Punto Banco Posted February 10, 2008 Share Posted February 10, 2008 Re: Lunatism’s Statistical Selections (lss) I must hit the hay with work time edging ever so close some sleep MAY be required however to that question about sample size i say depends... main thing is to compare performance of the CS predictions for matches you bet on and the rest or perhaps find which grouping perform better as you just mentioned perhaps those that predict 1-0, 1-1, 0-1 etc also of course average odds on these CS is also important as these tend to be shorter you may hit higher strike % but make less due to unappealing odds... though from what I see from all this you have it covered, I will just stop being impatient and wait until you get it sorted (My thoughts best not to rush anyways plenty of time.... says me...:lol:lol:lol) but anyways I will be able to think up a real answer tomorrow... 4 now :zzz Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 10, 2008 Author Share Posted February 10, 2008 Quote Link to comment Share on other sites More sharing options...
Punto Banco Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) :eek .... I'm glad I brought it up.... more later.... Quote Link to comment Share on other sites More sharing options...
F. Posted February 11, 2008 Share Posted February 11, 2008 Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 11, 2008 Author Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) what were the average odds you got for the correct score market? Your preliminary results in the correct score market look very promising (to say the least !!) but I'd rather try to find any potential flaws first before getting too excited... i know exactly what you mean - i've been staring at those figures for AGES trying to "figure out" what could possibly be the catch here, but i can't find one..... i'm wondering whether i should sit it out for another month to be sure (would probably double the sample size), or just go straight in right now!! the average odds are shown directly underneath the scores on the little tables....but to be easy 1-0 was 6.50, and 1-1 was 5.85. I guess these numbers could change over time too, i might just artificially shorten the average odds to cover for that in that - what do you think? .....and of course different bookies would have different averages Quote Link to comment Share on other sites More sharing options...
F. Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) Do you remember that I suggested you to gather a minimum of 150 bets and then draw conclusions? Well, the goal was to test the robustness of your selection system. Basically, a 15% yield on 150 selections means the robustness of your system is above 99%. Obviously these figures are well above the 99% level of confidence, and the only catch I spot at first sight could be the variance due to the high odds, but even then, a 166% yield on 300+ selections looks strong enough to me... Let me make some quick calculations and I'l get back to you... F. Quote Link to comment Share on other sites More sharing options...
palaceman Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) I'm just trying to play devils advocate here mate, but in many years (mostly successful) experience of gambling I have learned that a new system or idea is often not as good as initial research suggests it might be. I have also learned that no great harm is done by testing it for another month or three before committing to it. There can only be two outcomes of such diligence, both of them good:- (i) It will demonstrate that the idea is indeed profitable; and although you've missed out on a couple of extra months profit whilst you researched it further, if it's as good as you think then you'll be reaping in the profits for the next 50 years anyway, so what's a couple of months in the total scheme of things. (ii) It will turn out to be not as good as you'd been lead to believe, and you'll be glad you shown such self-restraint because it has saved you money. If there is a drawback to what you have written, it seems to me to be likely to relate to the strike rate. Obviously I haven't checked your research, and I'm happy to assume the results as accurate. However, I believe you are likely to struggle to maintain a strike rate of anywhere near the 61% you say you have achieved for 1-1 draws. But I'd love you to prove me wrong. :ok Quote Link to comment Share on other sites More sharing options...
F. Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) I'm just trying to play devils advocate here mate, but in many years (mostly successful) experience of gambling I have learned that a new system or idea is often not as good as initial research suggests it might be. I have also learned that no great harm is done by testing it for another month or three before committing to it. There can only be two outcomes of such diligence, both of them good:- (i) It will demonstrate that the idea is indeed profitable; and although you've missed out on a couple of extra months profit whilst you researched it further, if it's as good as you think then you'll be reaping in the profits for the next 50 years anyway, so what's a couple of months in the total scheme of things. (ii) It will turn out to be not as good as you'd been lead to believe, and you'll be glad you shown such self-restraint because it has saved you money. If there is a drawback to what you have written, it seems to me to be likely to relate to the strike rate. Obviously I haven't checked your research, and I'm happy to assume the results as accurate. However, I believe you are likely to struggle to maintain a strike rate of anywhere near the 61% you say you have achieved for 1-1 draws. But I'd love you to prove me wrong. :ok Those are wise words indeed, and I agree with you. As I said before, even with such promising results we shouldn't get too excited. Let's try to find as many objections to this proposed system and we'll all be better of, Luna included. As for me, Luna passed my first test (see next post) F. Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 11, 2008 Author Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) Do you remember that I suggested you to gather a minimum of 150 bets and then draw conclusions? yep, and now that i have that many - i'm still unsure! if this strike rate & yield had been lower, i'm sure i would have been more confident - but because it's so high (compared to what i would have expected), i'm a little hesitant.....a little hesitant, but still a whole load excited about the potential of something like this Let me make some quick calculations and I'l get back to you... sounds ominous :tongue2 i'll be hanging around waiting to see what you come up with Quote Link to comment Share on other sites More sharing options...
F. Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) the average odds are shown directly underneath the scores on the little tables....but to be easy 1-0 was 6.50, and 1-1 was 5.85. I guess these numbers could change over time too, i might just artificially shorten the average odds to cover for that in that - what do you think? .....and of course different bookies would have different averages I just checked the average odds you gave me and they are right on spot, with an average overround of 30%, which is the norm in this market. In other words, your (theoretical) profit has been made using standard odds. :eek:eek What else could we try to sink your system?? :rollin F. Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 11, 2008 Author Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) I'm just trying to play devils advocate here mate, but in many years (mostly successful) experience of gambling I have learned that a new system or idea is often not as good as initial research suggests it might be. I have also learned that no great harm is done by testing it for another month or three before committing to it. There can only be two outcomes of such diligence, both of them good:- (i) It will demonstrate that the idea is indeed profitable; and although you've missed out on a couple of extra months profit whilst you researched it further, if it's as good as you think then you'll be reaping in the profits for the next 50 years anyway, so what's a couple of months in the total scheme of things. (ii) It will turn out to be not as good as you'd been lead to believe, and you'll be glad you shown such self-restraint because it has saved you money. If there is a drawback to what you have written, it seems to me to be likely to relate to the strike rate. Obviously I haven't checked your research, and I'm happy to assume the results as accurate. However, I believe you are likely to struggle to maintain a strike rate of anywhere near the 61% you say you have achieved for 1-1 draws. But I'd love you to prove me wrong. :ok hey palaceman - i need a devil's advocate right now - you've answered my prayers :D it's really great that you guys are helping out with this..... like you i want to have as many anti-points as possible, so perhaps we could bat them all down together (well....that, or watch the "idea" crash and burn!). you're of course right - time won't be a bad thing, since in the end, the money now v money later debate shouldn't be an issue if there is more data and more substance behind the stats...since that would just bring more confidence bla bla bla (i'm waffling, and it feels dirty) i can see what you mean about the strike rate - and over the coming days, weeks, months that is what i'll need to keep an eye on most of all, hopefully it can buck the trend of the "good for a month" systems and hover around even 50% (which would be fine) if you think it would help, i can post an excel file with my results/cs predictions in if you like, but i honestly think that that will not prove much.....since it's after the events in question. hopefully you guys trust my data ;) Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 11, 2008 Author Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) What else could we try to sink your system?? :rollin erm....well if ANYONE has any past form tables and league tables that can be looked at on a week by week basis for the leagues i listed above, that would be of great help...... the only reason i couldn't go further back than the 6th january is because i discovered i wasn't keeping complete records - and of course i need the full league table and form details for each team for each day of the week. (it might sound OTT, but that's how i've done it, to cover for any table corrections or the odd game that gets played during the week). just means that even though i'd pretty much written the code to run all these checks, my pc was "frozen" for the duration of "Welcome To The Jungle" (t4 last night.....nothing else to do while i waited), as it processed the league/form table! so really, what i need is a super computer to calculate it that way...... another way to do it is this (but i "can't do it" i don't think i'm "good enough"!!!! :loon), some code that could look at a list of results, and make form tables and league tables based on those results for each given date. The biggest problem would be ensuring that the league start dates were adhered to by this "code", but if anyone could somehow make such calculations, i do have the results at hand. once the tables are made, all i need to do is generate the CS prediction and then compare the predictions to the real scores....of course that presumes i could get my hands on a super computer, or face my pc being out of action for the next week or so as it trudged through the decades of results i have! :D any takers? Quote Link to comment Share on other sites More sharing options...
F. Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) i can see what you mean about the strike rate - and over the coming days, weeks, months that is what i'll need to keep an eye on most of all, hopefully it can buck the trend of the "good for a month" systems and hover around even 50% (which would be fine) if you think it would help, i can post an excel file with my results/cs predictions in if you like, but i honestly think that that will not prove much.....since it's after the events in question. hopefully you guys trust my data ;) I am pretty sure the system won't keep a 61% strike rate in the correct score market, but if it manages to achieve half of that (30%) it will be well into profit. I personally don't think posting an excel file would help much. I've seen in previous post of yours the kind of research you like to conduct and I'm sure it's well done. F. Quote Link to comment Share on other sites More sharing options...
F. Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) erm....well if ANYONE has any past form tables and league tables that can be looked at on a week by week basis for the leagues i listed above, that would be of great help...... [...] another way to do it is this (but i "can't do it" i don't think i'm "good enough"!!!! :loon), some code that could look at a list of results, and make form tables and league tables based on those results for each given date. The biggest problem would be ensuring that the league start dates were adhered to by this "code", but if anyone could somehow make such calculations, i do have the results at hand. once the tables are made, all i need to do is generate the CS prediction and then compare the predictions to the real scores....of course that presumes i could get my hands on a super computer, or face my pc being out of action for the next week or so as it trudged through the decades of results i have! :D any takers? That's *precisely* what I mean when I say that I know the kind of research you like to conduct :loon:loon F. Quote Link to comment Share on other sites More sharing options...
F. Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) Just for the record, here is the quick robustness test I like to perform on candidate systems: It's an old thread in here, but high quality stuff posted by very knowledgeable members from that day (including Joseph Buchdal himself) and it's very quick and simple to perform. I'm pretty sure you will achieve confidence levels above 99.999999% (and I'm not kidding) F. Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 11, 2008 Author Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) I personally don't think posting an excel file would help much. I've seen in previous post of yours the kind of research you like to conduct and I'm sure it's well done. haha, thanks (compliment, right?:)) it seems i've a reputation for going way over the top when it comes to research...... probably something to do with all this "free" time that i get by substituting sleep for more "excel related" challenges i guess ;) Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 11, 2008 Author Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) Just for the record' date=' here is the quick robustness test I like to perform on candidate systems: It's an old thread in here, but high quality stuff posted by very knowledgeable members from that day (including Joseph Buchdal himself) and it's very quick and simple to perform. I'm pretty sure you will achieve confidence levels above 99.999999% (and I'm not kidding) F. a little gem hiding here all this time! you'll have to help me out a little bit though - or did i (or i guess it's we, since you seem to have similar) calculate it correctly that there are 20 0's following a nought for the chance of chance? i've got my calculations wrong, right?! what did you get as the average fair estimation and the bookies percentage estimation? Quote Link to comment Share on other sites More sharing options...
F. Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) a little gem hiding here all this time! you'll have to help me out a little bit though - or did i (or i guess it's we, since you seem to have similar) calculate it correctly that there are 20 0's following a nought for the chance of chance? i've got my calculations wrong, right?! what did you get as the average fair estimation and the bookies percentage estimation? hahaha... I think you did it right :rollin Forget about stats and think about it for a while: what's the chance of *randomly* achieving a 160% yield in a market with 30% overround after 300 bets? Now take a look at the result you just calculated,and that's your answer :lol F. Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 11, 2008 Author Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) hahaha... I think you did it right :rollin hmmm, well, as you said before - we'll need to find more tests...other ways to poke the system...... i'm lost - generally (as far as testing has gone with me), i've always just looked at yield and strike rate and dashed any system/idea that wasn't showing profit - was aware of the Chi Squared Tests, but never applied it (forgot i guess), i have no idea what else i could do....(apart from getting hold of more past data) but in the next few hours (or minutes), i'll get to work on some graphs that will show how each league performs....since i've nothing better to do :sad Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 11, 2008 Author Share Posted February 11, 2008 Quote Link to comment Share on other sites More sharing options...
