Is fractional kelly any good for value betting? Building my bot

Pages: [1]
Author Topic: Is fractional kelly any good for value betting? Building my bot  (Read 1600 times)
esi
Gaining experience
*

Karma: 3
Posts: 32


View Profile

esi

Is fractional kelly any good for value betting? Building my bot
« on: October 19, 2019, 08:59:37 PM »

Hello, im building my own value betting bot and im using pinnacle as the provider for the "true" odds.
My bot is almost finished but i cannot decide on the staking plan.
I know nobody likes to share such thing with others but im really in a big dilemma and afraid all this work will be for nothing.
What i have done so far is as below:
1. set a max stake
2. only accept bets with 3%+ value
3. use the real odds from pinnacle to calculate real probability
4. calculate if i will get a positive roi from that bet

So based on the above what would be a decent staking method, is fractional kelly any good?
Thank you
Logged
VidaBlue
Becoming a Pro
***

Karma: 47
Posts: 236


View Profile

VidaBlue

Re: Is fractional kelly any good for value betting? Building my bot
« Reply #1 on: October 20, 2019, 04:26:58 PM »

Hi!

I think that the only simple answer at this stage is probably also a useless answer: it depends. The criteria you're after are also answers that many sport bettors are constantly trying to pursuit. In your case it would depend on so many factors and a setup that might be unique for your situation, such as, among many other things, the bookmakers available in your country.

Quote
1. set a max stake
Calculated stakes may get very high, if the calculated EV is very high and odds very low. I find it uncomfortable to deviate very much from the flat stake size into the high regime, because there's always the risk that the calculated hypothetical EV is wrong, so I would decide for a max stake size too.

Quote
2. only accept bets with 3%+ value
Sounds like a good idea, maybe set the limit even higher. This is my own hypothesis that I use, so I wouldn't claim it is the truth, however: One sharp bettor's action may shift the lines way more than 3%. Therefore I think that a true odds reference, such as pinnacle's, represent an interval of possible true odds and not just one value. In order to ensure that the bet has value as per definition, I would factor in the juice of the sharp reference as well. Any arb is therefore a value bet according to this hypothesis, anything lower MIGHT have value.

Quote
3. use the real odds from pinnacle to calculate real probability
If you don't have anything more elaborated, this should be a good starting point.

Quote
4. calculate if i will get a positive roi from that bet
I don't understand the meaning of this bullet.

In your shoes I would just start out, maybe even run the bot without betting, paper trading for a while, with the sole intent to gather as much data as possible. This way you won't get your soft accounts limited. But if you can't wait for action, then by all means start out, preferably very conservatively - you can always increase the Kelly fraction. Betting for real instead of paper trading may give more commitment to the project. Overall your bot should be profitable basing it on pinnacle lean, but there are so many setups where no side is sharp or pinnacle even is the softer side, so the more data you gather, the clearer this should become.

I am not suggesting that the following is true in your case, but for me I often find that automation is so great, that I easily get seduced by such a wonderful solution and let to think that it can do things better than me. It will never be able to do betting better than myself, but Mr. Robot is very persistent in doing what I instruct it to.
Logged
esi
Gaining experience
*

Karma: 3
Posts: 32


View Profile

esi

Re: Is fractional kelly any good for value betting? Building my bot
« Reply #2 on: October 23, 2019, 03:16:00 PM »

Thank you for the reply, I thought nobody would answer.
The bookie I'm using is bet365 for now.

Quote
3. use the real odds from pinnacle to calculate real probability
If you don't have anything more elaborated, this should be a good starting point.

- What do you suggest is a better option? I'm using betburger so I have the option to compare with all other bookies

Quote
4. calculate if i will get a positive roi from that bet
I don't understand the meaning of this bullet.

- What i mean is based on the pinnacle probability i calculate the expected value if positive or not:
https://www.pinnacle.com/en/betting-articles/Betting-Strategy/how-to-calculate-expected-value/EES2VE46TM4HTT32
"(Probability of Winning) x (Amount Won per Bet) – (Probability of Losing) x (Amount Lost per Bet)"
I found this way to be kinda flawed as this filter removed higher odds eg 2.5 to 3+ since their probability is small, anyway i need to have a better look at this if im doing something wrong or just remove this filter after all.


