hammy wrote:
I do appreciate the problem. The thing is that what temsd to happen is a few decent players find an army that suits them and that they know how to play. The army makes its way up to near the top of the rankings and then get noticed by lesser players who do badly with them and as a result the rankings of the army drops. Just look at the volatility of the army ELO graphs. I am not sure how much can actually be read inot the army stats as they stand.
The army ELO does provide interesting reading but if you compare the top 10 armies now with the top 10 6 months ago you are quite likely to see significant changes.
Not to much and we stated just that in the FAQ (you did read that, right?)
There is a reason why the preselected sorting is by popularity and not ELO, you know.
Even if we included a player ELO score, the discussion would then be how to weight it properly. In a way the army ELO scores often normalize themselves somewhat. Because as you stated, after some good players used an army to good effect a number of other players try to emulate them with various levels of "success". So army ELOs are more of an roller coaster then player ELOs, but with enough samples the end results
may still have some significance.
What might be possible is to give an average of the ELO scores of the players that used the army. I.e. current army ELO is 1520, average player ELO of the players that used the army is 1640. Something like that might help a bit, still the problem remains that I would have to use the current player ELO scores instead of the player ELO at the time of using the army.
