[Cutler'sThumb]

In general, yes, I definitely consider the time frame. That's why I'm compiling results for the last 3 seasons, as it allows me to see if a trend that has been good historically but is flat or down in recently.
This morning, however, I just had an "off with their heads" approach. If this trend of small trends hurting results continues, I'm sure I will have to go back thru and look at some of them more carefully. Overfitted and small is probably where the real problem lies. It's a little crude, but whenever I see a small sample and a trend code that can't fit in the SDQL query box I start looking at it more critically.

[dmitean]

Why this query doesn't show last night's game between Memphis and Hawks?
Hawks won last game as home dog and faced Memphis as road dog.

WP>60 and p:HWD and AD and season>2012 and playoffs=0

[b1slickguy]

Why this query doesn't show last night's game between Memphis and Hawks?
Hawks won last game as home dog and faced Memphis as road dog.

WP>60 and p:HWD and AD and season>2012 and playoffs=0

The data base has Warriors vs Hawks as a pick with line at 0.
Good luck.

[Cutler'sThumb]

Nice work Cutler. One thing I have come to believe about the sample size is that it might be good to not consider it by itself, but as a function of time. For example, two queries could have the same sample size of 200, but one goes back 20 years to get that 200, while the other goes back 2 years to get the 200. So the sample sizes are not really the same. Are you factoring that into deciding which queries to eliminate?

Another point to be made here is that the NHL database only tracks back to 2006, so essentially every query is at minimum "season>=2006", while the NBA tracks clear back to '95. Gotta be on the look out for trends that made all their profit 10+ years ago. Favor larger, more recent trends and I think the results will improve.
Obviously there are guys who are much more adept than me at jumping in and out of trends to find what is "hot" this season, but I think we have a lot of good trends to leverage so I'm trying to figure out the best way to systematically use what we have to generate consistent results using the volume model.

[Heart]

I'm wondering too if lines makers have become aware of certain sdql trends this year and have adjusted accordingly.

[pip2]

I'm wondering too if lines makers have become aware of certain sdql trends this year and have adjusted accordingly.

That would be kind of awesome! In that case fading the quries might actually work until the lines makers adjust again!

[pip2]

Another point to be made here is that the NHL database only tracks back to 2006, so essentially every query is at minimum "season>=2006", while the NBA tracks clear back to '95. Gotta be on the look out for trends that made all their profit 10+ years ago. Favor larger, more recent trends and I think the results will improve.
Obviously there are guys who are much more adept than me at jumping in and out of trends to find what is "hot" this season, but I think we have a lot of good trends to leverage so I'm trying to figure out the best way to systematically use what we have to generate consistent results using the volume model.

Another aspect of this is I still remember how well the spitball queries did. Part of that might have been lucky timing, but the spitball queries, up until the Nash sheet got made, were like 85% composed of Nash-jmon-mako-hiyahya queries, because those were the ones that I tended to pick out from the earlier pages of this thread. Maybe I should go back and try to count results of mako-jmon-nash-hiyahya queries...

[JMon]

Geez, I haven't read this thread in awhile so excuse my ignorance. Are you guys auto playing every single query? Or just tracking? And if you are tracking, what are you tracking and why? Just because a query is active doesn't mean it should be played. I constantly add, delete, edit queries everyday and I have 1000s. A legitimate query that is active for the day still needs to be dissected each and every time- even one's that have hit 75% for the duration and the current year.

[nash13]

Geez, I haven't read this thread in awhile so excuse my ignorance. Are you guys auto playing every single query? Or just tracking? And if you are tracking, what are you tracking and why? Just because a query is active doesn't mean it should be played. I constantly add, delete, edit queries everyday and I have 1000s. A legitimate query that is active for the day still needs to be dissected each and every time- even one's that have hit 75% for the duration and the current year.

Nothing to excuse here, I think the whole group reached the hardest point of evaluation, when to take a play and on which standards. This is harder than finding queries alone.

[JMon]

Nothing to excuse here, I think the whole group reached the hardest point of evaluation, when to take a play and on which standards. This is harder than finding queries alone.

