Recently, I ran some experiments to clarify (for myself) some observations that I have made WRT FB Mogul. I wanted to determine whether computer controlled teams are given an advantage over human controlled teams and what the extent of attribute loss was in relation to a team's final win/loss record.

These are my observations:

The computer appears to be given an edge over human controlled teams when the game is played in simulation mode. On average, a computer controlled team seems to win about 0.25 more games per season than human controlled teams. I do not know what factor(s) the programmer manipulated to attain this result -- perhaps an increase in the computer players' attributes?

There are far more tie games in a FB Mogul season than happen in the NFL. Since 1974 (when overtime was initiated) there have only been 20 tie games in the NFL. That is about 1 every other year. A typical year in FB Mogul (with 32 computer controlled teams) has about 10 tied games. The fewest for a year in the 10 season trial (running the same saved file 10 times) was 6 and the most 16!

WRT attributes (pass blocking, pass defense, accuracy, turnover, etc):
During a season your players (and the computer teams' players) will not lose (or gain) attributes unless they have a serious injury -- in which case the injured player may lose attribute values, sometimes massively.

When you advance to the next season but before the draft, your (and the computer's) players will lose/gain attributes based primarily upon your record for the season. The worse your record, the better your players do. This is accomplished by nearly twice as many players on an 0-16 team gaining in attributes (as compared to an 8-8 team) and half as many losing attributes. The average amount gained/lost per player is roughly the same. But with a lot more gaining and a lot fewer losing attributes it makes a big difference. Heaven help you if you go 15-1 or 16-0...

Free Agents will also lose/gain attributes. For every FA who gains (on average 10 points) there will be four who lose attributes (on average 8 points).

Undrafted players also lose/gain attributes before the draft. About 60 will be affected. About 20 will gain (on average 1.2 points) and 40 will lose (on average 1.5 points).

Immediately after the draft, there is another gain/loss of attributes. Typically, 120 players (in the league) are affected. Drafted players tend to lose many more points than veterans. Human controlled teams are more negatively affected than computer controlled teams. Fewer of the Human controlled teams' players gain and they gain about 1.5 points less per player than computer players. For the players losing attributes, the number of players and the amounts lost are about the same for Human and computer controlled teams.

If a human controlled team has an excellent record the previous season, some (perhaps most) of your draft picks are going to lose attributes massively. In some cases resulting in a lost of 10-15 points in their PEAK attribute.

The Blocking attribute for RBs, FBs, TEs, and WRs is routinely decremented and can not be restored in versions 10 and earlier. So it is best not to select/acquire a player because they can block. You are better off going for other attributes in these players.

Attribute gains in a single year for an individual player are rarely over 40 points. Attribute losses can easily exceed 60 points.

The version of FB Mogul that I used for this analysis was 10.02. I doubt that anything has changed in the new version since the posts indicate there has been no significant change.