According to their site, they realized their mistake with season 1 and 2. See, the big issue (for me) is that all they were comparing was weapon data and that was it. Also, the Spartan vs. Ninja was actually more of a toss up. Ninja aren't completely silent when stealthing (no one is) and the Spartan was trained to wake at a hair trigger of any 'odd' sound (what constitutes as odd was left to the imagination) so it would have probably be closer to 50/50, than the utter raping that happened.
The big turn off to me was the Knight vs. Pirate. A) Pirates are NOT warriors. They're raiders, they'd have run away long before the Knight showed up. B) the Grenado was not something they used in a fight. C) The technology gap was too wide. The Gonne was designed to defeat plate, long before the 18th century Pirate even existed, giving him an unfair advantage from the beginning.
However, it was cool to see that broadsword myth busted. Damn, that thing sheared right through that pig clean! The saber/cutlass tore it's way through, leaving a ragged, if equally deadly, cut.
Season 3, they claimed would add things like tendancies, tactics and skills to the mix of data. Haven't been able to watch any more episodes to see what sort of difference their talk made.