So it's my understanding some are back testing every single query as a whole- to see how they perform if one played every query? Is that right? Or am I misunderstanding what's going on? If that is the case, that is foolish and a complete waste of time. Not everything is black or white, nor can be statistically proven or hold logic. Logic..give me a fukkin break..square fukks. You are fool if you think otherwise. Some speak of small samples....I will take on anyone here in a contest using small samples/current year queries to those that claim bigger samples are better.

Sdql comes with experience and learning the hard way. Not scrapping, not copying and pasting, not using someone else's thoughts/queries and trying to make money off it. It's about learning the hard way through trial and error. There are around 4-6 contributors to this thread that should even have a say. The rest are piggy backers and you know who are are.

[pip2]

So it's my understanding some are back testing every single query as a whole- to see how they perform if one played every query? Is that right? Or am I misunderstanding what's going on? If that is the case, that is foolish and a complete waste of time. Not everything is black or white, nor can be statistically proven or hold logic. Logic..give me a fukkin break..square fukks. You are fool if you think otherwise. Some speak of small samples....I will take on anyone here in a contest using small samples/current year queries to those that claim bigger samples are better.

Sdql comes with experience and learning the hard way. Not scrapping, not copying and pasting, not using someone else's thoughts/queries and trying to make money off it. It's about learning the hard way through trial and error. There are around 4-6 contributors to this thread that should even have a say. The rest are piggy backers and you know who are are.

You sound really pissed. But you make top notch queries and probably good money as well, so it's not clear why you would be so pissed off. If people aren't doing it the way you do it, they will probably pay the price in terms of losing money, so they will suffer all you could want. And if they don't lose money, then maybe there is more than 1 way to do it. Who cares, either way?

[JMon]

Not so much pissed off, but the truth, eh?

[pip2]

mil/bkn O 2 units -- +2
atl L 2 units -- -2
mia L 1 unit -1
ut O 1 unit -- +1
ut L 2 units -- -2
ind W 8 units -- +8
phi L 2 units -2
phi U 1 unit -- +1
sa U 1 unit -- +1
okc W 2 units -- +2
okc U 2 units -- -2
___________________
total +6

[nash13]

My Inbox was full, if anyone was trying to write me messages, now I cleared it.
Just a heads up report for the tracking starting 29/01/15.
NHL: 45-30-4
+49 units profit (on variable staking)
28% Yield

NBA:
59-56-1
-29 units profit (on variable staking)
-11% Yield

NCAAB:
81-74-3
+5 units profit (on variable staking)
3% yield

So i think JMon has a point and is spot on with his observations that blindly playing picks is not the way to go, as far as for NBA with its large possibilities. NHL seems to be a different story. My main field of expertise was MLB so far, so I am not sure how my observations form there might transfer to other sports. MLB is working in circles and has monthly awesome working queries. Year in year out. NBA and NFL are the sports i can not figure out yet.

The discussions here going a bit out of hand, which was never my purpose to do this in the first place.

[Cutler'sThumb]

Not so much pissed off, but the truth, eh?

I'll freely admit that I'm nowhere near the expert level of many of the guys on this thread with regards to constructing my own queries. As JMon correctly states, it takes years of hard work to really be good at that, and I only discovered SDQL 3 months ago. The concept of trends and SDQL appealed greatly to the way I approach this game, and I'm sure I'll pick up the whole package in due time. In the meantime, however, I joined the group just as there was a massive amount of info being uploaded, and I've spent a lot of time trying to work out the best possible way to use all that info. JMon, you obviously have a successful system down, so feel free to ignore my efforts. If I crash and burn trying it my way, it'll just be an expensive lesson.

It may well be a waste of time to back test all the trends, but doing it has allowed me to familiarize myself with the language and see how individual trends behave over time, and which ones have flat lined recently and should be put on the bench. The NBA is still a work in progress (still over 1/2 of the trends to work thru), but I've completed this process in the NHL, and have seen great results (thanks Oilers!). I don't play every trend, but rather the trends that fit my filter and are not opposed by a valid trend on the other side. I'm sure this could get even better if I knew each trend inside and out and took the time to cap each trend for each game, but I just don't have the skill (or the time) to do that at this point.

