Originally posted by chopperocker
???????????????????
Very peculiar % given here...
-
~~~~~~~~~~~~~~~~~
???????????????????
Very peculiar % given here...Comment -
Well if you add a few more question marks to each line it may become more clear...Originally posted by 19th HoleComment -
~~~~~~~~~~~~~Originally posted by Mako-SBR
Yeah ... Crystal clear.Comment -
bk'ing for later....
ThanksComment -
don't think this is one of mine, but had it saved and noticed the playoff record.
p:HFL and op: W and H and -10<=line<-5 and o:WP>50 and playoffs=1 and 2011<=seasonComment -
points in 4th quarter vs 3rd quarter
SDQL experts,
I'm trying to use SDQL to determine the average points scored in the fourth quarter compared to that of the other three quarters of NBA games (or specifically playoff games if we would expect to see different results than from regular season games). All I've been able to come up with is thebfollowing, but its not giving me what I want.
total and playoffs=1 and series game=2 and 2011<=season and P4
I ask because I'd like to do live in-play betting at the end of the third quarter to try to predict the over-under total. If we thought the total points scored by the end each quarter of a 200 pt-game would increase linearly (e.g., quarter 1=50; quarter 2=100; quarter 3=150; quarter 4=200), then at the end of the third quarter we can predict the final total by using the following formula:
3/4*x=150. (x=final total points)
x=200
So, if we know that more or less points are scored in the 4th quarter than in the other quarters due to variables like fatigue, a very close scoring margin, etc., we can modify the above formula accordingly.
Any help towards pointing me in the right direction to get this imformation would be helpful and appreciated.Comment -
To put it simple, the problem is that scoring between quarters is practically unrelated. The only thing that differentiates 4th quarter from any other quarter is the possibility of intentional foul game down the stretch. Otherwise it is pretty random. Even such factor as fatigue can play different roles from game to game - sometimes it can resemble as poor defence, which would lead to more points scored, poor offence, or both...I don't want to crash your dreams, but it is just way too many variables to put in a formula of the kind you proposed.Originally posted by emceeayeComment -
Thanks for the input.Originally posted by JAnthony
Yeah clearly there are innumerable variables that contribute to the points scored in the 4th quarter vs 3rd (or any other quarter for that matter) to determine a model and the correct regressors to include in it to accurately predict the point totals for each quarter.
Now, is it "random" as you say? Definitely not. You yourself even pointed out some factors that would influence the outcome, like poor defense or poor offense.
Regardless, if you can bring to bare any general information at the end of the third quarter that would influence the final total of the game, specifically whether the average 4th quarter points exceed that of the 3rd or of each do the other two quarters, then you can crudely calculate an estimate of the final score.
And if through this crude estimate I'm able to accurately predict the OU 51% or more of the time, then I'm happy. So really, please don't worry about "crashing" any 'dreams"
Do you happen to know how I can determine average points scored in the different quarters of the game?
Thanks in advance.Comment -
shortcuts for points scored in the 1st, 2nd and so forth are P1, P2.....Originally posted by emceeayeThanks for the input.
Yeah clearly there are innumerable variables that contribute to the points scored in the 4th quarter vs 3rd (or any other quarter for that matter) to determine a model and the correct regressors to include in it to accurately predict the point totals for each quarter.
Now, is it "random" as you say? Definitely not. You yourself even pointed out some factors that would influence the outcome, like poor defense or poor offense.
Regardless, if you can bring to bare any general information at the end of the third quarter that would influence the final total of the game, specifically whether the average 4th quarter points exceed that of the 3rd or of each do the other two quarters, then you can crudely calculate an estimate of the final score.
And if through this crude estimate I'm able to accurately predict the OU 51% or more of the time, then I'm happy. So really, please don't worry about "crashing" any 'dreams"
Do you happen to know how I can determine average points scored in the different quarters of the game?
Thanks in advance.
to get averages you need to use summatives, which can be found how to use at sdql dot com. Then you could follow by group-bys to look at wide range of numbers in one query.Comment -
4th quarter scoring
616-318 (4.79, 66.0%) 490-432-12 (0.69, 53.1%) avg line: -4.1 447-471-16 (-0.13, 48.7%) avg total: 190.2 35.32 45.1 20.52 75.3 6.27 35.1 5.25 11.23 42.07 22.05 20.11 12.63 24.5 24.2 23.9 24.0 97.4 33.91 43.5 18.63 75.2 6.20 34.2 4.65 10.64 40.13 23.59 18.24 13.43 22.8 22.9 22.8 23.3 92.6
The amount of points scored in the 4th is under "Q4" above, the average points scored for 1st=47.3, 2nd=47.1, 3rd=46.7 and 4th=47.3
You can vary the query to get different results by putting a total in the query box, for example....
