Topaz
Programmer/TheAwesomeGem
Let me give you a brief introduction. I recently formed a small indie game studio online and we had an idea to create a Football Management Simulation game. We have thoroughly discussed our boundaries and on the fact that we want to make a fair and realistic football management game and researched a bit on Football Manager 2018. We had found a lot of interesting info and data about how Football Manager 2018 made their Match Engine(which is a integral part of a Football management sim) and we oppose the idea to replicate their system. Basically the way their match engine works is that every quarter of a second, all players, referee, linesmen and the ball runs their logical AI concurrently. While the idea sounds good on paper, you can see when playing Football Manager 2018, how you get so much random moments and errors that really makes no sense and draws you out of the game. We won't do that. Our idea is simple. Get a lot of statistical data from formation, style of play, player traits, player stats, etc between two teams and generate probability equation and then use a discrete event simulation to then pre-generate all the events and stats that is going to be displayed in real time. We do have to pre-generate the stats and events every time a match starts, half time kick off, when you make tactical changes/subs, etc. Kinda like how Football Manager 2018 does it. However, we won't be simulating all 22 players, referee, linesmen and the ball since that leads to a lot of error and quite hard to get it balanced(aka Chelsea losing to Chelsea u-23 without any reason in FM 18).
Right now, we are trying to form data from formation. Things such as the number of players on defense, midfield and attack and the compactness of the position as well as the role each player to generate some data. What kind of data could we generate from just a formation apart from the ones we had? The best thing about the way we are doing the system is that we have endless possibilities of a realistic simulation as well as being able to test and balance out any issues easily. We are using weight-based probability system since it's dynamic and we can also add on more variables to keep track on rather than a chance-based probability system.
Right now, we are trying to form data from formation. Things such as the number of players on defense, midfield and attack and the compactness of the position as well as the role each player to generate some data. What kind of data could we generate from just a formation apart from the ones we had? The best thing about the way we are doing the system is that we have endless possibilities of a realistic simulation as well as being able to test and balance out any issues easily. We are using weight-based probability system since it's dynamic and we can also add on more variables to keep track on rather than a chance-based probability system.