Fair or not fair ?

[ChrisB]

Hello everyone, I read some systems to determinate valuebets. But why it's not interesting to trusth books about odds calculating. bookmaker use powerful algo for major leagues, their estimations are correct. Ok, books adjust a little pre-opening odds to equlibrate the future weight of money. But apart this, the opening odds of majors bookies are fairs. an example to determinate value bet. average books opening odds : 2.50 - 3.20 -2.50 (36% - 28% - 36%) average books closing odds : 2.20 - 3.40 - 2.80 ( 41% -26% - 32%) value should be 2.80 (+8.9% from the opening odd). In conclusion, fair estimation can be found with books opening odds. This is my point of view. I would like to know what do you thhink guys ? Sorry fo my english. Thank you for your comments and point of views. :)

[v-zero]

Re: Why values bets are not determined like this ? Not all odds are fair, the assumption that the bookies are amazing at calculating fair odds is inaccurate.

[Lee26]

Re: Why values bets are not determined like this ? 2/9 for arsenal to beat blackburn at home a few weeks back didnt seem fair, but i went for it anyway . . . 0-0! see your point, they can have their moments but i do like the exchanges if i can get money onto them!

[ChrisB]

Re: Why values bets are not determined like this ?

Not all odds are fair' date=' the assumption that the bookies are amazing at calculating fair odds is inaccurate.[/quote'] Hello V-zero, I suppose that a major book use the same algo for EPL and i don't see how this major book can calculate wrong odd in opening for Blackburn and not for Newcastle on other match. And what determine 100% secure that the bookie has made a bad calculation ?

[v-zero]

Re: Why values bets are not determined like this ?

Hello V-zero, I suppose that a major book use the same algo for EPL and i don't see how this major book can calculate wrong odd in opening for Blackburn and not for Newcastle on other match. And what determine 100% secure that the bookie has made a bad calculation ?

As far as what determines completely fair odds, that's the dream ticket, but it's difficult to attain. However, I have come up with a system for draw no bet odds which in backtesting has produced between 15% and 55% ROI across all leagues available on football-data applied to average odds from betbrain.

[Mulkis]

Re: Why values bets are not determined like this ?

I would like to know what do you thhink guys ?

This is exactly what I've been contemplating and going to papertest in near future.

[ChrisB]

Re: Why values bets are not determined like this ? do you think this is a good approaches or not ? what are your arguments ? thanks ;)

[Mulkis]

Re: Why values bets are not determined like this ? My main asumption is that bookies' starting odds are almost "true" (of course making account for overround). This asumption could be easily checked by statistical test and R2 is something around 0.97 I dont think that enyone could do better. Then starting odds are somehow distorted by the market. Teams news, fans money, etc. Bookies try to balance their books and this makes odds to drift. This drift is not statistically grounded, this is something like "madness of the crowd" which distorts prices. I am mostly on financial markets and always trying to exploit this emotional drifts (like stock crash after Japan earthquake). Emotions, not data or facts mooves the markets. So, I assume, there might be some "value" in distorted odds. This is just an theorethical assumtion, it is not checked for now. In paralell I do my own simple Poisson distribution just to have a "benchmark" for "real odds". But I will never be able to beat sophisticated models that bookies use, so I think it is more viable to exploit unefficiencies of the market if there is any.

[ChrisB]

Re: Why values bets are not determined like this ?

My main asumption is that bookies' starting odds are almost "true" (of course making account for overround). This asumption could be easily checked by statistical test and R2 is something around 0.97 I dont think that enyone could do better. Then starting odds are somehow distorted by the market. Teams news, fans money, etc. Bookies try to balance their books and this makes odds to drift. This drift is not statistically grounded, this is something like "madness of the crowd" which distorts prices. .

its not necessary to beat shophiticated bookie models because we use this model (starting odds). Does it mean if all starting odds goes up better than 3% are values ?

[Mulkis]

Re: Why values bets are not determined like this ? Dont know, really. I would say drif 10% is absolute minimum to start looking at. But I just guessing for now.