JMon, your smaller trends are absolutely golden, but imho this simply isn't the case with some of the other smallish trends that have really been massaged. I look at it from a statistical standpoint (it was a many years ago, but I lived and breathed the scientific method for a long time, and I did take statistics along the way): the smaller the sample size, the more volatility there is likely to be. This is statistics 101, doesn't matter if we're talking about what % of dogs like to eat their own poop or if the Nuggets are going crap the bed (again). That doesn't mean small trends should be shunned, it just means they should probably be looked at with a little more caution. I actually won the largest bet I've ever made last month based on a trend with a sample size less than 40, but the situation was perfect. I knew about it from before I found SDQL and had actually PMed you asking about a start time filter, as I was trying to come up with the correct query for it. Anyway, sorry if I got your dander up with my posts.

[Heart]

Uh i'm not sure who his hectoring was targeted to. It has all the makings of a drunken rant so hopefully we can chalk it up to that and move on.

I've started filtering my NBA queries by filtering on current Day, Month, Season, rest, and o:rest and over the past week the queries aren't hitting. I only started with a subset of the queries in the spreadsheet since I think a lot of them were "forced" queries and use only the ones that make sense to me why they would hold an advantage. If the books adjusted I don't think they would over adjust to the point where fading the picks would be profitable.

[pip2]

Fading the pip query library:

mem beats bkn 1
cha/det U 1
lal/den U 3
den beats lal 3
phx beats hou 1

[JMon]

Uh i'm not sure who his hectoring was targeted to. It has all the makings of a drunken rant so hopefully we can chalk it up to that and move on.

I've started filtering my NBA queries by filtering on current Day, Month, Season, rest, and o:rest and over the past week the queries aren't hitting. I only started with a subset of the queries in the spreadsheet since I think a lot of them were "forced" queries and use only the ones that make sense to me why they would hold an advantage. If the books adjusted I don't think they would over adjust to the point where fading the picks would be profitable.

No rant, I wasn't targeting anyone that contributes to this thread. Just been getting pms from lurkers I suppose.

[JMon]

I'll freely admit that I'm nowhere near the expert level of many of the guys on this thread with regards to constructing my own queries. As JMon correctly states, it takes years of hard work to really be good at that, and I only discovered SDQL 3 months ago. The concept of trends and SDQL appealed greatly to the way I approach this game, and I'm sure I'll pick up the whole package in due time. In the meantime, however, I joined the group just as there was a massive amount of info being uploaded, and I've spent a lot of time trying to work out the best possible way to use all that info. JMon, you obviously have a successful system down, so feel free to ignore my efforts. If I crash and burn trying it my way, it'll just be an expensive lesson.

It may well be a waste of time to back test all the trends, but doing it has allowed me to familiarize myself with the language and see how individual trends behave over time, and which ones have flat lined recently and should be put on the bench. The NBA is still a work in progress (still over 1/2 of the trends to work thru), but I've completed this process in the NHL, and have seen great results (thanks Oilers!). I don't play every trend, but rather the trends that fit my filter and are not opposed by a valid trend on the other side. I'm sure this could get even better if I knew each trend inside and out and took the time to cap each trend for each game, but I just don't have the skill (or the time) to do that at this point.

JMon, your smaller trends are absolutely golden, but imho this simply isn't the case with some of the other smallish trends that have really been massaged. I look at it from a statistical standpoint (it was a many years ago, but I lived and breathed the scientific method for a long time, and I did take statistics along the way): the smaller the sample size, the more volatility there is likely to be. This is statistics 101, doesn't matter if we're talking about what % of dogs like to eat their own poop or if the Nuggets are going crap the bed (again). That doesn't mean small trends should be shunned, it just means they should probably be looked at with a little more caution. I actually won the largest bet I've ever made last month based on a trend with a sample size less than 40, but the situation was perfect. I knew about it from before I found SDQL and had actually PMed you asking about a start time filter, as I was trying to come up with the correct query for it. Anyway, sorry if I got your dander up with my posts.

With your learning you will find that the majority of sdql (not all) are unstable-always changing. For instance, most of the queries I have posted here, I don't even use or have been personally deleted. Yet as you stated, I can see how backtesting them would be beneficial to the learning process.

[dmitean]

Uh i'm not sure who his hectoring was targeted to. It has all the makings of a drunken rant so hopefully we can chalk it up to that and move on.

