NFL - Situational Plays - SDQL

[Mako-SBR]

Here's one from Dr. M for today, but he has since taken it down from his site:

Thoughts?

A and rest>4 and -3<=line<=3 and NDIV and p:L and WP=50 and NB and season>=2004

Queries like these are why Dr. M has rarely (if ever) managed to end a full season in any sport with a win rate of 52.4% ATS or higher in years, despite his mastery of SDQL.

Too many of his have no logical or easily explained reason why the scenario performs...as an example in this one just because a team has won "exactly" 50% of it's games and then has a line between -3/+3, doesn't mean anything in terms of predicting the future performance of said team).

It's just random, which is why the 22-0 ATS result isn't "real".

And it's likely why it didn't win today and recorded it's first loss in a decade...because it likely wasn't ever real in the first place.

SDQL handicapping is tough because it can only provide the data, it can't provide the reasons why a particular situation features an advantage for one team over another or versus the total.

It's up to us to figure those reasons out for ourselves, and then decide whether the previous performance was simply random occurrence, or if there is actual causation between the scenario and the outcome that led to the record.

Thanks for posting though emceeaye, keep it up, it always furthers the discussion and knowledge!

[chopperocker]

10 penalties for 80 yards hurt the Bears. that aside looked like the Bears were the right side.

[Alex Vaile]

Also from what I saw the query of the day was Bears panthers over. And that hit. Crazy enough guys I hit a 6 team parlay yesterday.
Bears over
Phi over
Browns ML
Bucs +11
Texans +5
Steelers -6.5
10 for 446. Nailbiter on Browns and Houston games.

[nash13]

Also from what I saw the query of the day was Bears panthers over. And that hit. Crazy enough guys I hit a 6 team parlay yesterday.
Bears over
Phi over
Browns ML
Bucs +11
Texans +5
Steelers -6.5
10 for 446. Nailbiter on Browns and Houston games.

next time share it with us, before the games start congrats

[KeepItCummin]

Also from what I saw the query of the day was Bears panthers over. And that hit. Crazy enough guys I hit a 6 team parlay yesterday.
Bears over
Phi over
Browns ML
Bucs +11
Texans +5
Steelers -6.5
10 for 446. Nailbiter on Browns and Houston games.

Heck of a day for u Alex, congrats buddy!

[chopperocker]

Also from what I saw the query of the day was Bears panthers over. And that hit. Crazy enough guys I hit a 6 team parlay yesterday.
Bears over
Phi over
Browns ML
Bucs +11
Texans +5
Steelers -6.5
10 for 446. Nailbiter on Browns and Houston games.

Nice! that's ROI

[chopperocker]

8 < t:rest and week<=6 and (2014,1)<=(season,week)

11-0-1 Over

AF and week<=6 and (2014,1)<=(season,week)

14-4 Over

likely the above 2 come back to earth at some point, but worth noting.


(2004,1) <= (season,week) and p:margin>=21 and op:margin<=-21 and A and playoffs=0

6-19 ATS

[KeepItCummin]

Nice! that's ROI

Chopper, what's ROI?

[chopperocker]

Chopper, what's ROI?

Return On Investment

[Alex Vaile]

Heck of a day for u Alex, congrats buddy!

Thanks guys. Felt good to hit 1 for once!

[emceeaye]

Queries like these are why Dr. M has rarely (if ever) managed to end a full season in any sport with a win rate of 52.4% ATS or higher in years, despite his mastery of SDQL.

Too many of his have no logical or easily explained reason why the scenario performs...as an example in this one just because a team has won "exactly" 50% of it's games and then has a line between -3/+3, doesn't mean anything in terms of predicting the future performance of said team).

It's just random, which is why the 22-0 ATS result isn't "real".

And it's likely why it didn't win today and recorded it's first loss in a decade...because it likely wasn't ever real in the first place.

SDQL handicapping is tough because it can only provide the data, it can't provide the reasons why a particular situation features an advantage for one team over another or versus the total.

It's up to us to figure those reasons out for ourselves, and then decide whether the previous performance was simply random occurrence, or if there is actual causation between the scenario and the outcome that led to the record.

Thanks for posting though emceeaye, keep it up, it always furthers the discussion and knowledge!

Thank you, Mako! I appreciate your point, and its precisely that challenge that I've been struggling with since working with SDQL, namely differentiating random from reliable query results. When you have query results with a small sample size, like for Dr. M's above query, this issue becomes even more pronounced. What makes it difficult is when you're working with a query where the rationale is not immediately obvious yet still logical and intuitive once you really sit down and think about it for awhile. Honestly, sometimes it is very difficult to figure out the logic. And sometimes, even when the sample size is large, the logic does not seem compelling or intuitive--It is situations like these that make it especially difficult to know whether to go with a query or not. Why would I go with a query whose rationale I dont understand or buy into? Well, if the sample size is large enough and the results are significant enough, then it really shouldnt matter if I understand it, right? My point is that I think we would all benefit from evaluating the logic/terms of queries in here, especially if the sample size is small, to make sure we aren't just making wagers based on random results. Although it may take some extra time, and the reading dry, I think trying to lay out the logic of particularly challenging queries here in this (or another) thread is invaluable to improve our success wagering on them!

