If you really want to test Mongoose, simply use a caster offhand (the orbs, books, etc), this will stop you from attacking with it. If you weapon is broken, you use your fist; doesnt that count toward the number of hits you got?

My original findings were that, on a level 70 dummy (to avoid misses), I would get a PPM between 1.5 and 2. However, when hitting a boss dummy it would be (with a minimal sample mind you) closer to 1. Thus, my question is: does the level of the dummy impact the PPM value? Aside from the fact you are missing more, what is the % that a HIT will proc?

~4600 attacks may be a too short test duration, at least I can imagine that there is space for variations. To give you an example: When I was doing this 5 hour test, I was surprised to see a glancing rate of 22.2% after ~3500 attacks. That was 1.8% less than expected. ~3500 attacks were simply too few, to give a distinct result of the real glancing rate. Maybe you can do an other test of longer duration with the same setting and spec to double check if there is a relation to Assassination, or if it was just "lucky"?

Unless I've misread something, Savage Combat (in the buff section) is currently listed as giving only a 0.01 multiplier per talent point when it should be giving 0.02.

I just noticed this.

In the Player/Raid buffs section, where the Melee Haste option is.. it says Improved Icy Touch. Imp IT is a thunderclap-type debuff on the boss, Imp Icy Talons is the haste buff.

Not really a big deal, just thought i'd point it out.

Usually, having a smaller sample space will only result in your confidence interval being wider. For example, the first test I concluded had about 2000 hits; it gave me an average PPM of 1.75 with a confidence interval 19 times out of 20 of 1.50 and 2 (thats all from memory). This would indicate that with a 95% certainty you can say your PPM is between 1.5 and 2. For it to actually be 1 would require you to be not only unlucky, but EXTREMELY unlucky (emphasis on the extreme which essentially means it doesnt happen). That was all on a level 70 dummy.

HOWEVER, and this is my question, could the level of the dummy actually influence your proc rate? Doing a test with only 300 swings gave me a PPM of roughly 1 on a boss dummy. Now 300 is very small, my confidence interval would probably have been between 0 and 2 PPM so... What is it worth? Not much.

You should not be changing that cell, because it automatically determines its value based on whether you have the Hyperspeed Accelerators equipped on the "Gear" sheet or not. If you change it manually, I can't guarantee anything about the results.

This change is not going in until patch 3.1.0.

In the current patch it is .01. Next patch it will be .02, but for the moment it's only .01.

True, though low-probability events exhibit less variance, so we get better data on Mongoose than we would on Glancing in 4600 attacks. More is, of course, always better - but it's a start. In terms of grinding the actual numbers:

Running Lokkun's data shows the proc rate to be 95% likely to be between .878 and 1.104 PPM.

DMM's Shiv test: as I recall PPM effects aren't normalized to 1.7 on instant attacks, they're simply not haste-adjusted. So I'd expect the proc rate there to be 1.3/60 if it's 1 PPM. In terms of the actual data, the range gotten is 1.82 to 3.13 PPM.

DMM's autoattack test: 95% likely to be between 1.286 and 1.908 PPM.

So... we have 3 different tests, and 3 almost mutually distinct probability ranges. Clearly, there is some mechanic in play, or an error with one of the tests.

I guess my first question would be for DMM: how were you counting procs? Is it possible that you were counting both gains and fades, and thus finding (almost) twice as many procs as were actually occurring? Could you give the breakdown of gains/refreshes/fades?

Edit: Just grabbed DMM's part 2, and am counting 52 procs in 3068 hits, which gives a 95% confidence interval of .643 to 1.118 PPM. So I suspect it is a gain/fade doublecounting thing that is causing his results to be different.

I've been able to theorize a maximum Mutilate dps of

5539 DPS
300 second duration
Non-murderable mob

Has anyone beat that on their spreadsheet of Best in slot gear?

vulajins spreadsheet btw

Getting 5674.92 with best in slot gear, and full raid buffs(flask, 1 lust, speed pot, AP food) as a Troll.

5706.92 DPS BiS on everything except 2nd trinket (FotFF never drops for me).
300 sec fight duration, blood elf, all buffs (except heroic presence).
WD/WD IP/IP 5e/5r, ToTT not used. insc/JC proffs.

I'm not interesting in having this thread turn into a "who can win the spreadsheet" dick-waving exercise. Please don't take it that direction unless you're going to reveal that you came up with a new theoretical DPS maximum involving a non-obvious cycle or some gear selection that isn't already covered in the "T7.25 Rogue Gear Discussion" thread.

The post immediately above mine would be an example of the kind of discussion that is actually worthwhile, since it concerns the differences between an IP/IP Eviscerate cycle and an IP/DP Envenom cycle.

(edit) I will be infracting any posts in this thread from here on out that simply list your DPS and a blurb about the gear/buffs used if you don't specify how your finding is important or useful to others.

I'm not entirely clear what you're trying to say here, but it seems like you are getting higher DPS output when you specify "use Envenom" even though you don't have DP equipped. That is because the Envenom calculations do not automatically zero out just because you don't have DP equipped. Using the Envenom damage formula in the sheet, an Envenom performed at 0 DP stacks (even though it's technically impossible) will still deal substantial damage. Plus, you'll still get the Envenom buff bonus to your poison procs, even though you technically shouldn't. So yes, selecting "use Envenom" when you don't have DP equipped will still increase your DPS, but only because it is modeling an impossible situation.