pinkisntwell
-
Posts
150 -
Joined
-
Last visited
Posts posted by pinkisntwell
-
-
Re: Stakes from Betfair Many people have tried playing like this and all of them have failed.
-
Re: Estimations based on Odds Data of local book
If you already have some mathematical knowledge (maybe you studied engineering or something similar) then I suggest you try an introductory statistics & probability book, like http://books.google.com.cy/books/about/Probability_statistics.html?id=HQ3vAAAAMAAJ&redir_esc=y But be advised that sports betting markets are extremely efficient. No matter what you do, no matter how advanced knowledge you have, it's still an extremely remote possibility that you'll be able to make a living out of sports betting. In previous posts I have proved (for certain interpretations of the verb "prove") why living on sports betting is impossible. I myself have built numerous models and the only thing I got out of them was improved statistical knowledge.Yes i am not sure if i would be able to follow something like that.... -
Re: Estimations based on Odds Data of local book
Only forward testing can determine performance. Do yourself a favor and never bet using a system that has only been backtested otherwise you'll find out what christos did: The system will start underperforming the moment you start betting it.No I am saying a model built correctly will not overfit. So back testing is a good thing to do to determine performance. -
Re: Estimations based on Odds Data of local book
So you're basically saying that a model has not an issue with overfitting if it has not an issue with overfitting. Nice.Its a broad and relatively vague definition. But basically it means soundly built model that is sufficiently powerful whilst not overfit. -
Re: Estimations based on Odds Data of local book
What is a "robustly built model"?Overfitting is not an issue with a robustly built model and not really a consequence of backtesting. -
Re: Estimations based on Odds Data of local book
For the simple reason that when looking back you already know the results of the matches. Your systems need to be tested out-of-sample, on matches that were not used during their training. Otherwise this occurs: https://en.wikipedia.org/wiki/Overfitting.Can you explain why you think backtesting is worthless? -
Re: Estimations based on Odds Data of local book
One approach is to model each team's merit as a probability distribution and use your data to infer the parameters of said distribution. I have some research that might interest you at http://www.statsfair.com/ What's your background? If you are not well versed in mathematics then I'm afraid it will be difficult for you to follow any in-depth discussions on the matter.Thanks pinkisntwell. What do you really mean by statistically-based system ? What i am doing is not a statistically-based system ? Can ypu please explain it more -
Re: Estimations based on Odds Data of local book
This kind of analysis will get you nowhere. The only positive fact is that you seem to recognize that backtesting is absolutely worthless and it can cost a lot of money. The data you have is very valuable, but only to be used as part of creating a statistically-based system.I try to explain a little more the system...I have odds database from greece local betting book for all leagues since 2000... I have also data from other sportsbooks (bet365, bet-at-home) but are not too many (only since 2008 and not for alla leagues) so thats why i am using only the greek book for all games. Also the greek sportsbook doesnt change odds too often. Over the years i was wondering what kind of analysis would give the better estimations to predict future games. I have tried any possible combination with so many backtesting. But when i really started to try it betting always went bad. Now i am sure that to see the odds combinations (example to see what happened to previus games with the same 1x2 odds on the same league) is nothing more than a hole in the water. Also looking at the favorites odds is under questions because sportbooks bring down favorites odds so much. So here i try to see what the outsider did in previous games with the same odd. I will give you an example of one of my estimations posted already. Lets take the game Ponte Preta - Santos. I the database if you see what santos did in previous games with the particular odd offered here in greece (Santos is given as an outsider) you can see that santos had 8/9 games both teams to score since 2000. Looking at the Vestsjaelland - Aarhus game Vestsjaelland won 6/7 games when offered at this odd as an outsider. I really dont know what will happen so thats why i am little restrained and use also the draw on the predictions (DC, or 0 AH) -
Re: Back tested system US Open 12 % yield, 12 years
but that's no longer back testing.Totally agree with Methuselah. I would imagine that the majority of systems posted on this forum were developed using back-testing (including my "Homes" and "Overs" systems). Only by back-testing can you establish "value" bets..... and if you can't establish theoretical "value", you will almost certainly lose in the long-term. Regarding your system, Gonza3591, you say you developed your cut-off criteria based on the last 12 years results (2000-2012??). I suggest you do your analysis on the results from 2000-2010, and then test it "live" on the results for 2011-2012. If the yield for the "live" testing is consistent with the previous theoretical results, then you may be onto a winner. If there is a sudden drop in performance for the "live" testing, then you have a problem. -
Re: Back tested system US Open 12 % yield, 12 years Back testing is absolutely worthless, since you already know the results of the matches. Do yourself a favour and don't bet this system.
