's a good example of a base query/scenario that works which can then be expanded on, which everyone should be doing with each query we post just to get good at seeing the whole picture.

[Mako-SBR]

There's a cleaner way to include that streak in this case:

o:streak = 4

Saves you the trouble of putting all those oppppppp entries.

Ha, never knew that, nice T.

[tonywayne]

Ha, never knew that, nice T.

Thanks. If you need a losing streak, use an negative number. And of course > and < work with it too. I've been using "streak" and "ats streak" routinely to look for trends.

[tonywayne]

What about:

HD and p:ADW and season >= 2009 and conference = o:conference = po:conference = Eastern

The road faves in this trend are winning by over 6.5 points on average (and covering nearly 64% of the time). That would seem to bode well for tonight's matchup with a pretty low spread.

Thoughts?

[dmitean]

Agree, not much long term value with them for the most part as they're just too volatile.

They're actually a big part of the reason that the few SDQL-based pay touts out there have rarely exceeded 50% for long before crashing back to Earth over the course of a long season.

For a wide variety of reasons the majority of team/player trends break and expire at a much faster rate than the more generic league-wide trends...assuming the league-wide trend was based on logic and has a large number of games to analyze.

Excellent practice though within the db for sure, like that angle of the "daily" broadcasts.

I actually prefer teams based trends than general ones.
I'ts obvious that bookies (and public) react quicker to teams based trends, because they are easier to notice, but they are usually much more accurate.

Like Boston playing Under second game B2B few years ago. Or betting Over on Bucks games, when Bogut sat out. Playing Over in Thunder games, the first season with Perkins was easy money (still is btw).
Thunder off a loss angle and so on...

Such trends usually come with a good reason and not just statistical coincidence.

General trends are much harder to understand, are we looking at statistical abnormality that should correct itself or there is some logic behind it.

[dmitean]

What about:

HD and p:ADW and season >= 2009 and conference = o:conference = po:conference = Eastern

The road faves in this trend are winning by over 6.5 points on average (and covering nearly 64% of the time). That would seem to bode well for tonight's matchup with a pretty low spread.

Thoughts?

Nice one, especially since it's 4 - 11 last two seasons and 1 - 5 last six times.

[JMon]

tA(points)>=100 and oA(o: points)>=101 and WP>=60 and tS(L, N=5)>=4 and game number>=30

Total oriented trend, when a playoff team has a chance of getting getting back on the track against a poor defensive team.

like it

[JMon]

There's a cleaner way to include that streak in this case:

o:streak = 4

Saves you the trouble of putting all those oppppppp entries.

the way I prefer

op:W and op2:W and op3:W and op4:W

[JMon]

Like this one quite a bit:

HD and season>2002 and p:AL and pp:A and ppp:A and pppp:A and ppppp:A and op:W

did you check out the total difference btwn conferences!

[JMon]

Here's another:

HD and season>2009 and p: points<100 and pp: points<100 and tA(points)>100

Probably a little too general. Changing to:

HD and season>2009 and p: points<100 and pp: points<100 and tA(points)>100 and p:W

More descriptive of the actual situation but drops our sample size significantly and shows less consistency over time as it's 2-9 since 2009 but only 8-17 since 2002 (6-8 between 2002-2009)

what about turning it into a small home dog and dropping the season filter

1.5<=line<=3

[Mako-SBR]

I actually prefer teams based trends than general ones. I'ts obvious that bookies (and public) react quicker to teams based trends, but they are usually much more accurate.

I'm sure you do, the public eats these up as being 'real', which is why touts use them so frequently on their advertising blogs.

And I can definitely see why they're appealing, no question. But the actual results from thousands of previous team/player-trend plays from dozens of logged handicappers over the past decade's worth of NBA seasons have proven them to be tough to win with the vast majority of the time.

The type that are like this example (not a real one, made up for the post) are nearly impossible to quantify as being anything other than a fluke, no matter how strong they may appear in the moment: "The Spurs are 14-2 when Tony Parker has a shooting % of 50% or higher in his previous game and 5+ assists".

Or something like: "The Spurs are 11-1 when coming off 2 days or more rest after winning on the road within their division if they're playing their 4th game in 7 nights".

