Recommendations for creating algorithms for beginners.

When working with retro analysis, pay attention to an important detail:

IF YOUR ALGORITHM IS NEGATIVE IN THE SELECTION FOR ALL LEAGUES, THEN REMOVING LEAGUES WITH NEGATIVE RETURNS WILL NOT WORK! THIS IS THE MOST COMMON MISTAKE MADE BY BEGINNERS!

What does the term “all leagues” mean?

These are the leagues that the bot covered during the collection of matches, and you formed your algorithm based on them. 

The list of leagues is formed before the start of the collection of matches and should be as extensive as possible. You can exclude the following in advance: friendly matches, women's football, and youth leagues for players under 18 years of age +/- as desired. Or you can run ALL LEAGUES from the constructor and discard the most unstable ones after forming the algorithm WITH A POSITIVE ROI.

Collector.

The most accurate and profitable algorithm can only be obtained with the help of a match collector who has worked for at least 1 season, and in the most reliable case, about 2-3. 

After obtaining such a match database, we load it into an Excel table and find the most profitable dependencies of values, both statistics and odds, after which we check the resulting algorithm for profit, drawdowns, and stability.

The most significant parameters when configuring algorithms.

Entry minute and odds. The entry point into the match literally determines the degree of value of your outcome. How this works is clearly seen in the retro analysis. Take the same prediction and check it at different minutes of the match (preferably with a small interval of +/- 5 minutes so as not to lose the number of matches), and the profit, passability, and ROI will differ significantly. 

Odds. It is important to choose the right range of odds at which your algorithm will give you the desired profit and stability. If you want a large profit at high odds, you will have to endure long drawdowns and protracted losing streaks, while smaller values will give you more confidence and peace of mind, but not as much profit. The most profitable odds in the long run for TB start at 1.6, and for TM at 1.9. 

These two most important parameters determine the lion's share of your bot's success. Add to this the advantages of attacks and shots, and you get the simplest working algorithm. Well, for more accurate tuning, we process the statistics more carefully. 

In addition, it is worth paying attention to the match score, which needs to be correctly selected for your prediction. If you are working with TM goals, keep in mind that the most suitable scores for this market are 0-0, 1-0, 0-1, and 1-1. If there are 2 or more goals starting from 0-2 or 2-0, the probability of goals scored increases.

Thus, we have obtained the most accurate order for creating a selection:

1) Entry point into the match. 

2) Range of odds. 

3) Match score. 

4) Shots, attacks, corners, yellow cards, etc.

Conclusion of this article:

You have learned about the simplest and most important conditions on which the success of your algorithm depends, but do not forget that there are hundreds or even thousands of variations of combinations of conditions, so if you have not found a suitable option, it is worth simply looking for it. “It doesn't work” is not an answer. “I didn't find it, or didn't want to find it” is much more truthful.

P.S. If you find it difficult to work with these dependencies, choose ready-made bot options or consult people with experience in analysis.