[Datapunter]

Re: Why values bets are not determined like this ? Bookmakers do not ( or should not ) predict odds based on the most likely winner of a match. They (should) set odds in accordance to how they think their betting customers are going to bet, then apply their custom overround and thus, if they get it right, make the most of that built-in overround. In fact they couldn't care less who is most likely to win a match as by the time the match kicks off they have already made their profit and it doesn't actually matter who wins. Which is one of the reasons value exists. Different bookmakers have different type of clients and therefore their expectation of how THEIR group of clients is going to bet can be different therefore they set different odds. With some bookmakers it is common to set really good odds on something in order to attract clients and manage the situation by severely limiting the stakes. Other difference for example are bookmakers that are exclusively on-line attracting a modern crowd and those that have brick-and-mortar shops which have a much more traditional type of punter. Those bookmakers that do not operate that way are not bookmakers but punters themselves. Whenever a bookmaker goes bankrupt, and they do from time to time, look at the causes. A losing bookmaking company does so because they take chances same as a punters do. If a bookmaker has a lot much money coming in on a particular selection he can do a number of things. First adjust the price and hope to attract more cash on the other selections, limit the amount coming in on that selection, lay some of it off with other bookmakers or exchanges. Fairly easy to manage when you got an on-line business only, bit more difficult with brick-and-mortar shops but still possible as the flow of money is more steady than abrupt. Failing to do that is taking chances and in the traditional sense of "running a book" it is bad management. Also remember that by setting the odds the bookmakers actually direct the betting. The general betting public only have a superficial sense of value. They will bet on something because of the offered odds. Let me say that again: they bet on something because of the offered odds. Odds set by bookmakers, with the purpose of making money from their overround, regardless of an events outcome. Think about it. Now the question is how often are the odds based on expected betting patterns and the odds based on an event "true" odds the same, or as close as it gets.We shall never know as neither one can be determined with accuracy. This is all greatly based on speculation of course. I don't know how bookmakers run their business. My purpose here is simply to add this view point to the overall subject, food for thought. Assuming for a moment i'm right here, where do we go from here ? The search for value should then be based on 2 things. First the odds as determined by the general betting public. That becomes apparent when the markets have been open for some time. I.e. NOT the opening odds. The thing to look at is the price movements with the bookies from opening odds to current odds as that indicates where they've had to adjust based on money coming in. Exchanges, but here one needs to take into consideration the amount of volume matched. Too early can be misleading as there's hardly any significant volume matched. Too late can also be misleading as very close to kickoff you get a lot of trading volume ( and bookmakers laying off large amounts ) Here you can do research and determine of accurate these odds are. In horse racing for example there is documented research that says clearly the SP as determined by the workings of the markets is in fact a fairly accurate representation of a horse's chance of winning. The same research can be done for other sports. The tricky bit being what odds are you examining. The second is your own perception of the true odds for an event. The bookmakers take money on all selections and try to make a profit from the overround and therefore are done by kickoff. We punters however usually bet on only one outcome and therefore we are dependent on the actual outcome of events, Compare those 2 and you will find value, or not.

In conclusion, fair estimation can be found with majors books opening odds

Conclusion, i disagree but at the same time can't completely dismiss it. You'll all have to make your own minds up :moon

[ChrisB]

Re: Fair or not fair ? I completely agree with you database. agree also, that bookmakers adjust their pre-opening odd in function of their customers. But according also that their adjustements odds is not so far than the fair odd isn't it ?

[Datapunter]

Re: Fair or not fair ? Let's take a moment to revise the definition of "fair odds", fair odds are those odds that reflect a selections true chance of winning it's event. Considering there is a substantial number of aspects that cannot be determined, collectively referred to as luck, such as the weather or if a person got laid the night before the event or not (and with whom) true odds can never actually be determined. The best one can hope for is an approximation based on those aspects which can with some degree of accuracy be quantified. Follows a very simple problem. If you determine than a team starting a match at odds of 2.00 has, based on historical data, a 50% chance of winning can you say the "fair odds" for that team are indeed 2.00 ? No, all you can say is that on average any team starting at 2.00 has a 50% chance of winning. But that does not say anything about the individual team you are looking at. The individual team you're looking at may have fair odds of 1.80 or 2.20 or whatever. Even if the bookies odds are an accurate reflection of a teams winning chance, based on a large sample of historic data, they are only a reflection of ANY teams winning chance. Which may or may not be applicable to the individual team you are looking at. A horse starting a race with an SP of 2.00 has a 50% chance of winning. Documented. That does not mean every individual horse starting at 2.00 has a 50% chance. One may have 40% another may have 60%, on average it is 50% So to determine value you can reference this data as part of your research, but the only actual measure is still your own perception of an individual chance of winning and from that your own "fair odds".