In your shoes I would just start out, maybe even run the bot without betting, paper trading for a while, with the sole intent to gather as much data as possible. This way you won't get your soft accounts limited. But if you can't wait for action, then by all means start out, preferably very conservatively - you can always increase the Kelly fraction. Betting for real instead of paper trading may give more commitment to the project. Overall your bot should be profitable basing it on pinnacle lean, but there are so many setups where no side is sharp or pinnacle even is the softer side, so the more data you gather, the clearer this should become.

I am not suggesting that the following is true in your case, but for me I often find that automation is so great, that I easily get seduced by such a wonderful solution and let to think that it can do things better than me. It will never be able to do betting better than myself, but Mr. Robot is very persistent in doing what I instruct it to.
[/quote]

I can do tests using python or php but is there any api to know if the match i intended to bet was won or not? How do you do it
I haven't implemented fractional kelly as of now but probably i should
Thank you
Logged
chrisgr1984
Gaining experience
*

Karma: 1
Posts: 44


View Profile

chrisgr1984

Re: Is fractional kelly any good for value betting? Building my bot
« Reply #3 on: October 24, 2019, 07:00:39 AM »

From my experience I would say that the staking plan depends of your model and you will have to check this doing some back tests your self using your data. There are value models that are working better with flat stakes , others with Kelly , others with just taking the max bet etc etc You need to check which working the best on you .
An example : Kelly is very aggressive on high values but there are models that are very good when the value is low , and when the value it gets higher anomaly appears and the model seems to be broken and losing so to apply kelly on such an example can be devastating
My 2 cents
Logged
Phoenixking
Guest


Email

Phoenixking

Re: Is fractional kelly any good for value betting? Building my bot
« Reply #4 on: October 24, 2019, 04:09:16 PM »

Hello, im building my own value betting bot and im using pinnacle as the provider for the "true" odds.
My bot is almost finished but i cannot decide on the staking plan.
I know nobody likes to share such thing with others but im really in a big dilemma and afraid all this work will be for nothing.
What i have done so far is as below:
1. set a max stake
2. only accept bets with 3%+ value
3. use the real odds from pinnacle to calculate real probability
4. calculate if i will get a positive roi from that bet

So based on the above what would be a decent staking method, is fractional kelly any good?
Thank you

Bro, do we need advanced skills in programming language to build simple bots ?
What i'm thinking is that, I will specify set of simple instructions that the bot will follow and collect data for me , That's it. Paper betting basically.
I wish I can help with your situation, But I'm a beginner myself.  ;D
Logged
esi
Gaining experience
*

Karma: 3
Posts: 32


View Profile

esi

Re: Is fractional kelly any good for value betting? Building my bot
« Reply #5 on: October 25, 2019, 03:07:23 PM »

Hello, im building my own value betting bot and im using pinnacle as the provider for the "true" odds.
My bot is almost finished but i cannot decide on the staking plan.
I know nobody likes to share such thing with others but im really in a big dilemma and afraid all this work will be for nothing.
What i have done so far is as below:
1. set a max stake
2. only accept bets with 3%+ value
3. use the real odds from pinnacle to calculate real probability
4. calculate if i will get a positive roi from that bet

So based on the above what would be a decent staking method, is fractional kelly any good?
Thank you

Bro, do we need advanced skills in programming language to build simple bots ?
What i'm thinking is that, I will specify set of simple instructions that the bot will follow and collect data for me , That's it. Paper betting basically.
I wish I can help with your situation, But I'm a beginner myself.  ;D
depends, but at one point you will need good programming skills
Logged
Phoenixking
Guest


Email

Phoenixking

Re: Is fractional kelly any good for value betting? Building my bot
« Reply #6 on: October 26, 2019, 12:51:09 AM »

Thx for the reply bro, Good luck with your bot. :)
Logged
Pages: [1]
Print