leespam Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) erm....well if ANYONE has any past form tables and league tables that can be looked at on a week by week basis for the leagues i listed above' date= that would be of great help...... Hi Luna, haven't read the whole thing but looks promising. I'd echo the previous posts in exercising caution though. This link might be of interest to you. I don't think it'll cover all your leagues but it can give some tables after X amount of games: http://www.stats.betradar.com/statistics/betway/ It's a bit drop-down heavy as well but hopefully it's of help in back-testing a bit further. Only right that I offer some input after all your help in my thread mate :ok Quote Link to comment Share on other sites More sharing options...
palaceman Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) I'm not dismissing the data, but the danger of sub-dividing into leagues like that is that within each division, the number of matches involved in your research is greatly reduced (naturally) so it might be dangerous to draw conclusions based on much more limited numbers. Just by way of example, those leagues that are showing 100% success at certain scores, could you let us know what number of matches that is based on in each one? Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 11, 2008 Author Share Posted February 11, 2008 Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 11, 2008 Author Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) Hi Luna, haven't read the whole thing but looks promising. I'd echo the previous posts in exercising caution though. This link might be of interest to you. I don't think it'll cover all your leagues but it can give some tables after X amount of games: http://www.stats.betradar.com/statistics/betway/ It's a bit drop-down heavy as well but hopefully it's of help in back-testing a bit further. Only right that I offer some input after all your help in my thread mate :ok thanks for that link leespam - i'm gna be up all night extracting tables from there!!!! but at least it does manage to have the form and league info, aswell as catering to around 85% of the leagues in the list i put up here..... only problem is the postponement of matches (which can have big effects on the selections) - this site can calculate the tables based on the round, but if there is a cancelled match, as soon as that game is played the round will be updated as if the match was played at that date - annoying, but that makes can make a HUGE difference, considering the way my system works..... hmmmmmmm, i don't actually know what to do about that - if i backtest with tables calculated in that way, i'll be producing inaccurate predictions, meaning inaccurate stats :wall:wall:wall looks like i might have to give this one a miss - damn my tricky system :@ but i'll definitely find a use for this site though leespam, and i'm glad you posted it here :ok Quote Link to comment Share on other sites More sharing options...
happygooner Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) Interesting thread here. Will definitely be looking over during the course of next few gameweeks. Just one other thing. Wondered if you'd considered the possibility of how well your system would perform in over/under bets? Figured that as the CS was showing such promise, actual scoreline could be very close to that predicted. Only a thought anyway. Gl with this:ok Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 11, 2008 Author Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) Interesting thread here. Will definitely be looking over during the course of next few gameweeks. Just one other thing. Wondered if you'd considered the possibility of how well your system would perform in over/under bets? Figured that as the CS was showing such promise, actual scoreline could be very close to that predicted. Only a thought anyway. Gl with this:ok now that's an idea :D and of course it would be linked to the scores, so i guess it's a logical step, i'll pull up some stats on it later in the evening, probably around the same time i'll post my LSS selection(s) for tomorrow (i'm actually going out for once tonight!)..... one thing i do wonder though, what are the odds like for the over/under bets? since i guess you'd need pretty high ones (or really high strike rate) to keep "up" the profit levels i did briefly scratch the surface of over/under's a few years back, and think i might have come up with a brick wall along the lines of - not enough bookies provide this data for the lower leagues. but that isn't so much of a problem - not with all this new data i've got my hands on :tongue2 so later on....for you, i'll put up some stats on that though (might be late though!) thanks for your input Quote Link to comment Share on other sites More sharing options...
happygooner Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) now that's an idea :D and of course it would be linked to the scores, so i guess it's a logical step, i'll pull up some stats on it later in the evening, probably around the same time i'll post my LSS selection(s) for tomorrow (i'm actually going out for once tonight!)..... one thing i do wonder though, what are the odds like for the over/under bets? since i guess you'd need pretty high ones (or really high strike rate) to keep "up" the profit levels i did briefly scratch the surface of over/under's a few years back, and think i might have come up with a brick wall along the lines of - not enough bookies provide this data for the lower leagues. but that isn't so much of a problem - not with all this new data i've got my hands on :tongue2 so later on....for you, i'll put up some stats on that though (might be late though!) thanks for your input Odds tend to be close to evens as it's a 50/50 chance. Lowest would be around 1.5 (fairly rare) with highest around 2.70 (again quite rare). Quote Link to comment Share on other sites More sharing options...