I've started filtering my NBA queries by filtering on current Day, Month, Season, rest, and o:rest and over the past week the queries aren't hitting. I only started with a subset of the queries in the spreadsheet since I think a lot of them were "forced" queries and use only the ones that make sense to me why they would hold an advantage. If the books adjusted I don't think they would over adjust to the point where fading the picks would be profitable.

I believe that you can find a query that will fit any bet, if you look long enough.
Under, Over, backing the home team, fading it - all in one game.

I also believe that any query can be found unfit - if you spend enough time testing it.

I'm not an expert in SDQL by any means, but I believe that you shouldn't blindly bet on a query, but I also don't believe that you should dissect the life out of it.
If you look at month, season, rest, o:rest, conference, wp - you will find some spot where the query doesn't work well.

I can tell you how I approach queries.

ATS queries - month and season and home/ away & fav/dog if the query didn't specify it.
I don't look at the query, but on the teams in the game - how they play with such rest and how they perform on this day of the week either this season or take it one season back if the results are too small sample (0 results or 1 result).
I usually play with the range of line given. If for example line is -5, I will set the range on -2.5 to -7.5 to see how query performs.
Then I look at the game for special angles - revenge, look ahead the next game stuff like that - if it's possible, I ran query to see how it did in this special angle.

Totals:
I don't care about D or F. I'm much more interested in month and season.
The rest is more or less the same.
I can play with the totals but casting a larger net. If the total is 205, I will set the range 200 - 210 to see how it responds.

One more thing I like to look in to, when I'm looking at season results - how the teams in question played in this query.
For example, big samples can have 30 results in 2014 season or more and I can see that query suggests to play Team A here and the query has 60% hitting range and all tests are good.
But, if Team A, fit this query three times this season before and lost ATS all three - I won't play it.

Just an example - the game between Lakers and Nuggets. Three queries suggest we play the Over. But, both teams in a huge drop scoring wise, drop that the queries don't take in to account.
They also don't take in to account the fact that Nuggets on the road have much much lower scores than at home.
If you play with number for Nuggets and with the Lakers, I think you won't play the Over here.
Can the Over hit? For sure. But I think that it would be smart not to touch here.

Two queries point to the Rockets tonight and this is much harder case.
Suns lost to Rockets this season already at home, also a play that had support from the queries.
Suns also off a loss to the Kings and this is the last game before ASG break, while Rockets will play one more game.
But, both queries seem legit and went through my tests and came out ok.
This one I still don't know what to do with...

[JMon]

Good breakdown dmitean.

[moisiFr]

I follow that thread and the NHL one closely because the information provided is very valuable. There is a lot of talk about which trends to play or not. As some said, the backtest will be mostly profitable because we are looking at profitable trends. Past Performance Is Not A Guarantee Of Future Returns.

What is important in a trend is the parameters. I think some of the parameters are added just to improve the ROI, but they are not explainable parameters. Let me explain. Let’s take the NHL trend F and o:streak > 3. To me this is a very good trend. Does it make sense for a hot team (streak > 3) being an underdog? The public is going to jump on the underdog. This is where the educated bettors are smarter, the bookmaker sets the “worse” team as a favorite. This trend is a typical, against the public trend which is usually profitable.

On the contrary a trend like AF and line < -3 and line > -10 and Average(margin@team and season) >= 3 and p: points > 104 and pp: points > 104 and ppp: points > 104 and rest < 2 and WP > 64.1 and game number > 16 is not good at all because there are too many parameters that are not explainable. Why would it be a play at -3.5 but not at -2.5. Why the number of points is 104 and not 105?

Basically I am not considering trends with too many parameters, and also the trends with parameter involving line, total, WP and others parameters that are too random to me.

My 2 cents
Keep up the good work

[TheLineShifter]

In my humble opinion, moisiFr has valid points. I am not an expert in this stuff but I would've cleaned all the less sensible trends with weirdo parameters attached. Some trend mathematically may present a very good percentage but it would be way better if it has a good rationale beyond the maths and numbers. Just a thought.

Dmitean is doing this the right way but with the right set of mind, tools and most importantly with a good bankroll management system, we can turn this system into a brainless investment especially for the ones who do not have extensive times to put in to analyze all the stuff daily and beat the bookies in the long run.

[emceeaye]

In my humble opinion, moisiFr has valid points. I am not an expert in this stuff but I would've cleaned all the less sensible trends with weirdo parameters within. Some trend mathematically may present a very good percentage but it would be way better if it has a good rationale beyond the maths and numbers. Just a thought.

