same flaw as in HLstats in that case.
This depends on your point of view as you have to start somewhere if you wanna generate a number like "skill".
I've checked the scripts now and PS offers two kinds of skill-computation. One old Elo-based and a newer which is the default since PS3.1:
If you're as a killer better than your victim you'll get this as a bonus (**2 means powered by 2):
Code:
$kbonus = ($kskill + $vskill)**2 / $kskill**2;
So taking me as a victim you'll get:
Code:
$kbonus = (929 + 476)**2 / 929**2
$kbonus = 2.29
At the same time the victim will loose some points:
Code:
$vbonus = $kbonus * $vskill / ($vskill + $kskill);
So:
Code:
$vbonus = 2.29 * 476 / (476 + $929);
$vbonus = 0.775
So if you shoot me once your skill is increased by 2.29 and mine dropped by 0.775.
The other way round the code is different:
Code:
$kbonus = ($vskill + $kskill)**2 / $vskill**2 * $vskill / $kskill;
$vbonus = $kbonus * ($vskill + $self->{baseskill}) / ($vskill + $kskill);
So this will result into:
Code:
$kbonus = (476 + 929)**2 / 476**2 * 476 / 929;
$kbonus = 4.464
$vbonus = 4.464 * (476 + 50) / (476 + 929);
$vbonus = 1.671
So here my skill will be increased by 4.464 and yours decreased by 1.671.
These values however will be get modified (multiplied) with a weapon's weight if this option is activated(?) and have a total cap of 10.
If you now check your play sessions you'll see that the last sessions from the 7th and 9th have a K:D < 1. So you got killed more than you killed which has resulted in the overall decrease of your total skill. All the sessions with a K:D > 1 resulted in a steady skill increase. So I think that the stats are running fine.
The big problem is that the stats are only pure kill-stats (assists are counted too). So all stuff like capping do not count. I haven't dug any deeper in this, but you can define you own stats-computation for a kill-event and the assist-event. Perhaps it's possible to add other events there to.