A betting bot is nothing more than a computer program that bets on Betfair in our stead.
In this article I will not be talking about the programming aspect of which I know little about. Instead I'll be talking about the most important aspects you should take into account when thinking about creating a statistics based bot, before you go spending your time programming.
First criterion
The first step towards building a statistics based bot is, finding a lucrative strategy.
First we formulate a hypothesis and then we must test it to figure out if it has any value. An example:
- Hypothesis: Always betting on the favourite, is it lucrative after X bets?
- Test: The larger the sample we use to validate our bet, the more relevant our conclusion is.
From the moment we find a lucrative statistic we must use our betting sense and:
- 1st figure out if this strategy makes sense,
- 2nd if this is a repeatable statistic.
After getting to our lucrative, sensible and repeatable strategy then, our next step is to translate this strategy into IF{} conditions.
Translating this strategy into an algorithm might seem like a simple task but I guarantee it is not! Most of the strategies I use as a live trader are untranslatable into IF{} conditions.
An example of a strategy that you cannot include in a bot:
- I see that Barcelona is controlling the ball but showing no rush getting into the opponent's box. They appear interested only in keeping possession and rest by passing the ball around.
- So I bet on the closest Under.
This strategy makes sense and is repeatable, I cannot use it with a bot though. This because the bot is unable to understand if Barcelona is, or isn't, trading the ball creating no danger.
A betting bot can only preform simple if{} analysis, which are majorly supported by statistical and odd analysis.
Example of a bot based on statistical analysis of the present (the match itself):
- If a team has over 5 corners before the 30th minute,
- Then bet for that team.
Example of a bot based on statistics of past matches:
- Analysing the last 30 matches of a team, if more that 30% of the team's goals were scored on the first 15 minutes of play,
- Then bet for that team before the match starts and close that bet before the 15th minute.
These were two hypothetical examples of strategies that we can apply to a betting bot. These strategies were not tested nor do they guarantee profits!
Second criterion
It is important that we, as bettors, understand that a bot is stupid, for real.
We, humans, are able to look at something and realize when it makes absolutely 0 sense. For instance we know that if the match is 2-0 for Benfica, it cannot end 2-0 for Porto. A bot though, if we do not tell him that, it might not be able to know or even understand this situation.
So, if it was not properly programmed, integrating all the mentioned details, the bot might end up betting on a 0-2 when the match is already 2-0.
Third criterion
The fact that the tactic that you discovered is lucrative and translatable into IF{} conditions is not enough, simply being lucrative doesn't cut it. This tactic must be super-lucrative!
Because that tactic has to be able to cover the costs of entering and leaving the market (higher for bots than humans).
I'll speak for myself, I'm experienced enough to know what the best odd is and if it is better for me to wait a little bit more before I place my bet or to take the immediately available odd.
That is, in my case I can have lower market entering and exiting costs because I do not blindly take the immediately available odds.
But, I cannot "teach" these considerations to a bot through IF{} conditions, this means that the bot has entering costs much higher than my own.
I hope I've heled, summarizing here the aspects you must take into consideration in order to successfully build a betting bot.
Hug,
PR
