First, I made a tabletop fighting RPG. The core idea was that there were a variety of advantages you could gain in combat (each associated with a stat), so the game was about struggling to achieve advantages while not letting your opponents have very many.
The core idea worked okay, but in expanding it into a full game I ran into the Mons Issue.
The Mons Issue is when a player character is essentially a walking bank of move bundles, and the primary play turns into choosing which move bundle to use. In something like Pokemon, these bundles are monsters - "mons". In this fighting game, it was fighting styles.
There's nothing inherently bad about mons games, but if I'm going to make a fighter game with that kind of mechanic, there's no reason to have the parallel advantage mechanic, since the two major play elements would just drown each other out.
I'm mulling over a new idea for a fighting game which is based around turn measures. I'm not sure whether I want to make it a tabletop game, or a video game. The problem with it being a tabletop game is that I need to figure out how to limit the move set so new players don't feel like they are drowning. Haven't figured out how to do that, yet.
I call this system the Measure Cascade system. It has two elements that make it unique and interesting, both of which are related to the "measure" concept.
A "measure" is a length of time. If you are acting at a certain measure, your actions are ones which take a certain amount of time. For example, a fast jab is a measure 1 action, while a high Brazilian hook kick is a measure 3 action. Movement is typically a measure 3 action, except for panicked (energy-wasting) movements, which can be faster.
All the participants show their chosen measure for a given round at the same time. So players A, B, and C might throw measures 1, 3, and 4. In a tabletop game, this can be done by just extending your hand with the proper number of fingers extended.
You start the turn phase at the lowest chosen measure - measure 1, in this case. Everyone at or below that measure gets a turn. Add one to the phase. Repeat until you finish the highest chosen measure.
So the turn order in this example would be phase 1: A; phase 2: A; phase 3: A + B; phase 4: A + B +C.
A would get 4 turns, but all of those turns have to be spend doing crappy measure 1 actions. C only gets 1 turn, but it's a powerful measure 4 turn.
So there's a constant seesaw between how many turns you want, how strong you need them to be, and whether you're granting your enemy too many turns.
The second half of the Measure Cascade system is the cascade. A cascade can be called when an attacker is striking at you, unless the attack specifically cannot be cascaded. This makes the attacker get a full success on their attack, but you retroactively get a turn at one measure lower. Essentially, an interrupt.
So if player B attacks player C with a measure 3 high kick, player C can declare a cascade. This means player B is going to hit full force with that kick, but first player C gets a measure 2 turn. He might use that turn to punch player B in the face, which would interrupt the high kick. However, because this is a cascadable attack, player B could declare a cascade and say "your punch will definitely hit me full force and interrupt my high kick, but first I get to do a measure 1 turn". Maybe he catches the punch, for example. If he succeeds in catching the punch, then player C's punch fails and player B's original kick is not interrupted, so it lands full force. None of this affects player C's turn any.
Cascades add a level of tactical concern, but most of the techniques that are most useful in a cascade are exhausting (use energy). So cascades are not going to resolve the same way every time. It also allows for "defensive" play - characters that are good at cascade play can block or counter very effectively, which means their opponents will generally want to stick to uncascadable moves (or measure 1 techniques, which are inherently uncascadable).
I also added in some range rules to allow for complex multiplayer behavior - most attacks are "clench" range, but any two characters in a clench are the only people in that clench.
If player A spends their turn moving to clench range with player C, then player A and C are clenched. But if player B then does the same, A is relegated to medium range, the clench broken. B and C are now clenched.
This is something to be aware of (you just wasted A's turn) and also to use to your advantage (get A out of clench if they can't weather the enemy's upcoming turn). Also, certain attacks work at medium range, and are therefore useful to do without clenching.
Anyway, I'm thinking about it, trying to decide whether to make it a video game I never complete, or a tabletop game I never complete. Right now, it's a rather tabletoppy system because that's what I initially designed it for, but I can see an easy changeout to a ticky fast-turn-based combat video game.
Anyway, that's the kind of rule design I do when I'm thinking up games.