[ChrisB]

Re: Fair or not fair ? yes datapunter, you are right :) But: why our perception, our method of calculation would it be more powerful than bookmaker ? or Is there a method of calculation may be more accurate than the bookmakers method of calculation ? never found answer , but it's crucial. my point of view : bookmakers know better than anyone calculate the odds. it is their job after all thanks

[Mulkis]

Re: Why values bets are not determined like this ?

Now the question is how often are the odds based on expected betting patterns and the odds based on an event "true" odds the same, or as close as it gets.We shall never know as neither one can be determined with accuracy. <...> In horse racing for example there is documented research that says clearly the SP as determined by the workings of the markets is in fact a fairly accurate representation of a horse's chance of winning. The same research can be done for other sports. The tricky bit being what odds are you examining.

Datapunter, what exactly prevents you to do you own simple statistical test on soccer starting odds (SP)? I've made one and the result is astonishing R2=0.97. So, this is "in fact a fairly accurate", isnt it? I agree, that bookmakers mostly are concerned about balancing their books, that they initially (by releasing starting odds) are trying to guess market reaction to their prices, not an outcome of the event. But this endevour somehow leeds very close to "true odds". And later on the market "distorts" the odds and drifting accours. The question we trying to answer in this post is how much this deviation could be exploited for profitable betting (long term!). And "long term" I mean 300,500,600 bets.

[Datapunter]

Re: Fair or not fair ?

So, this is "in fact a fairly accurate", isnt it?

Yes it is accurate, but it is not the fair odds we are looking for. The odds are not the "fair odds" for an individual selection, that is my point. The odds you obtain that way are the fair odds ON AVERAGE, not for an individual. So the conclusion that the opening / betfair / market odds are a fair representation of a team's chance of winning is NOT correct. A team starting a match at odds of 2.00 will on average win 50% of the time, but only on average. The "fair odds" reflect the average, theoretical, team's winning chance. That does not mean the individual team you are looking at have 2.00 as fair odds. The individual team you are looking at can have 1.50 or 3.00 as fair odds. Given 10000 matches then on average it works out to 2.00, so what. You need to work out for an individual team it's fair odds, and compare them to what's available in the market, if the markets has 2.00 and you determine it should be 1.50 then back them, if the markets has 2.00 and you determine it should be 3.00 then lay them.

[ChrisB]

Re: Fair or not fair ? Datapunter, If the starting market has 2.00. This is 2.00. Why try to determine another odd ? i don't understand...

[Datapunter]

Re: Fair or not fair ? Tonight there's Blackburn vs Manchester City, currently the odds on Betfair are 2.02 for Man City ( lets say 2.00 for arguments sake ) Let's assume for a moment that these Betfair odds are accurate, a reasonable assumption. Does that mean Man City has a 50% chance of winning this match ? NO It simply means that on average any team starting out at 2.00 will win 50% of the time. It does not mean Man City have a 50% chance. Their chance could be 40% or 60% and that is exactly what we try to determine, if one chooses to do "value betting". Ok, common sense tells you ( me at least ) that it is much more likely that the fair odds for Man City are indeed 2.00 than any other value. But more likely does not mean it is so. You cannot say Man City's fair odds are 2.00, they are most likely quite close to 2.00 but that remains to be determined by a method that determines an individual team's chance, not an average team's chance.

[Godders]

Re: Fair or not fair ? Interesting thread here. I'll reserve judgement on what fair odds are for the time being (mostly because when I think I've found fair odds the market goes in the other direction!) I would be interested to know though Mulkis, if you've done an R2 test on odds at kick off as opposed to the opening odds? Also, (and statistical maths isn't my strong point, so I may be completely wrong), but doesn't an R2 test show what chance there is that winners selected by a system is down to chance or fluke, and does not measure at all whether the odds of a selection are value (and if so by how much), or fair?

[Datapunter]

Re: Fair or not fair ?

If the starting market has 2.00. This is 2.00. Why try to determine another odd ?

If you assume the market odds of 2.00 to always be correct then value cannot exist.

[ChrisB]

Re: Fair or not fair ?

If you assume the market odds of 2.00 to always be correct then value cannot exist.

yes of course value can exist. odds of 2.00 is the primary odd, and we assume team have 40-60% to win. if the final odd is 2.6. then there is a value. ok it's not a good example because 2.00 to 2.6 is rare. another example with odds of 4.00 is the same thing. really, i think that bookmaker do a good job with their powerful algo. Why we must challenge this work ? its a nonsense.

[Datapunter]

Re: Fair or not fair ? Have you actually understood a single word i've said in this thread, i'll wish you good luck and just leave it at that.

