I would agree with LordNathan's assessment of the merits of Tecmo Super Bowl -- it is indisputably the best sports game for the NES, and quite possibly the best NES game, period. Basically, what made this game awesome is the number of options and statistics that it has, which perfectly complement its above-average gameplay.
First of all, it has complete rosters from the 1991 NFL season. The rosters all seem reasonably accurate (or at least, all the stars are in the right place), and each player is accompanied by a picture of themselves. Of course, it's not really a picture of themselves; it looks like they had about 10 models with adjustable hair color and facial hair style. But, it's the thought that counts, and it's still pretty cool.
Each team's offense was very customizable. There was a starting and backup QB, 2 starting and 2 backup of both running backs and wide recievers (although run-and-shoot teams like the Oilers and Falcons had 1 RB and 4 WR's), and a choice of kick returner and punt returner from any of the offensive ball-handlers. So, if you were a big Steve Young fan, you could sub out Joe Montana, and put Steve in.
In addition to the substitution of players, each player had his own individual stats. All of the players had certain basic stats, like running speed and running power. In addition, every player had position-specific skills. Quarterbacks had passing accuracy, kickers had kicking accuracy, and defensive backs had an interception rating. All of these stats varied on a scale from 1 to 100.
Not only that, but the players themselves could vary during the season. Their physical condition varied on a scale from Poor to Excellent, with Average and Good in between. A player who was feeling poor would have his stats slightly deflated -- Jerry Rice, for example, would only be catching passes at 75%. However, when he was feeling average, he went up into the 80's, and when he was excellent, he was well into the 90's. Players seemed to spend most of their time at average, going to the positive side randomly and the negative side when they got "tired" (or, at least that's how it subjectively seems -- I have no clue how they actually determine it). There are also injuries, which sideline the player for 1-3 weeks. Injuries only happen to an offensive player who was carrying the ball and was tackled to end the play. Sorta limited, but still not that bad.
OK, so we've got all this stuff. Are we done yet? No! Each team has four running and four passing plays. The designers included a fair number of gimmicky plays. There were several flea-flickers and quite a few reverses. In addition, there were normal plays that looked like those trick plays but instead developed normally. By default, teams had these sets of normal and trick plays all together, but you could substitute out any play for one of your choice. It was really cool -- if you wanted to have 3 types of hail marys, you could. (Unfortunately, interesting plays were limited to offense. On defense, you could only pick a play to defend against, as stated above.)
The actual gameplay was quite well executed. Aside from the inability to switch players once the play starts (on offense you are always the person with the ball), I don't have any major complaints. Once the computer with a fast runner (i.e., Bo Jackson), got past your secondary, he was pretty much gone, which was sort of disappointing. In addition, once you've played the game for 8 or 9 years (/me looks around guiltily), you have a pretty good idea of how the AI's going to go after you. The result is somewhat comical to the casual observer of the game. Having caught a screen pass, your player procedes to juke and weave around the field, sometimes turning at right angles or going completely backwards, leaving the computer's defenders either overrunning you or diving futilely in your wake. It's amusing, at least for a little while.
And finally, we come to one of the crown jewel of the game: the statistics. Everything, and I mean everything (note: I do not actually mean everything) is recorded. QB stats can be sorted by completeion percentage, TD's, yards, or even average yards/pass. Similar stats were kept for running backs and receivers. Sacks were recorded, punt distances were measured, points were tallied, and punt and kick returns were tabulated with the utmost care -- not a yard was lost or a touchdown misplaced.
... or so I thought. Just the other day, I was playing my way through a season as the San Francisco 49ers, with the idea of seeing how much I could run up the statistics for an individual player. (Ordinarily, I'd be playing the Washington Redskins, who have the double bonus of being my favorite team and actually winning the Super Bowl that year, but the combination of Montana to Rice was just too good to pass up). So anyway, I go through the season, pretty much passing deep to Rice whenever I could (I think I ran the ball for 100 yards total, over 16 games, to give you an idea). Come the 13th game, I go to check my stats. I see that I have 4009 yards for Rice, with 60 TD's. Feeling pretty smug, I decimate my lowly opponent by 40-something points. I check the screen at the end of the game -- sure enough, Rice gained 300+ yards, and I remember that he scored 5 TD's. I go look at the leaderboard, and find him still in first place (duh), but with 4095 yards and 63 TD's.
...Huh?
That's not right.
D'oh!... you mean that I'm limited to (2^12 - 1) yards and (2^6 - 1) TD's?
Screw this game -- I'm gonna go play Tetris. (Wanders off muttering about lazy programmers and poor programming techniques).
(In defense of the game, scoring more than 40 TD's in a season has never been done by a WR, and 2000 yards is considered phenomenal. But still, in the game it was possible. Would it really have killed them to have added an extra