I do like team trends in one way, which is as a tertiary piece of information when weighing other more important metrics/scenarios. But I won't make or cancel a bet I'm already leaning towards or against based on whether a team or player trend is pointing the opposite way.

But all of us are always open to new angles though, if you find some that offer more than 50+ game sample sizes we'd definitely like to look at them!

What about:

HD and p:ADW and season >= 2009 and conference = o:conference = po:conference = Eastern

The road faves in this trend are winning by over 6.5 points on average (and covering nearly 64% of the time). That would seem to bode well for tonight's matchup with a pretty low spread.

Thoughts?

That's not bad either, I'd want to examine it a bit further but it looks like a solid one to build off and play with T. I hesitate to break out specific Divisions or Conferences (Eastern) but you never know, the data might point to some sort of hidden travel anomaly or circumstance that is present in a specific region that isn't in others.

Either way it's something to work on, like it.

[JMon]

Nothing for me tonight. Few leans, but nothing strong enough

GL fellas...and great contribution recently

[Mako-SBR]

Nothing for me tonight. Few leans, but nothing strong enough

GL fellas...and great contribution recently

I'm on this one today, but only a half unit as it's a scenario that's overfit and based off of ATS streaks and happens to feature pretty spectacularly shitty teams...all three of which aren't things I like to play on, but it's been winning so why not.

Plus it works for the Under as well as the Spread pick for more action. Can't remember where it came from, looks like a cofaga special though :

AD and (o:rest=0 or rest=1) and ats streak<0 and 0 < o: ats streak < 3 and line > = 12 and season>=2006

[hyahya]

U lost me at "0=12"

[TheLuckFactor]

Hi all,

Both of these trends are admittedly biased and have small sample sizes. Returns conflicting results for the ATS pick, but both favor the over. Probably a no bet for me but figured I'd share to confirm that these trends are not reliable. (Still learning!)


tS(U, N=9)>= 8 and total>=201 and HD


tS(A,N=5)>=5 and HD and rest <=1 and line <4 and total>200

Thanks!

[Mako-SBR]

U lost me at "0=12"

Whoops, typo, edited it and now it's correct:

Edit: Damn, still wrong, something about the copy/paste. One sec

Edit 2: Ok, working once I put spaces into the part of the query the board had trouble with:

AD and (o:rest=0 or rest=1) and ats streak<0 and 0 < o: ats streak < 3 and line > = 12 and season>=2006

For the lurkers or inexperienced: Remove the spaces in the second half of the query above around the red letters/symbols, starting at the "and 0 < o: ats streak" area. First time I've seen that, no idea why it happened. So any red symbol or letter in the query above, remove the space in front or behind it.

Both of these trends are admittedly biased and have small sample sizes. Returns conflicting results for the ATS pick, but both favor the over. Probably a no bet for me but figured I'd share to confirm that these trends are not reliable. (Still learning!)

tS(U, N=9)>= 8 and total>=201 and HD

tS(A,N=5)>=5 and HD and rest <=1 and line <4 and total>200

The second one is interesting to me. I like the concept of a small dog at home getting points with no rest after being away for ages and how that would conflict with the public bettor's mind and thus the corresponding line adjustment (value).

Worth looking into and seeing if we can expand it to a larger sample.

[FortySix]

Sorry Mako, i have been trying this one AD and (o:rest=0 or rest=1) and ats streak<0 and 0 < o: ats streak < 3 and line > = 12 and season>=2006 like you said, taking out the spaces but I keep getting an error reading.

[Mako-SBR]

Sorry Mako, i have been trying this one AD and (o:rest=0 or rest=1) and ats streak<0 and 0 < o: ats streak < 3 and line > = 12 and season>=2006 like you said, taking out the spaces but I keep getting an error reading.

It works, just tested it, but you have to remove ONLY the spaces that were artificially entered. NOT the spaces that every single SDQL query always has. Use your head in other words and leave the spaces that have to be there of course.

SO: take out the spaces before and after the "<" symbols, take out the spaces around the "=" sign, take out the space after the "o:", you know, typical SDQL language typed the proper way.

