I've thought a lot about this because sports games have always been a passion of mine.
Traditionally, there's two or three core aspects of most "simulation" sports games:
- Experience on the field / in play (think, a boss battle in Dark Souls)
- Franchise / Owner / Team Builder / Player Builder mode (think, the character development in an RPG)
- Competitive, player v. player modes (PvP in a fighting game)
(There's more, but I'll address these primarily, FWIW, I find #2 and #1 most important, #3 is unimportant to me in a lot of ways but I get why it's still important to others. Personally, I think #3 often works against #1 and #2, but its still a very important aspect of the game for many players... Similar to how Fallout enthusiasts are skeptical of how Fallout '76 online will affect the Fallout game they love, or how RDR enthusiasts are very worried that RDR2 will borrow too much from GTA Online and lose focus)
In #1 Experience on the field, this dictates player behavior, CPU opponent AI (e.g., the behavior of the NPCs that the videogame player is not controlling), strategy of NPCs (e.g., the other team that the videogame player is not controlling), graphics, sound, presentation, and everything that would go into a match, game, or single event sports contest.
There is a huge amount of improvement that can be done in all of these areas, in every sports game. Some games replicate some aspects very well, like baseball games. Baseball games are one of the best games at replicating the behavior of NPCs in play, because baseball is a sport that is limited in the functionality of specific players. A catcher, for the most part, will never play 2nd base. An outfielder, for the most part, will never play 3rd base. A base runner will never run from 3rd base back to 1st base, or go back to the batters box and hit the ball again once its on the ground. There is a very sequential nature to baseball: A ball that is out of play will never suddenly return to play unexpectedly; a ball thrown will never be thrown by the pitcher to an outfielder.
This is one reason why a lot of baseball games from 10-15+ years ago are still considered really good (High Heat 2002 and MVP Baseball 2005 are still considered among the best baseball games), because once you nail down the majority of scenarios, you can have a good playing baseball game, and then you can focus on refining those scenarios to perfection. Sure, advanced AI will be difficult to pin down, like logic on when outfielders should throw to specific bases, but a lot of the most difficult logic in baseball games is in forcing NPCs to make logical mistakes, instead of be logically perfect. I'm not talking about errors (an aspect of baseball where a player makes a mistake executing a routine play on the field), but more... "Does outfielder Z have enough arm strength to throw out baserunner A while baserunner A is trying to advance from 2nd base to 3rd base; Does baserunner A have enough speed to run faster than the speed and accuracy of the ball; Does infielder Y have enough fielding ability to catch the ball thrown by outfielder Z and tag out baserunner A?" These are all logical math operations that a computer can calculate, and then it's "just" a measure of replicating that visually on the screen and ensuring that what you see visually matches up to what the computer has calculated. The challenge, with baseball games in this case, is making them play realistically: IF outfielder Z makes that throw 100% of the time, or if baserunner A beats out the throw 100% of the time, then it removes the variability of what makes baseball an enjoyable pasttime.
Sports videogames are unique in that you have to simulate mistakes, fatigue, and thousands of other aspects because that's what makes it special. James Harden, the NBA basketball star and reigning MVP, was known for most of his career as being a lazy defender. He took plays off, watched his opponents score, didn't even try to stop them sometimes. It's part of the joy and madness of a star like Harden, one of the most talented guards in the NBA who could play great defense, but for specific reasons, often chose not to. Think about that in terms of a videogame. What if you had a challenging boss in a game like Dark Souls, one of the hardest bosses of the game, the proverbial "MVP-candidate boss of Dark Souls," but, like, 5% of the time he make a conscious decision to not bother trying to defend your attacks... but 95% of the time he was incredibly effective. This would get reported as a bug or if it happened often and was intentional, it would be frustrating for players -- "I can't find the pattern, this boss is completely random... 90% of the time he defends himself as I expect he will, but 10% of the time he stands there and lets me attack him!"