-
Re: Facebook accounts/sites with "fixed games, 100%" Why do you think it's a scam? More often than not it's easy to spot a fixed match if you watch the betting action. UEFA spots hundreds of such matches every year just by watching the betting action.
-
Re: Tennis and snooker statistical rankings and ratings Just added a page that lets you select two players to calculate the win odds. Example:
Find it at http://www.statsfair.com/predict -
Re: Tennis and snooker statistical rankings and ratings
I've moved your thread here' date=' hope you don't mind - I'm sure you'll get much better response from guys who regularly follow tennis.[/quote'] I don't mind, as I see that the link has remained in the Systems & Strategy forum. "Time navigation" goes forward or backwards in time by the selected number of weeks. 260 is not the default, it's just the highest number of weeks that can be selected at once. I have 4 weeks, 26 weeks, 52 weeks and 260 because they correspond roughly to one month, half a year, one year and five years, respectively. You are absolutely correct about the "Race to" option, I need to explain it somewhat better. -
I know there are plenty of people here that have undertaken research on statistical rating methods (Elo, Glicko, etc.) So have I, and I have made available some of my findings here: http://www.statsfair.com/ . There's tennis and snooker for now, and more will be added as suitable data is found. I would be interested in getting some feedback on the ratings and the website in general. Has anyone had luck following such systems? What were the problems you encountered during your research? The site is pretty bare-bones so far, so what would you like to see added?
-
Re: The reality of sports betting/trading for a living.
I am one of these people. You are absolutely correct about the house having the edge and meaning you can never win in the long run. you are wrong however in thinking that the house has the edge with every sport and every bet type. I fell very ill a few years ago and like pretty much everyone on this site I decided to dedicate my time trying to find a way of consistently making money. I was bed bound for around a year and spent all day every day trying to achieve this 'dream'. After that long experimenting with every sport and every staking plan I came across my own strategy for sport betting which I have yet to come across on this site and I now make a living from gambling (a very good one at that). The route of it comes down to' date=' as you say, having an edge on the house on every single bet. I have been discouraged from ever telling people what this is and I will not do so now but I will say that it is out there. No one may ever read this post but if you do. There is a sport on traditional bookmakers and within that sport there is a way of consistently beating the house and if you want to find it you will.[/quote'] Can you give a hint on that strategy of yours? What sport is it about? Does it involve a computer program or can it be done by hand? How many bets per week does it give? -
Re: Historic Data Questions & Model Development
Split your data into training and testing segments. Fine-tune your system until you have reason to believe it works. Then apply to your testing segment and see how it stacks up.Yeh understand that. Just getting used to working with this type of data, writing an initial algorithm to calculate the ratings, backtesting and comparing odds. Any suggestions re potential improvement? I plan to have a play tomorrow with things such as different # games used to calculate the GS, weighting more recent games, capping outliers and so on. Cheers. -
Re: Historic Data Questions & Model Development
ratings.pdf it's just a simple example. Don't expect to get any money out of it.Unspectacular back test results of: [TABLE=width: 128] [TR] [TD=class: xl70, width: 64]2012/13[/TD] [TD=class: xl71, width: 64]0[/TD] [/TR] [TR] [TD=class: xl70]2011/12[/TD] [TD=class: xl71]22[/TD] [/TR] [TR] [TD=class: xl70]2010/11[/TD] [TD=class: xl71]9[/TD] [/TR] [TR] [TD=class: xl70]2009/10[/TD] [TD=class: xl71]29[/TD] [/TR] [TR] [TD=class: xl70]2008/09[/TD] [TD=class: xl71]-3[/TD] [/TR] [TR] [TD=class: xl70]2007/08[/TD] [TD=class: xl71]-32[/TD] [/TR] [/TABLE] Thinking of adding a weighting parameter to more recent games and capping outliers (Arsenal's 8 for example this season). Also want to account for an away team factor. Thoughts and comments welcome! -
Re: The reality of sports betting/trading for a living.
Thanks' date=' so should you have a level return plan instead?[/quote'] Yes, this way you make better use of your bankroll. -
Re: "Sports-Betting as a Business" strategy
Bookmakers don't set up odds, the public does.Key of my method is that I do keep track of real form of the team. Sometimes one team play a decent game but lose the game in some mysterious unlucky accident. Bookmakers are not sophisticated enough when they make lines for this hundreds of bets every week. They take rather widely available stats and set up odds. -
Re: The reality of sports betting/trading for a living.