LEAVE the spaces around the logical areas where typically SDQL queries always have spaces (around the "and" obviously, in between parameters, etc). It's just a few spaces, no other way to describe it.

Have never seen SBR's software do this before, no clue why (this is not related to the smiley spaces glitch we've already worked around for those lurking).

[Mako-SBR]

Testing the previous query to identify a board post workaround, ignore:

AD and (o:rest=0 or rest=1) and ats streak<0 and 0=12 and season>=2006

[Mako-SBR]

Test 2:

AD and (o:rest=0 or rest=1) and ats streak<0 and 0=12 and season>=2006

[Mako-SBR]

Test 3:

AD and (o:rest=0 or rest=1) and ats streak<0 and 0=12 and season>=2006

[Mako-SBR]

Yeah no idea why, just tried to Go Advanced and disable the automatic HTML parsing the board does (along with smileys) for posting but none of it worked, all mangled the query the same way.

It doesn't like the way we define the ats streak parameter for SDQL for some reason, that particular "and < o: ats streak < 3" phrase (but without the spaces around the symbols and colon obviously).

Luckily we don't use it much but what a pain in the ass when we do.

Testing again, first with the spaces: < o:ats streak<3 and line>

Now without:

Ok, the specific error the board has lies in the < symbol when followed without a space by the o: symbol. That screwed up the whole query in the board, so here it is with just that one single space added (you must remove it to cut and past into SDQL):

AD and (o:rest=0 or rest=1) and ats streak<0 and 0< o:ats streak<3 and line>=12 and season>=2006

There we go. So to get the board to post it right I had to add a space after the red < symbol above. Take out that space and the query works in SDQL. It only seems to occur with that specific issue, a < or > symbol directly followed by the o:

[parlayin]

Here's a nice one I've been using that is active on the Bulls/Heat game today (bet the over).

Premise behind it is that you're catching a team that usually doesn't have strong offense after a game in which they established a rhythm and scored far more points than was expected...and that the higher ouput will then continue the next game to push it over the predicted total:

AD and rest<2 and p:HW and p:dps>=15 and p:margin>=10 and season>=2006

Sorry I'm late, this was from page 3 but just a few questions as I'm still learning.

1. How are you capturing the "usually doesn't have strong offense" part?

2. If the basis is that a team's unusually high output will continue the next game, do the query results hold up in normal B2B situations, or what's the significance of using away dogs coming off home wins? I know that was the situation that given day with the Bulls and Heat but I'm wondering if the theory is supported more generally. Does home/away or previous win/loss matter? Isn't the p:dps the main factor?

[JMon]

Geez, thanks for the heads up mako

[Mako-SBR]

Sorry I'm late, this was from page 3 but just a few questions as I'm still learning.

1. How are you capturing the "usually doesn't have strong offense" part?

2. If the basis is that a team's unusually high output will continue the next game, do the query results hold up in normal B2B situations, or what's the significance of using away dogs coming off home wins? I know that was the situation that given day with the Bulls and Heat but I'm wondering if the theory is supported more generally. Does home/away or previous win/loss matter? Isn't the p:dps the main factor?

Good Qs!

For #1, it's the dps result in that previous game being so much higher than was expected from the team (see the KillerSports dot com NBA SDQL Manual .pdf for more information on dps, or for the lurkers trying to learn, any info about SDQL queries. I'd link it but SBR doesn't allow external links so Google it).

For #2, momentum is the general answer, but all of the various permutations you're suggesting can be explored in SDQL itself with a little effort if you're interested. Run a variety of scenarios and report your results back to us, you may improve that query considerably.

Geez, thanks for the heads up mako

Tell me about it, I want that half hour of my life back...

[cofaga]

p:H and P:margin<=-15 and AD and 60>o:WP>50 and conference=o:conference and (rest>0 and o:rest>0) and game number>=42

Seems like a solid play to me, lemme know what you guys think (on the over play)

[JAnthony]

I like the chances of a team being a bit underrated by bookies, because of recent poor performance:

p:ats margin<0 and 10<=p:line<=13 and 11<=line<=13

Principle mostly the same:

p:L and pp:L and -1<=line<=1 and rest=0 and o:WP<43 and A