Dmitean is doing this the right way but with the right set of mind, tools and most importantly with a good bankroll management system, we can turn this system into a brainless investment just to beat the bookies in the long run.

Just because you don't understand the rationale for queries with many variables does not mean it's not valid or even that they're over-fitted. Yes, certainly the most intuitive ones are more face valid, and therefore the results are likely attributable to variables that make sense to you. But variables that predict the outcomes of sports games are very often counter-intuitive or even entirely escape any obvious, logical rationale. This very likely may mean that you (or anyone for that matter) don't understand the influence of certain variables that by themselves may mean nothing but when interacting with others may be very predictive.o

[nash13]

In statistics we use the neural networks to create patterns and strategies for certain situations. In macro economics there are 1000s of variables in a equation and tweaking one and wighting the others is essential. same here. the more you dig into this field the more crossroads will lead to loopholes, dead ends among other things.

my second field of analyzing games is by using Rating Systems like Massey, Dunkel Index and DCI. That works well too.

[JMon]

EMC couldn't have said it better myself.

[nash13]

to add a point to emceeaye: let's not mix up over fitting and back fitting. in the google group weather wizard wrote a short article about that assumption.
We all know if you flip a coin 100 times the statistical outcome is 50 heads and 50 tails. If we flip one 500 times it is very likely that the outcome will be within 1% of the expected outcome. So try this: Flip a coin 500 times, keep records Flip it with your left hand after a heads on some occasions Flip it with your right hand after a heads on other occasions Flip it with your right hand after a tails on some occasions Flip it with your left hand after some occasions Flip it at a higher height after a heads on some occasions Flip it at a lower height after a heads on some occasions repeat the last 2 after a tails While the expected outcome of this is 50-50 there will be subsets that are much higher and lower. I found that flipping a coin after a heads with my lefthand at a higher height than the last flip came out 61-42 tails or 59.22% I also found that flipping a coin with my right hand after a tails turned u heads at 54-40 57.44%

So are these systems now that we found such great subsets that deviate from the expected outcome? NO NO NO they are expected deviations. The reason is we plugged in superficial variables to enhance the outcome after the fact. Itgave us false hope that we found something significant. The only way something has meritorious considerations is if you plug in known meaningful variables, then you have at least a chance to have found something.

but meaningful is relative. that's the whole point.

[emceeaye]

to add a point to emceeaye: let's not mix up over fitting and back fitting. in the google group weather wizard wrote a short article about that assumption.
We all know if you flip a coin 100 times the statistical outcome is 50 heads and 50 tails. If we flip one 500 times it is very likely that the outcome will be within 1% of the expected outcome. So try this: Flip a coin 500 times, keep records Flip it with your left hand after a heads on some occasions Flip it with your right hand after a heads on other occasions Flip it with your right hand after a tails on some occasions Flip it with your left hand after some occasions Flip it at a higher height after a heads on some occasions Flip it at a lower height after a heads on some occasions repeat the last 2 after a tails While the expected outcome of this is 50-50 there will be subsets that are much higher and lower. I found that flipping a coin after a heads with my lefthand at a higher height than the last flip came out 61-42 tails or 59.22% I also found that flipping a coin with my right hand after a tails turned u heads at 54-40 57.44%

So are these systems now that we found such great subsets that deviate from the expected outcome? NO NO NO they are expected deviations. The reason is we plugged in superficial variables to enhance the outcome after the fact. Itgave us false hope that we found something significant. The only way something has meritorious considerations is if you plug in known meaningful variables, then you have at least a chance to have found something.

but meaningful is relative. that's the whole point.

Right, and it should be noted that in the context of expected deviations from the mean that you are referring to, the smaller the sample size of these separate sets of flips at different heights and using different hands, the higher the likelihood of a larger deviation from the mean. As sample size of flips increase, the more regression to the mean (i.e., 50% heads and 50% tails) you are likely to see. Therefore, attributing a seemingly significant effect to the variables of "height" and "handedness of flip" would be an error. This also highlights another issue, which is how we determine statistical significance. Our sample sizes (as with the subsets of coin flips at different Heights with different handednesses), are often too small to have enough power to be statistically significant. Significance, provided a large enough sample size, is usually achieved by reaching an effect size from running a statistical test at a probability level of .05 or better--An effect with this probability level will happen by chance 5 times out of every 100 times the experiment is run. The problem is that we never actually run statistical tests on the data here in order to determine whether an effect is really significant or not. Furthermore, to make matters worse, we don't know which variables have more relative weight with respect the overall outcome so that we can weigh some variables more or less valuable than others, in order to better rank our queries with respect to their relative degrees of predictive abilities.