AI is difficult in sports videogames because if every player makes the right decision 100% of the time, then it feels scripted and robotic. In older versions of Madden and NCAA Football there was a name for this, "Robo-QB (Robotic Quarterback)." It's debated whether RoboQB ever truly existed, but there's a decent body of evidence that suggested that in mostly older versions of EA developed football videogames, that in specific scenarios in the game, the NPC offense would trigger an AI setting that would, for the most part, make them choose the perfect play for your defense and make the NPC quarterback make the perfect decisions, perfect throws, and perfect play execution. In the old NCAA Football games (R.I.P.), this would result in mostly bad QB's having these unstoppable streaks of amazing performance virtually no matter what you, the player playing defense, did. It was natural: This is a computer trying to be a person, and the sport of football is designed in a way that there is -- basically -- always a designed play that will always succeed against another designed play, as long as the play is executed by the players on the field.
SImulating mistakes (or fatigue, lapses in judgement, split second decisions etc) is very difficult, and if it feels arbitrary in a game, then fans will hate it. Tom BRady, the greatest quarterback of all time, throws interceptions, sometimes inexplicably. He throws fewer of these than every other player in the game, but, sometimes he
does throw a bad pass that is intercepted. In videogame world, the game will sometimes compensate for this and throw a bad pass inexplicably -- sports devs/gamers usually refer to this as "a dice roll" mechanic, it's common in all sports games because simulating mistakes/imperfection is harder to do than calculating perfection, so they rely on a calculation of randomness in specific scenarios -- and this feels
random in a bad way, like the game scripted a bad pass from you the player even though you did
exactly the same mechanic you've always done. Imagine this in another game like, say, Street Fighter 2, where you execute
the perfect button combination to throw a fireball, you've done it 50 times already perfectly, but this time the 51st time, the dice roll mechanic takes effect and instead of throwing a fireball at your opponent, Ken throws nothing and leaves himself open to a devastating, match losing counter-attack... For the player, this would feel cheap and that game would quickly be shoved off of the competitive circuit because elements like that should have no place in a fighting videogame.
Realism, in sports games, is very difficult to achieve on the field without it feeling arbitrary and random.
Another aspect of gameplay on the field is that you want to simulate the events of a sports game, but it's bad when you take control out of the players hand. I feel this is the biggest problem with the NBA 2K or WWE 2K games. Now, WWE 2K is widely panned, the game is horrible and I'm a big critic of it, I could write pages on how terrible it is. But, NBA 2K is a game that is still -- mostly -- praised. But, personally, I think that basketball games started to lose something around the move to 3D rendered players, because in an effort to simulate what you see on TV, they have to take control out of the players' hand. In a basketball game like NBA 2K, when you're advancing the ball down the court, if you hold a trigger and then perform a motion on the analog stick, your athlete -- based on his attributes -- will perform a maneuver to try to shake his opponent. On defense, the opponent will be able to hold a trigger and then perform a counter move at the same time which will try to defend the play. The game will then "play this out" in a short animation, taking control away from the player almost like a quick time event, until the next opportunity to do something else. Attributes will go into the calculation, an athlete with poor handling skills might collapse on his face or lose the ball, a defender with poor defending ability to make a paultry attempt to swipe at the ball and "break his ankles" (basketball lingo for getting your feet tied up while trying to defend a ball carrier), but this isn't the sport of basketball. THis is a videogame fighting mechanic implemented in a basketball game, and as basketball games increasingly rely on animations trying to replicate what you see on television, and they map those to button presses, it makes the animations feel cheap and unrewarding. In real life, if Lebron James is advancing the ball, does a cross-over to a spinmove, breaks the ankles of a defending James Harden, and follows it up with a monstrous slam over another player who gets posterized, that will be one of the seminal moments of the year... a 5-second play that will go down in the annals of the basketball season. If, in a videogame, you hit LT + right stick waggle -> X ... and Lebron does that every time, and James Harden's ankles break every time, it feels cheap, it loses the effect, it feels routine, and you don't feel like you just made this devastating play, you feel more like you're playing a Tomb Raider escape sequence and you just "pressed X for amazing."
Simulating sports games
is hard.