The lower the odds, the higher the stakes.When I googled proffesional gambler I only scanned the pages and did not realise they offered tips or how much these tips may lose. I have been using tips online at ****, EasyOdds and SelectaBet and offline buying the Racing Post, Racing and Football Outlook and Racing plus, and with mixed success on sports betting. I am looking at tipster perfomrmance websites which show what longtime ROI these people acheive as most do not publish this with their tips. Out of interest why would you not back level stakes? I have not bet level stakes in the past but have had bad money management betting 25-30% of my bank on one bet and then loosing and then having to bet the same again to win it back. My worry long term is "money management" and would level stakes not help me with this? What other method would you suggest? -
Re: Systems and backing at true value
The math of this is very easy and it's not open to discussion. The expected profit per unit staked is equal to: ( (odds taken) / (real odds) ) - 1 For example, if a bet has a 50% chance of winning (i.e. real odds of 2.0) and you manage to bet at 2.1 then your expected profit per unit staked would be 2.1/2 -1 = 0.05 units.Interested in FF's response too. Isn't this the definition of value? i.e. 10:1 on a 5:1 shot? -
Re: Historic Data Questions & Model Development
First of all, you can forget about line shopping. This is because if you are a winning player you will be banned from all bookies apart from Pinnacle and the exchanges.Hi Pink, I'm not sure, that's why I'm having a go :) - I believe good line shopping combined with an estimation of the "true" line might work?
I don't think it is possible for a piece of software to outsmart thousands of punters. Even if it was though, the software would have to very complex and it would have to be self-modifying. The money you can get from sports betting would not compensate for the costs of maintaining such software. If there was value in something like this there would be companies doing it already.That is a really interesting point and one that I hoped we might discuss in this thread - is it possible to beat the market using qualitative analysis mentioned in your post? -
Re: Historic Data Questions & Model Development
I believe it should go without saying that the system should take into account all relevant information, and not merely team names and results. In other words, if your data is simply a series of timestamped results between team names then any model that is generated by them is going nowhere fast. The odds market evaluates all sort of data, lineups, transfers, what everyone says to the media, weather, fixes (yes they do happen fairly regularly), how much each team wants to win each particular match, etc. Do you believe that you can make a system that can collect and successfully model this data? I find it extremely unlikely. There is zero literature that claims to have produced such a model. To answer your question, I would suggest you forget about computer models and just spend the time on studying the aforementioned data yourself, as it is vastly easier and less time-consuming.Hi guys, new user here. A little background: I spent 5 years working in demand forecasting for a large power company and now work in energy trading. I'm reasonably comfortable with stats, programming and risk/expected value etc. I also used to play poker to a reasonable standard until the games got much tougher and I left uni and started commuting to work :) I am looking to develop a model mostly out of curiosity to calculate the odds UK football matches. I have a couple of questions as I start to plan my approach (currently focussing on a basic combination of home/away team expectations and refining): - How are promotions/relegations treated with historic data? Assuming I have data from the last 5 years. - On a similar note, do we run into issues of sample size at the start of a new season? - I have read much about poisson/zero interest poisson online. ZIP seems to provide a better fit but even then there are proponents of regular poisson. -
Re: The reality of sports betting/trading for a living.
Getting limited is the number one problem with gambling for a living. If it "didn't cross your mind" then evidently you are a light year away from being a pro gambler (assuming such a beast exists).Whilst my fine tuning of the 12 systems I have been using was going well, I stumbled across arbitrage betting one day and thought I have the money and time to do this, and can't lose, studied all the potential pitfalls and started Arb betting this Tuesday, only to get my account at Stan James closed after placing 5 arb bets with them this week. I have been trading with Stan James for 6 weeks staking £14,800 and winning £19,700 and as such they do not want my business. This to me was a big set back, and why I am researching before risking losing my other accounts. So far Bet365, Bet Victor and Sky have taken bets of many thousands and paid some wins of huge amounts without problem, so it never crossed my mind I would have this problem.
Most of the information there makes sense, some of it doesn't (e.g. the suggestion to always bet flat stakes), but in any case you should still keep in mind that that website is nothing more than a tipping service, and one that consistently loses money. People have tracked their bets and they are not winning anything near to what they claim they are winning. Steer well clear.Ps. Google > for some excellent advice on doing this for a living, it has some great information which makes a huge amount of sense to me.
The reality of sports betting/trading for a living.
in Betting Systems & Strategy
Posted
Re: The reality of sports betting/trading for a living.
No, we have not heard of them. How did you learn about them?