HF and playoffs and total<200, which will obviously change the results. And if you wish to see how a scenario will change by the game situation, put into the margin between the two teams after the 3rd quarter. For example to query how a team does when winning by >6 points after the third quarter, use the query....
playoffs=1 and HF and M3>6
This is covered in the query manual I believe. In the above query, fourth quarter scoring increases by an average of .2 points over the other quarters, which when you acknowledge that overtime happens once in while which is included in fourth quarter scoring, fourth quarter scoring on average would actually decrease.
For your purposes of finding whether you can find an edge by betting over in the fourth, there doesn't appear to be one.Comment -
Can't you get rid of the overtime factor with "overtime = 0"?Comment -
just type in overtime and see if it's a parameter, easiest way to checkOriginally posted by pip2
Comment -
Green,
Thank you for this. It'll be really useful. I now have the commands I'm going to run at the end of thr 3rd period or into the 4th for the every commercial break betting. What I'm wanting to do is to bet on the OU on the game final total. I just want to experiment with this to see if using the 3/4x=(total after 3 quarters) formula to estimate the final score and to consider also the average 4th quarter points you provided (taking into account the M3 margin) to estimate the final score, and to see if after comparing it with the Vegas OU line at the time if my estimate deviates significantly from their provided total line. Then I can experiment and bet very small amounts accordingly to see how often its possible to out predict the Vegas line totalComment -
You took there only half of my point about 4th quarters. Like I said, it can be the sloppiest and it can be the toughest of the quarters. For example, yesterday in MEM v. OKC game both teams scored 65 points combined in 4th quarter, which honestly made no much sense considering these both teams, today also look at the points scored in TOR v. BKN game in 4th quarter and look at IND v. ATL. What conclusions exactly can you draw from that?Originally posted by emceeayeThanks for the input.
Yeah clearly there are innumerable variables that contribute to the points scored in the 4th quarter vs 3rd (or any other quarter for that matter) to determine a model and the correct regressors to include in it to accurately predict the point totals for each quarter.
Now, is it "random" as you say? Definitely not. You yourself even pointed out some factors that would influence the outcome, like poor defense or poor offense.
Regardless, if you can bring to bare any general information at the end of the third quarter that would influence the final total of the game, specifically whether the average 4th quarter points exceed that of the 3rd or of each do the other two quarters, then you can crudely calculate an estimate of the final score.
And if through this crude estimate I'm able to accurately predict the OU 51% or more of the time, then I'm happy. So really, please don't worry about "crashing" any 'dreams"
Do you happen to know how I can determine average points scored in the different quarters of the game?
Thanks in advance.
And hitting 51% won't get you anywhere. That is a fact.Comment -
I was looking to something similar, except it was using halftime results to predict final results. The great thing about either idea is that you're removing a lot of surprises from the equation as far as somebody turns out to be hurt, orOriginally posted by emceeaye
Pop suddenly decides to rest half his team...maybe there is some kind of connection that isn't obvious but might be useful -- for example if one team is ahead or behind a certain number of points at the end of the third quarter, then perhaps more or less points tend to be scored during the 4th quarter...Comment -
I was looking to something similar, except it was using halftime results to predict final results. The great thing about either idea is that you're removing a lot of surprises from the equation as far as somebody turns out to be hurt, orOriginally posted by emceeaye
Pop suddenly decides to rest half his team...maybe there is some kind of connection that isn't obvious but might be useful -- for example if one team is ahead or behind a certain number of points at the end of the third quarter, then perhaps more or less points tend to be scored during the 4th quarter...
