Just found a study about what a lot of people seem to do nowadays. Compare lines of bookies, devig them then bet if one is much more different. But i thought it was interesting to read
arxiv.org/pdf/1710.02824
Scientific study about comparing odds lines
-
arb12
- Totally Pro
- Karma: 36
Post
When an observed event becomes an academic study (whether it is high-level aviation/space science, e.g. Sergey Korolev/ Igor Sikorsky/ Konstantin Tsiolkovsky or even a Mathematical Method for predicting the most likely lines in Asian Handicaps in Soccer / or even for compiling true ELO of Batsmen & Bowlers and building a reliable Mathematical/Statistical covered Forecast in Cricket), almost all mysteries are largely solved and the connections between natural phenomena and the forces behind the object are discovered, hence the probabilities of X or Y happening are known (at least within a calculable interval) and are far from a mysterious degree of “luck”.
Agencies and the odds compilers know this and try to reduce the odds offered below the calculated probability of event Z happening. I mean they just try to do this, I’m not saying they can calculate objective real odds all the time.
The following Hypothesis arises: is it possible for an independent market participant to independently develop a series of Math/Stats-driven Models from scratch? That is why these kinds of articles are too incomplete, no one will reveal the deepest details of their work...
Re: Scientific study about comparing odds lines
When an observed event becomes an academic study (whether it is high-level aviation/space science, e.g. Sergey Korolev/ Igor Sikorsky/ Konstantin Tsiolkovsky or even a Mathematical Method for predicting the most likely lines in Asian Handicaps in Soccer / or even for compiling true ELO of Batsmen & Bowlers and building a reliable Mathematical/Statistical covered Forecast in Cricket), almost all mysteries are largely solved and the connections between natural phenomena and the forces behind the object are discovered, hence the probabilities of X or Y happening are known (at least within a calculable interval) and are far from a mysterious degree of “luck”.
Agencies and the odds compilers know this and try to reduce the odds offered below the calculated probability of event Z happening. I mean they just try to do this, I’m not saying they can calculate objective real odds all the time.
The following Hypothesis arises: is it possible for an independent market participant to independently develop a series of Math/Stats-driven Models from scratch? That is why these kinds of articles are too incomplete, no one will reveal the deepest details of their work...
- niel
- Gaining experience
- Karma: 6
Post
I'm pretty sharp on certain markets: politics, economics, culture, ... Usually prediction markets but i don't know how to model. I just do deep research till i have a pretty good idea about how likely something is to happen. I just don't know much myself about any ball sports, i used to suck at sports back in school. I was only good at running, i'm quite fast.
The only sports i want to get deeper into myself are boxing and ufc.
Would love to get deeper into how modeling works. I'm just more a guy that would try to model politics then soccer or something.
Re: Scientific study about comparing odds lines
But what does that have to do with it? This is based on the efficient market hypothesis. Markets are on average pretty accurate. That is why if you take the average market price, devig it and find a much better deal somewhere at some bookmaker (that likely made a mistake), you make money on the long term. What you describe is quite opposite, you assume markets are wrong and try to predict with an own model.arb12 wrote: ↑Thu May 01, 2025 2:02 pmThe following Hypothesis arises: is it possible for an independent market participant to independently develop a series of Math/Stats-driven Models from scratch? That is why these kinds of articles are too incomplete, no one will reveal the deepest details of their work...
I'm pretty sharp on certain markets: politics, economics, culture, ... Usually prediction markets but i don't know how to model. I just do deep research till i have a pretty good idea about how likely something is to happen. I just don't know much myself about any ball sports, i used to suck at sports back in school. I was only good at running, i'm quite fast.
The only sports i want to get deeper into myself are boxing and ufc.
Would love to get deeper into how modeling works. I'm just more a guy that would try to model politics then soccer or something.
- Meise
- Gaining experience
- Karma: 6
Post
But there is no reason why someone, or a group of people, cannot independently do the same thing, assuming they have enough knowledge and practical experience with statistical modeling.