[v-zero]

Re: Fair or not fair ?

Yes it is accurate, but it is not the fair odds we are looking for. The odds are not the "fair odds" for an individual selection, that is my point. The odds you obtain that way are the fair odds ON AVERAGE, not for an individual. So the conclusion that the opening / betfair / market odds are a fair representation of a team's chance of winning is NOT correct. A team starting a match at odds of 2.00 will on average win 50% of the time, but only on average. The "fair odds" reflect the average, theoretical, team's winning chance. That does not mean the individual team you are looking at have 2.00 as fair odds. The individual team you are looking at can have 1.50 or 3.00 as fair odds. Given 10000 matches then on average it works out to 2.00, so what. You need to work out for an individual team it's fair odds, and compare them to what's available in the market, if the markets has 2.00 and you determine it should be 1.50 then back them, if the markets has 2.00 and you determine it should be 3.00 then lay them.

Don't really understand this post to be honest. If the Betfair odds are 2, and they are fair, they are accurate odds, then the chance that team will win is 50%. The key to betting is finding those odds that are better than fair.

[Mulkis]

Re: Fair or not fair ?

I would be interested to know though Mulkis, if you've done an R2 test on odds at kick off as opposed to the opening odds? Also, (and statistical maths isn't my strong point, so I may be completely wrong), but doesn't an R2 test show what chance there is that winners selected by a system is down to chance or fluke, and does not measure at all whether the odds of a selection are value (and if so by how much), or 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.

[Datapunter]

Re: Fair or not fair ?

If the Betfair odds are 2, and they are fair,

This is the whole misconception in this thread. If they are indeed "fair odds" for that team then yes the chance is 50%, but who says they are the fair odds for that team. You don't know they are the "fair odds", that is the whole point. And it is a wrong assumption to say they are 2.00 / 50% because the odds are accurate. The odds may be accurate on average, but that does not mean they are accurate for that individual team.

[Mulkis]

Re: Fair or not fair ?

This is the whole misconception in this thread. The odds may be accurate on average, but that does not mean they are accurate for that individual team.

1. I dont think Betfair odds are valid example for this topic - I've been talking about opening odds of biggest bookies like bet365, pinnacle, etc. 2. If the odds are "accurate on average" in the longer run you will break even. Consequently, if these "accuracy on average" is exploited correctly, you will be in profit in the long run. EDIT: and there seems to be one more misunderstanding: by "opening odds" I mean early odds presented by bookie a few days befor he event, not at the start of the event. This is major difference in odds!

[Datapunter]

Re: Fair or not fair ?

if these "accuracy on average" is exploited correctly, you will be in profit in the long run.

Exactly, but to do that you need to determine your own "fair odds" for each individual selection, otherwise how will you know when to back or lay. If you assume those average odds to be the "fair odds" then you are basically saying there can never be any value. So to get back to the original question: do the opening odds reflect "fair odds" ? No they may reflect accurate odds on average, but not individual "fair odds".

[davidc]

Re: Fair or not fair ? In my opinion one persons (or bookies) estimation of odds is as good as any others even though they might price up the game differently and he only way to find if the odds are fair, or indeed if they provide value, is to compare the odds on a long term basis and assess where the value exists by comparing the odds which you perceive as fair (or provide value) against the best bookies odds. The down side to this is that you have no control over how the bookies (or other people) assess each game and so although you might maintain the same process to select your odds they may not be directly comparable over time. Value is in the eye of the beholder!

[Mulkis]

Re: Fair or not fair ? But lets us go back to the primary idea of this post: if bookies starting odds are in general correct, then any further deviation from those odds should represent potential value. This is something need to be checked, but still the idea in itself looks quite valid.

[Datapunter]

Re: Fair or not fair ?

But lets us go back to the primary idea of this post: if bookies starting odds are in general correct, then any further deviation from those odds should represent potential value.

Correct. But in order to be practically applicable you need to address 2 problems. First the deviation will needs to be big enough to overcome the built in overround. (setting aside a moment the costs of constantly transferring cash from bookie to bookie) In itself not a big problem, a simple matter of only betting if the difference is above a set threshold. But then you also need a strike rate big enough to make a profit. Here we're back to determining value on the basis of your perceived "fair odds". Imagine for a moment you got a match with a team starting at 2.00 and you can get a price with a bookmaker at 2.10 . Wahey. But what if the "fair odds" for that individual team are actually 2.20 ? That fact will result in a strike rate lower than expected and therefore there will be no profit at the end. 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.