I'll use the MobA games to illustrate what I'm asking since it's easier to explain even tho I want to apply that algorithm to time series.
In a MobA game you have more than 100 characters and every game involves 5v5 with 10 different characters. So it's 10 choose 100 meaning there are 17,310,309,456,440 possibilities.
What I would like to do is use a database with millions of games and work out probabilities for 10-characters setups to come up. Since players will often choose the same characters the high combinatorics won't stop multiple games from having the same setup in a pool of let's say 1 million games.
However given the high combinatorics I can't just make a table with all possibilities and put every new game setup to a case. How should I proceed?
I've been thinking to maybe create a case with first game checked, and then for each new game either higher the count of that first case or create a new case if it's doesn't correspond and so on. Is that the most efficient way to proceed?
