Anyway, nobody (well, no one good) has such large groups of units.
Using a high number of soldiers over a huge number of battles allows to reliably use the results to infer general features of the game, since it limits the impact of randomness.
Also, since the battles are head on between soldiers that are simply told to slaughter each other, there is no need to split the troops.
It may be done anyway, though... I'll post the results after altering the script so that it uses split troops.
And fights are seldom head on. Also, consider that there is lots of shooting before a fight (bowmen shooting and going back). There are so many tactics at play here that aren't considered.
True, there may be other advantages for Sword Fighters, that I haven't been able to standardize and automate over hundreds of battles.
But in these tests Sword Fighters have been given 3 clear advantages, and still didn't prove to be better than Pikemen... this is a strong proof against the notion of Sword Fighters being better than Pikemen.
It means that you need
at least 4 advantages for them to start being more useful than Pikemen.
It means that Pikemen are generally stronger melee warriors, not that they are so universally superior that they outshine anything else in every situation.
Also, please note that not acknowledging this is much akin to claiming that Knights are worthless because you might face an army double your own, entirely made of Pikemen and a few CrossBowmen, while you have no ranged support... sure, you
might.
But when you recognize that such an army is being made you can stop recruiting Knights and adapt earlier.
Likewise, if you recognize the presence of many of the factors that limit the power of Pikemen (lots of tight spaces, no enemy cavalry whatsoever, high numbers of Bowmen, etc...) you
must consider changing you strategy accordingly and recruit someone else than Pikemen, or you'll deserve your defeat.