[Alex Vaile]

Today Dr.M has Seahawks as the play as Skins are 0 and 17 against teams since

team=Redskins and H and oS (PFD)/oS (FD)<0.575 and date>=20061201

[Mako-SBR]

Thank you, Mako! I appreciate your point, and its precisely that challenge that I've been struggling with since working with SDQL, namely differentiating random from reliable query results. When you have query results with a small sample size, like for Dr. M's above query, this issue becomes even more pronounced. What makes it difficult is when you're working with a query where the rationale is not immediately obvious yet still logical and intuitive once you really sit down and think about it for awhile. Honestly, sometimes it is very difficult to figure out the logic. And sometimes, even when the sample size is large, the logic does not seem compelling or intuitive--It is situations like these that make it especially difficult to know whether to go with a query or not. Why would I go with a query whose rationale I dont understand or buy into? Well, if the sample size is large enough and the results are significant enough, then it really shouldnt matter if I understand it, right? My point is that I think we would all benefit from evaluating the logic/terms of queries in here, especially if the sample size is small, to make sure we aren't just making wagers based on random results. Although it may take some extra time, and the reading dry, I think trying to lay out the logic of particularly challenging queries here in this (or another) thread is invaluable to improve our success wagering on them!

Yes, totally emcee, agree. It can be really tough to cut through the static to find the actual reason why a situation is producing a predictable result, especially if it's not obvious. And even then, sometimes the sample size is large enough that you just don't care and throw caution to the wind as you mentioned, simply because the numbers are so favorable heheh.

[Mako-SBR]

(2004,1) <= (season,week) and p:margin>=21 and op:margin<=-21 and A and playoffs=0

6-19 ATS

Very nice chopper, essentially that query is identifying +EV from artificially-impacted spreads after blowout wins/losses the previous week...as we just saw happen with the Patriots and Bengals (media/public crushed the Patriots all week while hilariously pumping the Bengals as a "Super Bowl team" lol).

I want to see the first person who uses it to bet on the Jets and Raiders though next week, raise your hand and let me salute your bravery haha.

[JMon]

Thank you, Mako! I appreciate your point, and its precisely that challenge that I've been struggling with since working with SDQL, namely differentiating random from reliable query results. When you have query results with a small sample size, like for Dr. M's above query, this issue becomes even more pronounced. What makes it difficult is when you're working with a query where the rationale is not immediately obvious yet still logical and intuitive once you really sit down and think about it for awhile. Honestly, sometimes it is very difficult to figure out the logic. And sometimes, even when the sample size is large, the logic does not seem compelling or intuitive--It is situations like these that make it especially difficult to know whether to go with a query or not. Why would I go with a query whose rationale I dont understand or buy into? Well, if the sample size is large enough and the results are significant enough, then it really shouldnt matter if I understand it, right? My point is that I think we would all benefit from evaluating the logic/terms of queries in here, especially if the sample size is small, to make sure we aren't just making wagers based on random results. Although it may take some extra time, and the reading dry, I think trying to lay out the logic of particularly challenging queries here in this (or another) thread is invaluable to improve our success wagering on them!

I read somewhere and (a point I believe in) is things trend with out the need of reason/logic behind it, especially when humans are involved. There is no reason why one can not make money of it. The problem is finding the trend in time and knowing when to stop betting it before regression. Much like the stock market; get in and get out but at the same time make sure you make money of it.

[Mako-SBR]

I read somewhere and (a point I believe in) is things trend with out the need of reason/logic behind it, especially when humans are involved. There is no reason why one can not make money of it. The problem is finding the trend in time and knowing when to stop betting it before regression. Much like the stock market; get in and get out but at the same time make sure you make money of it.

Very true, your post reminds me of the popular story of Bob Voulgaris, who developed a nice point-prediction model for the NBA using quantitative metrics, programmed for him by a few sharp engineers he had hired away from Wall St.

The model saw massive success for him in it's first year or two of use but then started to inexplicably fail despite his best efforts, to the point of uselessness. Impossible to know "why" it failed, even though he and his team were doing the exact same things to create success that they had done before, but your comment about timing and regression are as good a pair of culprits as any.

[chopperocker]

Very nice chopper, essentially that query is identifying +EV from artificially-impacted spreads after blowout wins/losses the previous week...as we just saw happen with the Patriots and Bengals (media/public crushed the Patriots all week while hilariously pumping the Bengals as a "Super Bowl team" lol).

