Mulkis Posted April 25, 2011 Share Posted April 25, 2011 Re: Fair or not fair ? So the deviation from the average can in itself generate profit, but only if the betting is done with at least a neutral, preferably a positive expectation of value. Otherwise the negative value will offset the positive deviation and you just go around in circles. Correct. All we need a volantier to papertest this "heresy". I would go for it myself, but have no time at the moment. Quote Link to comment Share on other sites More sharing options...
slapdash Posted April 25, 2011 Share Posted April 25, 2011 Re: Fair or not fair ? R2 is the correlation coefficient. It shows how much of varyability in Y (actual results of the events) is explained by the X (opening odds of the underlying events). For my test I used football-data .xls bet365 historical odds. My model was segmented into +-0.05 odds (if odds were 1.44, my model consolidated these odds into group of 1.39-1.49). I've tested only favorites odds. The resut (r2)was 0.97 for more than 1000 sample. Can you explain in a bit more detail how you carried out this test? I suspect you're not testing how accurate bookmakers' starting odds are, but how closely "actual win rates by group" correlates with "win rate by group if bookmakers' starting odds were fair", which, as Datapunter has tried to explain, are not the same thing at all. Quote Link to comment Share on other sites More sharing options...
Mulkis Posted April 25, 2011 Share Posted April 25, 2011 Re: Fair or not fair ? Can you explain in a bit more detail how you carried out this test? I suspect you're not testing how accurate bookmakers' starting odds are' date=' but how closely "actual win rates by group" correlates with "win rate by group if bookmakers' starting odds were fair", which, as Datapunter has tried to explain, are not the same thing at all.[/quote'] With pleasure. Glad someone finally got interested in it. Please note, that this excercise does not show us "true odds" or anything like that. It only shows us that correlation between bookies odds and actual results are very high. So, I've took historical odds for certain number of soccer matches (more than 1000). Only odds for favorites were tested. It implies range something from 1.10 to 2.80. This range was devided into sub-groups with +-0.05 deviation. Then for each group number of favorite actually won the match was calculated (Excel DSUM, odds range criteria -+0.05, True=1; False=0). This gives us a percentage how many times favorite actually won the match (x axis). And we already have y axis (odds sub-groups). We plot all this dada into Excel scatter chart. Then we apply trendline to the chart and get R2 value. It is close to 1. Quote Link to comment Share on other sites More sharing options...
slapdash Posted April 25, 2011 Share Posted April 25, 2011 Re: Fair or not fair ? Sorry, I'm still not quite sure I understand exactly what you did. You took the data for > 1000 matches. How many (x,y) values did you feed into the R^2 calculation? Was it one per match, or one per subgroup (with the x value determined by the percentage of wins within the subgroup)? Quote Link to comment Share on other sites More sharing options...
Mulkis Posted April 25, 2011 Share Posted April 25, 2011 Re: Fair or not fair ? One per subgroup: eg 1.01-1.10 subgroup, then 1.11-1.20 ... 2.70-2.80 subgroup and so on. 17 in total. Quote Link to comment Share on other sites More sharing options...
slapdash Posted April 25, 2011 Share Posted April 25, 2011 Re: Fair or not fair ? Hmm. I'm afraid I'm going out shortly, but I'll try to explain later this evening or tomorrow what I think is the flaw in that approach. Quote Link to comment Share on other sites More sharing options...
Mulkis Posted April 25, 2011 Share Posted April 25, 2011 Re: Fair or not fair ? Ok. I am more than happy to hear some critics on this:ok. Quote Link to comment Share on other sites More sharing options...
Datapunter Posted April 25, 2011 Share Posted April 25, 2011 Re: Fair or not fair ? Hmm. I'm afraid I'm going out shortly, but I'll try to explain later this evening or tomorrow Arghhh, nooooooooo, please the suspense is unbearable :lol Quote Link to comment Share on other sites More sharing options...
bacardirum Posted April 25, 2011 Share Posted April 25, 2011 Re: Fair or not fair ? Yeah its all about what you feel is value, like when chelsea were away to wolverhampton in the middle of their dire run, I thought Wolves at 9.00 or thereabouts was amazing value, they won 1-0, imo way overpriced simply cause the name Chelsea, not even taking into account how rank they were playing at the time. Quote Link to comment Share on other sites More sharing options...
ChrisB Posted April 25, 2011 Author Share Posted April 25, 2011 Re: Fair or not fair ? Meanwhile, Mulki is the person who have the best approache to determine value on this post. :) Quote Link to comment Share on other sites More sharing options...