Actually I always wanted to ask the experts -- I'm pretty sure the answer is no, but then I never saw the "in seed " parameter before last night either -- is there any way to pick up information by halves? I would love to query what happens to end of game o/u's and spreads given a certain free throw percentage or number of turnovers in the first half...Comment -
Yes, the less time in between making a bet and the end of the game minimizes the number of unexpected variables to affect the outcome of the game.Originally posted by pip2
This reminds me of what happened in the Bulls/Wizards game last night. If we could look at the mathematical difference between points scored in the first and second quarters and the average points that are scored in those quarters while holding at least some of the situational variables constant (e.g., first quarter or 2nd quarter margin, play off game, etc.), and see if there is a relationship between the values of those differences and number of points scored in the 3rd or 4th quarters (e.g., if a team is scoring 6, 7, or 8 points more than average for those first quarters, and notice that the 3rd and 4th quarter points scored become increasingly less than average as the more above average the points get for the first quarters), then we can make more informed decisions about OU in the last quarter of games.Originally posted by pip2Comment -
Good fortune to those that want to look into this....there are some in other forums that have worked out second half systems that are predicated on the score of the first half. If I remember correctly, it was based on if the first half varied by greater than either 6 or 8 points from the expected outcome. For example, if the Bulls were favored over the Celtics by -8 and the total was 200 for the whole game...then the expected first half score would be Bulls 52, Celtics 48 at half. If for example, the Celtics were leading at half 48-41, then there would be a play on both the total and the side in the second half. I am not sure, but I think the play would be on the Celtics and the under in the second half, in other words things did not revert to the mathematical mean. I did some research on this earlier, and there was a small edge doing this...but I was just not that interested, for whatever reason.
From observation and a cursory glance, it seems that second quarter scoring is higher than first quarter, and if the books have not studied this (highly unlikely that they haven't) then there would be an edge there.
For SDQL, the way to query results for the whole game with known results at the half, is "M2" or "margin at the half".
An example of a query would be
HF and playoffs=1 and total>200 and M2<0 and line<-3
17-21 (-1.55, 44.7%) 6-31-1 (-7.75, 16.2%) avg line: -6.2 21-16-1 (2.49, 56.8%) avg total: 209.6 37.76 45.0 22.05 75.1 7.71 35.1 4.68 11.82 42.95 22.58 20.13 13.47 25.3 23.3 27.8 27.7 105.3 39.47 47.7 20.21 78.0 7.68 39.5 5.26 10.13 42.26 24.63 20.18 13.42 27.5 27.7 25.3 24.9 106.8
Keep in mind that the books will vary the total and side according to the first half results and as far as I know there is no place to access historical second half lines and totals, so to go back and do retroactive studies on this as I attempted may be problematic.
It'd be nice for you folks to come up with something good if you go ahead....I'll be rooting for you.Comment -
This doesn't seem to work in series games 1 and 2 too well.
HF and playoffs=1 and p:M2<4 and line<=-5 and round<3Comment -
there was a late moneyline drop on the bulls last night, as well as the rockets on sunday, have to weigh some of that stuff in when evaluating gamesComment -
I haven't looked closely at those games yet. They could be anomalies or instead reflective of the influence of very objective factors that have yet to be identified. but if they were anomalies in some way as you're implying, it wouldnt matter much in the big picture because they only account for a small fraction of a much much larger sample of games in a season and therefore inconsequential in the long run. What conclusions can I draw from what seem to you to be anomalies/random but which may in fact turn out to be very predictable? None yet, because I haven't had a chance to do the research to determine the answer. Why don't you go ahead and use those examples not as a way to support your hypothesis that what you observed was random, but rather explore what variables/systems, if any, can account for it? Go take a statistics class and discover how a t-test, multiple regression or statistical modeling can help you figure that problem out.Originally posted by JAnthony
Actually, that depends on how much money you are wagering. That is the factOriginally posted by JAnthonyComment -
You don't have to take any statistics class to conclude, that score does not increase linearly throughout the quarters. What anomalies? Can you show me at least 10 games in which score progression is somewhat the same throughout the quarters?Originally posted by emceeaye
The conclusion you can draw is, that you don't have to search hard and far to prove that those scores are more or less random in a sense, that you won't be able to predict final score after 3 quarters.
