This is an expansion of @vlaz excellent comments
Knowing all possible moves vs knowing the exact move are not the same.
Let us use a bounded system - such as noughts and crosses (or Tick-Tack-toe for my American readers) to demonstrate this:
I know my opponent will make a move and there is a small and ever decreasing number of places - such that any competent opponents will end-up in a stalemate (Gotta include the classic scene from War Games here)
So - how do you win a game? Well there are two elements required for a victory:
1: Your opponent makes a mistake2: You force your opponent into a situation where they can only stop 1 of 2 paths to victory.
e.g. see this GIF: As an example
Regardless of what the opponent chooses to do, there they cannot block both vectors of attack - so they loose.
This is the first option - it isn't true precognition, it is knowing the all the possible moves and rank-ordering them by probability - the two fighters attempt to force each other into a position whereby the other cannot counter all the possible attacks and so leaves themselves open.
Option 2 - Long Term planning, limits of human agility and cool-downs
You mentioned that Pre-Cog only allows for a few seconds of vision into the future. So let us assume our protagonist knows this - and so prepares the fight with a series of long-term (e.g. past the point of Precog vision) items to give them the victory.
If my opponent can see 2 seconds into the future, but they do not know that 4 seconds away (at a run) I have hidden a rifle - by the time they are within precog vision of knowing I have a gun, they have at most 2 seconds to find cover or attempt to counter it - 2 seconds is not a lot of time to get out of line-of-sight or find hard cover. I am using a Rifle just for illustrative purposes - luring them into a trap where they will not be able to escape the perimeter of the trap in the time that precog allows will still work.
In addition - is your Precog ability always on - or is it something that you have to consciously summon - if it is the latter, then you could have some form of cool-down (e.g. it takes a certain amount of effort to cast it and you need to recover a bit before casting it again) - in which case, a Precog vs Precog fight would revolve around who can cast it in the most tactically advantageous moment.
Finally - and I kinda touched on this above - there are limits to human agility and speed. Those that have ever been in a Car Crash will know that there is a moment when you can see what is about to happen, yet you have no ability to significantly change the outcome - all the precog in the world is not going to help you if an Avalanche is thundering towards you or a massive explosion goes off - no matter what you do, you cannot get out of the danger area fast enough with only a few seconds.
Manipulating this fact is how two precog fighters with always on precog would attempt to battle - getting them into a position where 2 seconds is not enough to escape their fate.