Grex Posted April 25, 2011 Share Posted April 25, 2011 Re: Fair or not fair ? ........ Mulki is the person who has the best approach............. :) ...... but Bacardirum definitely has the best a*se ;) Quote Link to comment Share on other sites More sharing options...
bacardirum Posted April 25, 2011 Share Posted April 25, 2011 Re: Fair or not fair ? :rollin Mulki, its a combination of everything aint it, whats value to you isnt to someone else, the two most important things, the odds you are getting and the selection. Like wolves at home would not have been value at 9.00 against an unbeaten chelsea firing on all cylinders, so happened they werent, value. So if the odds are better than you believe they should be, and you are confident in the selections ability thats about it I guess, I am sure there are a lot more technical ways of looking at things if you so happen are into all that, but at end of day its those two things, and its only those two things that matter in the longrun, your strike rate and the ability to get the best prices. ...... but Bacardirum definitely has the best a*se ;) Quote Link to comment Share on other sites More sharing options...
slapdash Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? One per subgroup: eg 1.01-1.10 subgroup' date=' then 1.11-1.20 ... 2.70-2.80 subgroup and so on. 17 in total.[/quote'] I think I'm really just repeating what Datapunter has said, but the problem with this is that you're not comparing the bookmaker's odds with "actual odds", you're comparing them with "average actual odds" over a subgroup. Your statistical test shows that for teams in each of the 1.11-1.20, ..., 2.70-2.80, ... ranges, the bookie sets the odds pretty well on average, but this tells you very little about how accurate the odds are for an individual team. Suppose I looked at all the football matches taking place one week, randomly picked one of the three possible results (home win, draw, away win) for each one, and offered odds of 3.00 for that result. About one in three matches would have the result I picked, and all that a statistical test like the one you've used would see is that I've offered 3.00 and that 3.00 is pretty close to the average actual odds for all the cases where I offer 3.00, and would pat me on the back. What it doesn't see is that actually a lot of these should have been 1.25, 6.50, or whatever. More realistically, it could well be the case that a bookie's opening odds are very good on average within each of your subgroups, but some are overpriced and some are underpriced. And your test wouldn't see this. If the underpriced ones tended to be the ones where the price drifted (which sounds reasonable) then there's no reason to expect that betting when you can get a price better than the opening odds would be profitable, even though the bookie's opening odds passed your statistical test with flying colours. What you need to know is how accurate the bookie's opening odds are for teams that open at a given price and then drift, since these are the ones you're considering betting on. It's quite possible, even likely, that the bookie's opening odds are a lot less accurate on average for these than they are on average for all teams opening at a given price. Quote Link to comment Share on other sites More sharing options...
Mulkis Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? Yes, as I've said in the begining, this test does not show "true odds" or anyhing like that. It shows that opening odds quite well explain (are correlated) with acctual outcome of the event. Of course this is on average, I saw no reason to stress it, because it is r2 test, it is always on average. So, we mostly agree on all points, but... What is odds 1.50 for examle? This number (1.50) represents 66,67% probability of coresponding event to happen (eg. fav. wins). It implies that here still is a signifiant chance that this outcome will not happen (33% - Fav. is not a winner). And there is the main misperception from your side, imho. You interpret that this outcome (fav loses) means that 1.50 odds were not correct, that "true odds" should have been 2.00, 3.00 or something. But it is not the case - even with less probable outcome (fav is not a winner), the odds 1.50 for the looser (ex.fav) are still correct! Dont be fooled by numbers! Quote Link to comment Share on other sites More sharing options...
slapdash Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? So, we mostly agree on all points, but... What is odds 1.50 for examle? This number (1.50) represents 66,67% probability of coresponding event to happen (eg. fav. wins). It implies that here still is a signifiant chance that this outcome will not happen (33% - Fav. is not a winner). And there is the main misperception from your side, imho. You interpret that this outcome (fav loses) means that 1.50 odds were not correct, that "true odds" should have been 2.00, 3.00 or something. But it is not the case - even with less probable outcome (fav is not a winner), the odds 1.50 for the looser (ex.fav) are still correct! Dont be fooled by numbers! I'm not sure what I've written that suggests that's what I think? :unsure Quote Link to comment Share on other sites More sharing options...
Mulkis Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? Ok. If this is not the idea behind your (Slapdash) thinking, then could you please explain this your statement "...that actually a lot of these (odds) should have been 1.25, 6.50, or whatever." Quote Link to comment Share on other sites More sharing options...
slapdash Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? Ok. If this is not the idea behind your (Slapdash) thinking, then could you please explain this your statement "...that actually a lot of these (odds) should have been 1.25, 6.50, or whatever." I wrote that while talking about picking results (home win, draw, away win) completely randomly and assigning them odds of 3.00. My point was that on average 3.00 will be very accurate odds, but if you look at individual matches, I'll have picked some overwhelming favourites to win, where perhaps the odds should be 1.25, and I'll have picked some underdogs to win, where perhaps the odds should have been 6.50. I wasn't saying that the actual results determined the "fair odds". My point was that it is quite easy to assign odds that are fair on average without your odds on individual matches being anywhere close to fair. Quote Link to comment Share on other sites More sharing options...