Re: Scientific study about comparing odds lines
And why would it not be possible? I mean, when the bookmakers first started to compile odds for a specific sport, they had to develop an independent model from scratch. Well, maybe they took something that someone developed for a different sport and adapted it, or they read an article describing a statistical model that could be used to model outcomes in this sport and implemented it, etc...is it possible for an independent market participant to independently develop a series of Math/Stats-driven Models from scratch?
But there is no reason why someone, or a group of people, cannot independently do the same thing, assuming they have enough knowledge and practical experience with statistical modeling.
-
arb12
- Totally Pro
- Karma: 36
Post
That's exactly what I meant when I asked that rhetorical question in my previous post.
With this I defend my thesis that the Authors of these materials will never publicly disclose categorically extremely useful information, which they have delved into creatively (assuming that they have come to it), thus having a significant lead over the mass market participant.
Whether the newly discovered Method (not mentioned in these articles, of course) is at a mature stage, ready to attack the markets, etc. is a completely different question, but you will never hear about it publicly from anyone, such things are not intended either for wide discussion or for sale. Officially they do not exist, but many are working on them.
Example: let's assume that an independent market participant independently tests its own series of non-standard developments, with a small number of them failing, a large number of them being mediocre, but some small number of these unusual methodologies, at a relatively acceptable risk, have withstood a huge number of backtests and over a long period of time have shown somewhat acceptable results. From now on, this researcher will only optimize in the next X years from now and model, hoping to capture new and more subtle Statistical dependencies, processed with a non-standard, specially adapted Mathematical apparatus.
We hypothesize that such an independent participant could operate with a 7-digit amount, reasonably distributed among counterparties in Asia and other places where legal entities realize a total turnover of billions daily and the participant's insignificant turnovers are not rejected by default.
Example 2: An investment fund specializing in sports markets manages a large capital. Naturally, it works with hired mathematicians and programmers, but will the most valuable developments really be shared by the skilled private individual with the fund?
Re: Scientific study about comparing odds lines
Meise wrote: ↑Mon May 05, 2025 8:46 amAnd why would it not be possible? I mean, when the bookmakers first started to compile odds for a specific sport, they had to develop an independent model from scratch. Well, maybe they took something that someone developed for a different sport and adapted it, or they read an article describing a statistical model that could be used to model outcomes in this sport and implemented it, etc...is it possible for an independent market participant to independently develop a series of Math/Stats-driven Models from scratch?
But there is no reason why someone, or a group of people, cannot independently do the same thing, assuming they have enough knowledge and practical experience with statistical modeling.
That's exactly what I meant when I asked that rhetorical question in my previous post.
With this I defend my thesis that the Authors of these materials will never publicly disclose categorically extremely useful information, which they have delved into creatively (assuming that they have come to it), thus having a significant lead over the mass market participant.
Whether the newly discovered Method (not mentioned in these articles, of course) is at a mature stage, ready to attack the markets, etc. is a completely different question, but you will never hear about it publicly from anyone, such things are not intended either for wide discussion or for sale. Officially they do not exist, but many are working on them.
Example: let's assume that an independent market participant independently tests its own series of non-standard developments, with a small number of them failing, a large number of them being mediocre, but some small number of these unusual methodologies, at a relatively acceptable risk, have withstood a huge number of backtests and over a long period of time have shown somewhat acceptable results. From now on, this researcher will only optimize in the next X years from now and model, hoping to capture new and more subtle Statistical dependencies, processed with a non-standard, specially adapted Mathematical apparatus.
We hypothesize that such an independent participant could operate with a 7-digit amount, reasonably distributed among counterparties in Asia and other places where legal entities realize a total turnover of billions daily and the participant's insignificant turnovers are not rejected by default.
Example 2: An investment fund specializing in sports markets manages a large capital. Naturally, it works with hired mathematicians and programmers, but will the most valuable developments really be shared by the skilled private individual with the fund?
- A2format
- Gaining experience
- Karma: 2
Post
Re: Scientific study about comparing odds lines
This is a 2017 article, and at that time I even downloaded their database (it was freely available, maybe it is now) for 500,000 matches. However, the database contains bets from almost only softbookmakers such as William Hill, Bet365, Betahome, Interwetten, etc. As you know, softbookmakers will not allow you to win, so the research is interesting, but it does not work in the long run.
-
