Melee Crit Chance
 

I recently produced many lengthy logs to parse melee critical hit chance accurately. Here are the results:

https://i.imgur.com/TSH9c9A.png
https://imgur.com/TSH9c9A

These logs had around 150k-700k hits each. They required that many hits to reduce the margin of error sufficiently. I came up with a function that fits the data well.

I had also made a couple of Ranger melee logs, and the results indicated that they might have a slightly higher critical chance, but I cannot be certain without producing 1-2 weeks worth of data and I'm unwilling to do so.

https://i.imgur.com/pKpMOW3.png
https://imgur.com/pKpMOW3

I also had some lengthy logs produced in 2003 that I made while attacking non-combative non-killable NPCs. The data is too thin to make conclusions, but the evidence suggests that all of the classes may have had the same multipliers from AAs back then instead of different ones like they do now. The warrior data point was from half a dozen overnight parses so it's very accurate, but it parsed well below what modern warriors should parse at with CF6. (I didn't parse CF6 on Live) The line on the chart is what warrior criticals would look like if warriors used the weaker multiplier that monks and knights get. The rogue log I had also parsed about equal to that of the monk even while on Live rogues get significantly higher multipliers at CF4-6. The paladin data points were from an old 2006 web page found online here http://giline.versus.jp/shiden/su2_e.htm

https://i.imgur.com/T6IMkDC.png
https://imgur.com/T6IMkDC

Archery logs took forever to produce. Three of the data points were each 2-3 days of logs and the rest were 1 full day. I suspect that if I had more data that the function to fit it would look slightly different due to these logs having far less hits as the melee logs, so margin of error is greater. Although we're talking tenths of a percent here, so users wouldn't notice.

Old EQ logs indicated a crit rate of roughly 22-23% with full PoP AAs, so that compliments the above base crit chance and a 150% AA multiplier like the melee data did. Crit chance before AAs in old logs also seemed to agree with the above data (or at least was way higher than melee rates) so the base crit rate was probably the same function then.