It's possible, but as with any simulation the realism of the results is proportional to the realism of the simulation. For example you could have a statistic per-player that indicates how likely he is to get injured in a game, say 0.05 for 1 in every 20 games. It might turn out to look ok but there is a whole bunch of information missing from determining whether he gets injured:

- Was the other team aggressive?

- Was he playing in or out of position?

- Was he 100% fit before the game?

- How many minutes did he play?

- How aggressively was your team as a whole tackling the opposition?

- How aggressively was the opposition told to attack your team?

- How aggressively did you ask him alone to tackle other players.

The list goes on, there are tonnes of metrics by which such a decision can be made. Football Manager didn't start out analysing all of these things. Back in 97/98 it was probably much more simple and since then it has evolved into a beast of a simulation. I would suggest just starting simple and gradually introducing other factors into your AI decision making as you find your existing systems limitations. Don't set out to beat Football Manager from the start because as a one man team it is simply too much work.

Source: I was a gameplay developer on Championship Manager 2008 and 2010.

Have fun!

Thanks for the reply.

In my last 6 months of fiddling with the probable algorithms, I realized that it is a herculean task. And this is the reason I thought about going with the results without any gameplay; fully aware that it won't give the players that thrilling experience. But then, I don't have any team at my disposal. It has really been a effort gone in vain searching for the gameplay developer in India. There just aren't enough people!!!!

I will ask your suggestion for one thing- should I wait until I find the willing guy or should I start with some what half baked version of the game? I really want to test my ideas about whether these will be acceptable by the community. So, it is becoming a kind of itch that is getting intolerable.

You will not find someone capable who is prepared to dedicate massive amounts of time to your ideas without proper compensation, you really only have three options:

1. Get the money to pay people to work for you.

2. Start working yourself, if you make something good and take care of all the heavy lifting others will be willing to help you make it better.

3. Forget about it.

One important thing to consider is how you are simulating. Is it every second, every minute, between events or just the match as a whole. This will drastically change how your simulate. I have been working on a Basketball Manager game recently and had to make this decision. Early on my approach was to simulate possesssions at a time. However I ran into issues figuring out what impacted each possession. While it could be done on a basic level it did not produce the desired results. I then switched to simming every second of the game and simulating each players actions at each point. This involved looking at the behaviour of each AI player, their ratings and at the current state of the game. Then with some randomness thrown in, the players would make decisions on what actions to perform.

Once decisions were made it was then a matter of calculating the results. Usually this involved some probability check, ie was the shot successful or not. This used a players rating in the calculation. These expanded over time taking into account more aspects that were discovered from research.

Things likes subs and manager tasks I used a different approach. For this I wrote different evaluation formulas that the manager would use to judge if a player should taken out, started, subbed in etc.... These formulas took into account, energy, performance and the ability of the player. Basketball uses rolling subs so Football would be a bit different.

Not sure if this helps and obviously I'm no expert on this, so this is all just based off of my approach to the problems.