Hi guys,
I'm not sure this is to be placed here but if the mods think it should be transferred to other parts of the forum just do it.
I have been using a website that provides data (live football) and filters to build up strategies and also backtesting possibilities.
From a sample of 75000 bets I have tunned my strategies and got the following results:
Now comes my doubts:
1. Should I run a MonteCarlo simulation on all the Handicaps individually? I know that 162bets from +0.5 is not a very good sample to test but i'm not sure if i should do it individually or if taking only into account the overall picture/result.
2. The results above reflect only selections between 1.7 and 2.4 odds. At an average 1.93. This low average helps with the results from MonteCarlo simulation right?
Also, i'm developing these live strategies to deliver picks to followers because I have had some info that soft bookmakers accounts (bet365 included) withstand a lot more time without being limited/banned if the bets are only from live betting. Imagine an account that is limited after 300/400€ being able to go to 1000€/1200€. If the Mods allow it I will share my telegram for you to follow the picks service. If not, i can share my results here periodically.
Inplay Betting Strategies
- tomtom70
- Gaining experience
- Karma: 6
Post
Re: Inplay Betting Strategies
Hi Cufs,
to answer your question whether you should run an MC I would first counter with a question: what do you want to prove with the MC?
Without having calculated the statistics, just from a quick look at the numbers, I would say that the strategy works well with the majority of handicaps. Where there are only few observations it is not quite unusual or unexpected that you get larger deviations from the mean (expected value), which is obviously something like + 8.77 %.
In an MC you should, however, make a reasonable assumption about the distribution of both the odds you use to test the strategy and the "true" distribution of results. That's where it is getting really difficult because the latter is obviously unknown.
As far as I know the (absolute) standard deviation of your results will rise with the square root of number n of simulations (this is because the sum of these individual distributions ultimatively converges to a standard normal distribution). This means, that the relative standard deviation (in % of your bank roll) will drop with 1/sqrt(n). So why should you do a MC? The results will show exactly this behaviour but are heavily dependent on the assumption of the edge you have (the mean of the distribution). I personally would not do it, but it obviously depends on what (and to whom) you want to show.
to answer your question whether you should run an MC I would first counter with a question: what do you want to prove with the MC?
Without having calculated the statistics, just from a quick look at the numbers, I would say that the strategy works well with the majority of handicaps. Where there are only few observations it is not quite unusual or unexpected that you get larger deviations from the mean (expected value), which is obviously something like + 8.77 %.
In an MC you should, however, make a reasonable assumption about the distribution of both the odds you use to test the strategy and the "true" distribution of results. That's where it is getting really difficult because the latter is obviously unknown.
As far as I know the (absolute) standard deviation of your results will rise with the square root of number n of simulations (this is because the sum of these individual distributions ultimatively converges to a standard normal distribution). This means, that the relative standard deviation (in % of your bank roll) will drop with 1/sqrt(n). So why should you do a MC? The results will show exactly this behaviour but are heavily dependent on the assumption of the edge you have (the mean of the distribution). I personally would not do it, but it obviously depends on what (and to whom) you want to show.
- Cufs
- Gaining experience
- Karma: 0
- Location: @Smartchoice23
Post
Maybe i'm getting this all wrong but I want to prove that pursuing this strategy is going to return more or less the same Yield that it has been while backtesting.
Yeah I took the other handicaps because they were hurting the Yield % too much. Of course we cannot infer anything with such a low sample on those handicaps but i still prefer to have them taken and run some backtests later on to see if this has changed. But still, the low sample problem will still persist.
Yes. Unfortunately I don't have the edge of each bet. Well, the website actually shares the edge of each bet but it's too far away from the Yield itself. For the table i have shared it presents an average EV of 15%. Can i infer anything by using this edge/EV% when running a MonteCarlo simulation?
I don't need to show the simulation to anyone. Just want to make sure that i'm delivering a good strategy to my followers. The premise here is that i'm dealing with someone's calculation of value and i'm filtering that value + inplay statistics + other filters to do so.
Re: Inplay Betting Strategies
Thanks for your answer.tomtom70 wrote: ↑Tue Oct 31, 2023 2:39 pmHi Cufs,
to answer your question whether you should run an MC I would first counter with a question: what do you want to prove with the MC?
Without having calculated the statistics, just from a quick look at the numbers, I would say that the strategy works well with the majority of handicaps. Where there are only few observations it is not quite unusual or unexpected that you get larger deviations from the mean (expected value), which is obviously something like + 8.77 %.
In an MC you should, however, make a reasonable assumption about the distribution of both the odds you use to test the strategy and the "true" distribution of results. That's where it is getting really difficult because the latter is obviously unknown.
As far as I know the (absolute) standard deviation of your results will rise with the square root of number n of simulations (this is because the sum of these individual distributions ultimatively converges to a standard normal distribution). This means, that the relative standard deviation (in % of your bank roll) will drop with 1/sqrt(n). So why should you do a MC? The results will show exactly this behaviour but are heavily dependent on the assumption of the edge you have (the mean of the distribution). I personally would not do it, but it obviously depends on what (and to whom) you want to show.
Maybe i'm getting this all wrong but I want to prove that pursuing this strategy is going to return more or less the same Yield that it has been while backtesting.
Yeah I took the other handicaps because they were hurting the Yield % too much. Of course we cannot infer anything with such a low sample on those handicaps but i still prefer to have them taken and run some backtests later on to see if this has changed. But still, the low sample problem will still persist.
Yes. Unfortunately I don't have the edge of each bet. Well, the website actually shares the edge of each bet but it's too far away from the Yield itself. For the table i have shared it presents an average EV of 15%. Can i infer anything by using this edge/EV% when running a MonteCarlo simulation?
I don't need to show the simulation to anyone. Just want to make sure that i'm delivering a good strategy to my followers. The premise here is that i'm dealing with someone's calculation of value and i'm filtering that value + inplay statistics + other filters to do so.
- GNMbg
- Has experience
- Karma: 4
- Cufs
- Gaining experience
- Karma: 0
- Location: @Smartchoice23
Post
Re: Inplay Betting Strategies
Sorry. I cannot give away the main source of my service.GNMbg wrote:can you share the website which provides the data? thanks
- Cufs
- Gaining experience
- Karma: 0
- Location: @Smartchoice23
Post
Re: Inplay Betting Strategies
For anyone who wants to keep following this inplay source, i'm updating betting information every two/three days here:
tinyurl.com/BetTracker-FootballLive
Bets are actually placed by Smartbet.io software so these are real bets placed.
tinyurl.com/BetTracker-FootballLive
Bets are actually placed by Smartbet.io software so these are real bets placed.
- lm10
- Gaining experience
- Karma: 0
Post
Re: Inplay Betting Strategies
hey mate, how's going with your strategy? the link you posted for progress isnt working and i'll like to follow and see how it has performed.