Hey:)
I'm sorry for the delay, i reckon i didn't see the posts earlier.
I understand what you mean PuppyChow. If one has way too many fights against one specific FG compared to all the others (say up to 10 fights against each FG but 30 fights against one very easy, or one very hard), then the total win rate will be influenced alot by this unequilibrium between battles.
But, for a fact, if one tends to go against more of a specific group of FGs and less against another (or if FGs have different chance of being fought against between each others), its certain that for that player, it matters more to beat the FGs he uses to face most of the time.
Although with same deck you and i can face different (common) FGs (For instance i myself go alot against miracle sometimes. And seism. But i almost never see Ferox. I even thought it was taken off FGs list for some time).
Averaging the single win rates is also not a bad way but it's not perfect also.
Hum. I'm thinking about it and indeed i don't know which way would be the better. Averaging the win rates can also be bad if we do indeed face some FGs more often than others.
For cases like this one (
http://elementsstatistics.comxa.com/getstatistics.php?dv=107358030), we have the following average of win ratios:
(100+100+0+100+0+80+80+100+33+57+86+22+33+56+60)/15 = 907/15 = 60.47% win rate (sorted by total of games played). According to the actual system: 55%.
Now, say you'll face Paradox (no, it didn't exist in 1.17 but the case would be similar if you had never faced Obliterator: You faced him only once. That was close to happening).
So, according to the current system:
if you win: (21+27+1)/88 ~ 55.68%
if you lose: (21+27)/88 ~54.55%
According to the win-rate-avg:
if you win:
100+(100+100+0+100+0+80+80+100+33+57+86+22+33+56+60)/16 = 1007/16 ~ 62.94%
if you lose:
0+(100+100+0+100+0+80+80+100+33+57+86+22+33+56+60)/16 = 907/16 ~ 56.69%
With 88 games played, the result of this battle may vary the total win rate from 54.55% to 55.68% (current system), or from 56.69% to 62.94% (win-rate average). Should one single game out of 88 be responsible for such a big difference (out of 88 games) ~ 6 to 7% with the win-rate average system? I'm not sure.
Of course, this is normal to happen (mathematically) if we calculate this way since we have a totally new entry in the formula.
This is quite interesting though, is there a better way to calculate the total win rate ? Hum. Either way i would like to read more opinions regarding this.