Bookmakers are doing exactly what you are trying to do - modelling possible total outcome. Yes, 15 years ago it was pretty much the way you are trying to convince that it is. Then those who noticed could take advantage of the second half lines, because bookies basically divided pre-game total in half and did not account for possible foul game and other factors towards the end of the game. But all those angles are long gone.Comment -
Yeah, of course not. Just to clarify, I wanted to use that formula just to provide a crude index of what it is in addition to something more precise akin to what Green suggested to take into account the margin. And, that linear equation is not what you learn from a statistics class that you can bring to bear here. You would learn stats tests to run on this data to discover new powerful predictive equations and to verify these trends.[/QUOTE]Originally posted by JAnthony
No, again, they are not random. You just don't have the tools to use to create models to account for the different variables that influence the outcome frequently enough to make a profit.Originally posted by JAnthony
Yes, the bookmakers likely are using sophisticated modeling techniques to better predict outcomes. Does that mean they they figured everything out already? Probably not--and Im sure there are plenty of people out there who have discovered their own models for predicting outcomes at a high enough rate to yield a profit using stats they learned in stats classes. You aren't going to hear about it, because those individuals aren't going to publish their findings. Your perspective sounds surprisingly defeatist given you are on this forum trying to better predict outcomes of games using these trends.Originally posted by JAnthonyComment -
Originally posted by emceeaye
No, again, they are not random. You just don't have the tools to use to create models to account for the different variables that influence the outcome frequently enough to make a profit.
Yes, the bookmakers likely are using sophisticated modeling techniques to better predict outcomes. Does that mean they they figured everything out already? Probably not--and Im sure there are plenty of people out there who have discovered their own models for predicting outcomes at a high enough rate to yield a profit using stats they learned in stats classes. You aren't going to hear about it, because those individuals aren't going to publish their findings. Your perspective sounds surprisingly defeatist given you are on this forum trying to better predict outcomes of games using these trends.[/QUOTE]
What I'm trying to say that you are doing the same thing here, just speculating with some wild assumptions. Without any reasoning behind it. By saying "random" I meant sophisticated and random enough for putting it all in one pot, in this case - formula or algorithm.
The best in the world who work with models break it down to a single possession and add values to every player individually. Your approach is way too general to say the least.Comment -
What I'm trying to say that you are doing the same thing here, just speculating with some wild assumptions. Without any reasoning behind it. By saying "random" I meant sophisticated and random enough for putting it all in one pot, in this case - formula or algorithm.Originally posted by JAnthony
The best in the world who work with models break it down to a single possession and add values to every player individually. Your approach is way too general to say the least.[/QUOTE]
What approach are you talking about? Same thing as what? I'm speculating, and I make no assumptions, wild or otherwise. Im trying to generate hypotheses--I would like to develop an approach like everyone in here but haven't done so yet. The best in the world will not tell you how they do it, but a good model I suppose would incorporate data on each player individually.
First you imply random by saying "random", and now you are saying you meant "sophisticated"... Its really confusing where you're going. But in any case, its certainly not random and unpredictable as you have been saying repeatedly. If you learn some stats, and have some faith that you can make some headway in predicting outcomes with those statistics, you'll be more successful with it. Good luck, buddy.Comment -
JAnthony and emceeaye, you've both posted your opinions on the matter, it's time to move on to the next query/scenario.
Lot of playoffs left to be examined, get on it.Comment -
Nice green, I'm aware of the second half threads you referenced and have always wondered why we haven't seen more SDQL queries that point in a similar direction.Originally posted by green7
Comment -
playoffs=1 and series game>1 and H and total<=190 and -11<=line<=-8Comment -
playoffs=1 and series game>1 and H and total<=190 and -11<=line<=-8 and p:W
That they won last game improves their chances ATS.Comment -
ATS not good
Actually, this one shows ATS not great for number one seeds in this situation:
playoffs=1 and H and seed in [1] and round<4 and streak>-4 and o:streak>-4 and o:streak<6 and series game=2 and p:W and total>167 and total<=190Comment -
why wouldn't this be showing up tonight???
total >= 200 and playoffs = 1 and series game = 3 and 2003<=season and rest<3 and HComment -
H and P:AL and P:season=season and P:margin<=-30 and p:AL and D and 2005<=season and rest<3 and line<8.5
A and -3 < line < 3 and p: ou margin >= 18 and 75 >= WP >= 60 and o:WP > 50 and 2005 <= season and rest <= 2
Comment -
DB had an error, playoffs=1 isn't functional and hasn't been all day.Originally posted by JMon
Comment -
Oh.. can't blame though... shit is done for free. Everything is done by human.Originally posted by Mako-SBRDB had an error, playoffs=1 isn't functional and hasn't been all day.Comment -
East always busts balls on them whenever anything isn't done chop-chop, tough position to be in for sure.Originally posted by JMon
Comment
