At the request of some friends, I will introduce the fuzzy method.
Because the fuzzy method covers a wide range, due to time constraints, I can only illustrate with one example.
For instance, if you take 5 sets of 16 red numbers and 5 sets of 17 red numbers, making a total of 10 groups of red balls, all randomly selected, according to probability, each group of red numbers should on average match 3 numbers. However, in reality, probabilities are random and uneven, so we need to allow for error. Based on verification over multiple periods, it turns out that among the 10 groups, the majority that match 3 red balls is only between 4-7 groups. Those that don't match 3 balls may have 5 red balls, or 2 balls and 1 red ball; very few will match all 6, and very few will match none of the balls. Next, you can select another 4-5 red balls from the 10 groups, with 1-3 groups matching accordingly. And so on. Note that the scope of error allowance is very important as it is an effective way to reduce the number of valid bets.
There are many such fuzzy methods, sometimes even allowing for fuzzy key numbers.