Tested Vlad for Proc Hits

I have not done a boss3 test for a few days because I ran out of things I was curious about that I could test.  (There are several heroes I'd love to test but I can't since I don't have.)  But as I was cheesing in the recent guild war, I was reminded again of how nasty Vlad is.  It always seems like he gets a lot of hits.  So I wondered if our understanding of how he works is correct.  Hey, I can test it!

Setup for Boss3 test of Vlad

The setup is a bit more complicated than the vanilla Boss3 test, because I was afraid that Vlad's proc would give the boss enough energy to proc.  So I added my HQ to make sure the boss stayed flat.  (My HQ has scatter 8.)  I was not sure if this would cause the test run to end early, so I watched carefully.  It didn't, though: the test went the full three minutes.  Here's the result:

Is this too much? 

99394792 damage.

My earlier test of just PD+HQ got 64953661 damage.  So we can see the delta from adding Vlad is 34441131 damage.  Vlad has 1200ms attack speed, so as in the previous tests, I model his ramp-up as 5 seconds of nothing followed by full damage.  So, 175s.  Assuming all hits are full damage, we find the number of hits by dividing by 35000: that's ~984 hits.

In 175s, there should be 875 auto-hits.  So, we find ~109 hits from procs.  Vlad procs every 6 seconds, so we estimate 29.16 procs during the run.  Thus, he's hitting ~3.74 times per proc.

The wikia article on Vlad says that his proc hits every 0.4s, for 4 seconds, with a 30% chance per pulse to hit.  That would suggest he averages 3 hits per proc, but that is clearly ruled out by my test.

One anomaly about Vlad is his fear infliction.  I don't think it varies when it happens, and I think it happens immediately every time.  So I'd suggest a different model for his proc: in the tick that it procs, it hits all enemies in range and inflicts fear.  For the remaining time, it works as wikia indicates (9 pulses, 30% chance per pulse).  In this model, we expect 1 guaranteed hit plus an average of 2.7 over the rest of the time.  This gives an expected 3.7 hits, which is within experimental error.