Some statistics.

Right after the author of this topic created it, I wrote almost an entire article regarding the issue. And then I deleted it, without even publishing it. It turned out to be too complicated.
I'll try again.

In general, the point is this:
a) when a Strong Player plays, no matter which Castle they choose, they increase the Castle's Strength with their reputation.
b) when a Weak Player plays, no matter which Castle they choose, they decrease (discredit) the Castle's Strength with their reputation.

Isn't that true? A Strong Player can defeat a Weak Player, even if the Strong Player is using a "weak" Castle and the Weak Player is using a "strong" Castle.

Taking the above into account, I propose:
a) based on the ranking table of online games subjected to statistical research for the season,
b) assign a coefficient (I think it can be called a "discretization" [or "accreditation"] coefficient), the essence of which is that we don't just sum up all the Castles, but we sum them up, "reducing" the Castle's strength if a Strong Player played with it, and vice versa.

For example, assign the following coefficients:
1st place in the ranking - 0.80
2nd place in the ranking - 0.82
3rd place in the ranking - 0.84
...
21st place in the ranking - 1.20

And now, let's say that players 1, 3, and 21 played with the Swamp Castle.
Players 1 and 3 won, and player 21 lost.
And then, in the overall statistics of won/lost games for the Swamp Castle, it will not be 2:1, but (1*0.80 + 1*0.84) : (1.2) = 1.64 : 1.2

Do you understand that the provided statistics do not take into account the strength of the players who play with a particular Castle, and who, accordingly, unintentionally introduce a bias into the statistics.

I wonder if anyone has the strength and patience to do such painstaking work and show what the statistics really are?