Mulkis Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? My point was that it is quite easy to assign odds that are fair on average without your odds on individual matches being anywhere close to fair. But it is not easy. In contrary, it might be easy for 3,5,15 matches (by simply guessing), but not for 100,200,1000. And if you looking for a long term profit, you should look at a huge samples. And, btw, talking about individual selections, this R2 correlation test has nothing to do with them. For individual selections I use Poisson. Quote Link to comment Share on other sites More sharing options...
betting-place Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? I know that my next suggestion will not contribute much to this topic, but we may not forget that @2.00, no matter how you interpret these @2, for one's victory does not mean that 50% of the mathematical chance. We must bear in mind the margin set by bookmakers. Quote Link to comment Share on other sites More sharing options...
slapdash Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? But it is not easy. It is easy. And I've told you one very easy, if rather silly, way of doing it. In contrary, it might be easy for 3,5,15 matches (by simply guessing), but not for 100,200,1000. And if you looking for a long term profit, you should look at a huge samples. That's exactly backwards. The larger the number of matches, the easier it is to get a high "R^2 score" from the kind of test you've done. Going back to the post that started this thread ... As I understand it the idea was that if bookies' original odds are accurate, then there might be profit to be made by backing teams whose odds drift, on the theory that the original odds are likely to be more accurate. My point is that the way you've tested the accuracy of the original odds says virtually nothing about whether this strategy is likely to work. It doesn't matter how accurate bookies' original odds are on average for all home teams. What matters is how accurate they are on average for the teams whose odds drift and that we are going to back. Quote Link to comment Share on other sites More sharing options...
Mulkis Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? Ok. Enjoy your gambling. Quote Link to comment Share on other sites More sharing options...
ChrisB Posted April 26, 2011 Author Share Posted April 26, 2011 Re: Fair or not fair ? I have an interesting question (i hope) :) 1st example : Fair odd = 2.50 actual odd = 2.55 = Low value 2nd example : Fair odd = 2.50 actual odd = 3.00 = High value Probability to win is the same or not ? Quote Link to comment Share on other sites More sharing options...
Mulkis Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? Yes, if 2.50 indeed "fair odds". Quote Link to comment Share on other sites More sharing options...
Datapunter Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? Yes, if 2.50 indeed "fair odds". Quote Link to comment Share on other sites More sharing options...
ChrisB Posted April 26, 2011 Author Share Posted April 26, 2011 Re: Fair or not fair ? ok guys thank you, i have another interesting question (i hope again) :) we know fair odd : 2.50 . how to determinate the placing bet ? is 2% above 2.50 ? 5% ? 10% ? And if we start placing bets for example 7% above fair odds. It means that the ROI will be better than 2% above ? Quote Link to comment Share on other sites More sharing options...
Datapunter Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? slight correction first: we ASSUME the fair odds to be 2.50 And if we start placing bets for example 7% above fair odds. It means that the ROI will be better than 2% above ? Yes, PROVIDING the fair odds have indeed been determined correctly. Quote Link to comment Share on other sites More sharing options...
ChrisB Posted April 26, 2011 Author Share Posted April 26, 2011 Re: Fair or not fair ? ok thank you datapunter for your answer :ok and sorry for my bad english. :cry Quote Link to comment Share on other sites More sharing options...
Datapunter Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? language should never be apologized for :ok so how do you arrive at the conclusion a team's fair odds are 2.50 ? and how do you know that value to be correct ? Quote Link to comment Share on other sites More sharing options...
betting-place Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? ok guys thank you, i have another interesting question (i hope again) :) we know fair odd : 2.50 . how to determinate the placing bet ? is 2% above 2.50 ? 5% ? 10% ? And if we start placing bets for example 7% above fair odds. It means that the ROI will be better than 2% above ? As far as I know Kelly formula would be fine for what U asked for! But U can also use simple system,Im using this one for taking local gifts. For example U have money edge for staking,for each bet.Lets say 300 and then U make math: 300/2.50=120 stake, and your real bet looks: 120*3.00! If your fair odd is correct U ll win 360, which is 60 more than 300, or in math that is 20%, and your value is 20%! Quote Link to comment Share on other sites More sharing options...
betting-place Posted April 26, 2011 Share Posted April 26, 2011 Re: Fair or not fair ? language should never be apologized for :ok so how do you arrive at the conclusion a team's fair odds are 2.50 ? and how do you know that value to be correct ? And if is that possible at all? Quote Link to comment Share on other sites More sharing options...
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.