I want to see the first person who uses it to bet on the Jets and Raiders though next week, raise your hand and let me salute your bravery haha.

agreed. interestingly when the win by margin team is off a BYE -

tp:week+2=t:week and (2004,1) <= (season,week) and p:margin>=21 and op:margin<=-21 and A and playoffs=0

0-4 SUATS

these would tell me some reasons why Bengals may be a bad play, but betting Patriots would be my own responsibility. I passed on that game, but enjoyed the research.

[nash13]

Is there a need for NHL Situational Plays? I have a few, but exchanging some of the would be useful.

[chopperocker]

Is there a need for NHL Situational Plays? I have a few, but exchanging some of the would be useful.

there is a need for all things my brother

[nash13]

Ok if it is ok, i will start a thread over there.

[nash13]

http://www.sportsbookreview.com/forum/hockey-betting/2972151-nhl-situational-plays.html

[JMon]

Is there a need for NHL Situational Plays? I have a few, but exchanging some of the would be useful.

I'm not in to hockey, nor do I understand it at all. But I do have one...I'll share over in your new thread. BOL bud. Make sure you keep it civil over there!

[b1slickguy]

(team=Cowboys or team=Eagles) and F and p:F and season>=2011

Good luck.

[Timmay]

What's SQDL?
I only read a couple comments and I'm lost.

What's this all about?

[emceeaye]

Seahawks ATS

Here is Dr. M's query of the day:

team = Seahawks and p:RY - tA(p:RY) >= 50 and season >= 2011 and p:W and NB

Thoughts?

[chopperocker]

Here is Dr. M's query of the day:

team = Seahawks and p:RY - tA(p:RY) >= 50 and season >= 2011 p:W and NB

Thoughts?

all information is great but maybe incorporating next opponents rushing yards allowed would be more thorough? much appreciation to mr. meyer for making his database accessible and making videos to demonstrate how to utilize it.

[neilb1973]

I can't get this query to work at all at KS. Just hangs. What are the results?

[emceeaye]

I can't get this query to work at all at KS. Just hangs. What are the results?

Add "and" after 2011

[nash13]

I have a question: I have two colliding trends. Is there a possibility to combine these in a query?
It is from the NHL general Trends.

Query 1:H and F and p:A and p:F and p:W and date>=20111022
Query 2:A and p:F and p:L and date>=20131012

On query 1 the play is on the Canucks.
On query 2 the play is on the Oilers.

Is there a possibility to code something like this:
"Play on the Home Fav and their previous was a Away Fav Win and their op previous game was NOT a Fav Loss" that is for Q1 without Q2
and Q2 without the plays from Q1 may be something like this:
"Play the away Team when they previous Loss was as a Fav and their opp had NOT Away Fav Win previously"

[emceeaye]

I have a question: I have two colliding trends. Is there a possibility to combine these in a query?
It is from the NHL general Trends.

Query 1:H and F and p:A and p:F and p:W and date>=20111022
Query 2:A and p:F and p:L and date>=20131012

On query 1 the play is on the Canucks.
On query 2 the play is on the Oilers.

Is there a possibility to code something like this:
"Play on the Home Fav and their previous was a Away Fav Win and their op previous game was NOT a Fav Loss" that is for Q1 without Q2
and Q2 without the plays from Q1 may be something like this:
"Play the away Team when they previous Loss was as a Fav and their opp had NOT Away Fav Win previously"

Unclear. You're saying you want a combination of the two queries but then negating the terms of the two queries when you explain it out.

I think this is a combination of the two:

H and F and p:A and p:F and p:W and date>=20111022 and o:A and op:F and op:L

Maybe I'm misunderstanding...

[nash13]

i wanted to know in which case it is better to take query 1 or query 2.
HF and p:AFW and date>=20111022 and (op or op:FW)
points out that it is safer long term to take the HF.

[Alex Vaile]

Dr.M.

Play Seattle -8.5 -110

team=Cowboys and p:F and p: passes - tA (p: passes)> 8.75 and NB and date>=19931125

Cowboys 0 and 27 ats in this spot since 93

[Alex Vaile]

Second play: ( Dr.M )

Play over 45.5 -110 in Washington vs Arizona game

team=Arizona and p:FDP <25 and p:A and rest<7 and date>=2004

19 and 0 over since 2004

[mdunlap3]

i wanted to know in which case it is better to take query 1 or query 2.
HF and p:AFW and date>=20111022 and (op or op:FW)
points out that it is safer long term to take the HF.

Makes sense.

[emceeaye]

Dolphins ATS

Dr. M's NFL query of the day today:

day=Sunday and REG and AF and p:H and p:line<=-7 and p:margin>=7 and tA(passes)-passes>=10 and p:PY<200