The Hat Man Posted February 11, 2008 Share Posted February 11, 2008 Re: Lunatism’s Statistical Selections (lss) Echo the comments above that this continues to look very interesting :ok I'd just like to make two further points to those raised above: 1) I know you've said before that you feel some loyalty to Bwin and this is why all your fixed odds prices are taken from their site. I'm not sure that this is practical. As an example, the Italian fixture that you posted on Sunday was 1.4 at bet365 compared to 1.35 at bwin. When you are dealing with heavy odds-on shots this is quite a significant difference and will have quite an impact on yield over time. In any case, i don't think there are many bookies that offer bets on this quantity of leagues so it may often come down to a direct comparison between bwin and bet365. Are you aware of any other sites which price up these "minor" leagues? The rule of thumb must surely ALWAYS go with the best price. 2) Have you looked into the possibility of multiples betting (eg. doubles/trebles and upwards)? While this will clearly result in a lower strike rate, bumper days like Saturday would have resulted in a high return. In fairness, I don't even know if pinnacle let you place multiples, but just a thought. Keep it up! ;) Quote Link to comment Share on other sites More sharing options...
palaceman Posted February 12, 2008 Share Posted February 12, 2008 Quote Link to comment Share on other sites More sharing options...
lunatism Posted February 12, 2008 Author Share Posted February 12, 2008 Re: Lunatism’s Statistical Selections (lss) Echo the comments above that this continues to look very interesting :ok thanks...hopefully the next batch of selections (for LSS) and the next weeks worth of CS predictions tell a pretty tale so we've all got something to smile about 1) I know you've said before that you feel some loyalty to Bwin and this is why all your fixed odds prices are taken from their site. I'm not sure that this is practical. As an example, the Italian fixture that you posted on Sunday was 1.4 at bet365 compared to 1.35 at bwin. When you are dealing with heavy odds-on shots this is quite a significant difference and will have quite an impact on yield over time. In any case, i don't think there are many bookies that offer bets on this quantity of leagues so it may often come down to a direct comparison between bwin and bet365. Are you aware of any other sites which price up these "minor" leagues? The rule of thumb must surely ALWAYS go with the best price. i can see what you're saying here, and you make an excellent point about bet365/bwin being the only two who cater to the minor leagues...but you wonder about my bias? :) basically, i DID contact the two bookies, and was pretty much given lists of all the leagues they catered to, and bwin came out trumps by a long way. i thought for peace of mind and for the sake of not splitting a bank over the two accounts (for when the system shifts to live), i'd stick to bWin. But in terms of a nice enough strike rate, you're right, i should always try and find the best odds...and so from now on, i will try and do that more often! 2) Have you looked into the possibility of multiples betting (eg. doubles/trebles and upwards)? While this will clearly result in a lower strike rate' date=' bumper days like Saturday would have resulted in a high return. In fairness, I don't even know if pinnacle let you place multiples, but just a thought.[/quote'] this, is a little more tricky - for starter, pinnacle don't do combo's, and because of the smaller number of minor leagues they provide, bet365 wouldn't be such a good idea either, leaving bwin. and the problem then, is that i cannot think of a suitable way to stick these together! if i picked the highest rated fixtures, and since i wouldn't (in most cases) be able to use AH on these, i might end up with a couple of 1.25's, combined to make a "meagre" 1.56, of course profit is profit at the end of the day, but when they combine for such low odds, i just wonder if there is any point. what would your take on that be? if you have any possible ideas for how i could apply combos to my selections, i'd love to hear them - since multiples are generally not my forte :tongue2 Quote Link to comment Share on other sites More sharing options...
.thumb.jpg.32384c3055cbc1254b0eb6212fda1359.jpg)
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.