Unfortunately, Deep Blue gets to utilize the Min-Max algorithm, which is a well-known algorithm for well-understood games:
For each legal move, assume your opponent makes the best possible response, and you make the best possible move after that, etc... look, say, seven moves ahead, or in the case of a supercomputer, eleven, even fifteen moves ahead. (Ideally, by "pruning" paths which you can see have already suffered horrible losses.)
That algorithm is simply not applicable to FM at all: the number of simultaneous opponents would raise the "possible response" moves to such an overwhelming size that even a supercomputer would be paralyzed trying to look seven moves ahead .. while the introduction of random numbers would render the entire concept moot.
I
do think that there are "look ahead" mechanisms which FM needs, but Min-Max isn't the answer.
What I would like to see is the AI "looking ahead" only in regards to its own squad. Each team should have "in mind" who it thinks its starters and second-choice are at each position "after this transfer window", "two-and-a-half years along the line", and "five years along the line".
The first would result in the AI, for example, not selling its only goalkeeper. In fact, ideally it shouldn't sell either of its
two best goalkeepers, unless there is a third goalkeeper on the squad whom it would classify as a "competent backup".
The second aspect would drive contract renewals. If a player would be projected as a starter, who is significantly better than the second-choice, and contributing at or above the club's current level 2.5 years down the line .. but his contract is running out .. the AI would be a lot more aggressive about re-signing the player.
Finally, the third time-frame would drive youth development: the AI would be a lot less willing to sell young players which it projected as having the ability to step into the starting lineup three to five years down the line.
You can do all of that without Min-Max.
Chess AI and FM 
Linear Mode