[nash13]

Right, and it should be noted that in the context of expected deviations from the mean that you are referring to, the smaller the sample size of these separate sets of flips at different heights and using different hands, the higher the likelihood of a larger deviation from the mean. As sample size of flips increase, the more regression to the mean (i.e., 50% heads and 50% tails) you are likely to see. Therefore, attributing a seemingly significant effect to the variables of "height" and "handedness of flip" would be an error. This also highlights another issue, which is how we determine statistical significance. Our sample sizes (as with the subsets of coin flips at different Heights with different handednesses), are often too small to have enough power to be statistically significant. Significance, provided a large enough sample size, is usually achieved by reaching an effect size from running a statistical test at a probability level of .05 or better--An effect with this probability level will happen by chance 5 times out of every 100 times the experiment is run. The problem is that we never actually run statistical tests on the data here in order to determine whether an effect is really significant or not. Furthermore, to make matters worse, we don't know which variables have more relative weight with respect the overall outcome so that we can weigh some variables more or less valuable than others, in order to better rank our queries with respect to their relative degrees of predictive abilities.

you are spot on sir

[pip2]

Fading the pip query library:

mem beats bkn 1--> -1
cha/det U 1-->+1
lal/den U 3 --> P
den beats lal 3 --> +3
phx beats hou 1 --> -1

Total +2

Running total +8

[pip2]

cle/mia U 1,
bos/atl O 2,
sac L vs mil 1,
ny L vs orl 2,
ny/orl U 1,
dal/ut U 1,
sa L vs det 7,
lal/por U 1,
mn/gs U 1,
gs L vs MN 1,
tor L vs wsh 1,
tor/wsh O 1,
hou W vs lac 3,
hou/lac O 1

[Heart]

The cool thing about sdql is they explicitly show you the win percentage of any given query up to the current date. One possibility could be to snapshot the win percentage of each query and then come back a month later (or whatever time frame might be considered significant) and re-run the queries and check to see if there is any downward trend in winning percentage of each of the queries or if they are holding steady.

edit: actually thinking about it though it's really the same method as just throwing out the queries that haven't been hitting in the past month (or whatever time period). Honestly my criteria for a query is that it has to be a winner in every season though which you would think is a good cross section to examine but for some reason 2014 overall is not hitting as well.

[nash13]

SDQL is a good tool and when you are familiar with it, it is way more than that. I would call my skills avg. I wish something like that would have been there for soccer or as we call it football in europe. The possibilities are endless. It is all just a matter of money and a group of dedicated people.

As for the status of the spreadsheet. I discussed with several users, main contributors, that it would be healthier to close the spreadsheet. Because of that the work of the pros in the Trend Market would not get abused. I would give every main contributor the chance to vote. It is not my main decision. The work of the guys in her is very valuable, so I would love to see them get paid for that.

As for my own process of being a seller in the Trend Market. Joe from Killersports will give me a spot in there, but at a side part for qualifying users. There will be Masters, who are contributing a long time.

[emceeaye]

SDQL is a good tool and when you are familiar with it, it is way more than that. I would call my skills avg. I wish something like that would have been there for soccer or as we call it football in europe. The possibilities are endless. It is all just a matter of money and a group of dedicated people.

As for the status of the spreadsheet. I discussed with several users, main contributors, that it would be healthier to close the spreadsheet. Because of that the work of the pros in the Trend Market would not get abused. I would give every main contributor the chance to vote. It is not my main decision. The work of the guys in her is very valuable, so I would love to see them get paid for that.

As for my own process of being a seller in the Trend Market. Joe from Killersports will give me a spot in there, but at a side part for qualifying users. There will be Masters, who are contributing a long time.

Yes, I will be selling trends on there as well.

[JMon]

Yes, I will be selling trends on there as well.

Yes, second