A common complaint is that people who sim games (let the AI run things) will not do as well as when they control things in-game (from micro-managing pitch by pitch to the genteel taking over when runners are in scoring position).
How about this for a quantitative test of your ability over the AI
[1] Form a league of two teams: (1) 2014 Yankees and (2) a duplicate of that team
[2] Turn off all story-lines, injuries, suspensions
[3] Create a 150 game season
[4] Play it out however you like best playing team 1
[5] If you win 95 games then you have a statistically significant result and you are probably better than the AI
[6] if you repeat this process for 5 seasons (repeat 2014 for 5 seasons, do not go to 2015) and keep an average of 88 games per season you are surely better than the AI
What do you think? Who is willing to give it a shot?
The results may help Marcus develop a better AI plus you can now make a comparison with other players. A nice controlled experiment.