View Single Post

KeyboardNinja's Avatar

10.29.2012 , 12:13 AM | #3
Quote: Originally Posted by grallmate View Post
I'd also suggest tweaking the 'death threshold' to 9 unmitigated hits. Given that even if the 10th hit is shielded (but not when defended) it is still going to kill the tank. Arguably, shielding 1 hit and not defending any in a period of 10 hits is still going to kill you, be it the first, middle or last hit. But for the sake of simplicity I'd stay with 9 sequential unmitigated hits. This gives a probability of 0.00091321.
I thought about that, but then we get into dicy arguments about "odds that you shield, but not defend", etc. It's not really effective HP anymore; it's more like expected TTK. This is, arguably, a more interesting metric, but also one that is much harder to work with. It also would tend to lengthen the TTK (due to lower expected damage) rather than shorten it as you suggest.

Quote: Originally Posted by grallmate View Post
I think you've made an error using 19.5 distinct time periods as well, the chance of seeing the worst case scenario is much higher than you have stated. It is possible for a string of hits to occur at any point after the last mitigated swing providing that there are still a sufficient number of swings remaining in the encounter.

Given the average encounter (using your numbers) has 195 swings, the string must begin before or on the 186th swing. as such, the probability of it not occurring in the hypothetical average fight is more like:
0.99908679 ^ 186 = 0.84372028
This gives a 0.15627972 chance of a string of 9 unmitigated hits occurring in an average fight. As such, doing 10 bosses a week you have about an 80% chance to see it each week. However, as Tam sated, its not really noticeable most of the time.
I'm not sure about that expression. Need to think about it. Seems like you're double-counting things.

Quote: Originally Posted by grallmate View Post
On another note, I also feel this discussion benefits more from keeping E/I eHP and K/E eHP separate since eHP has much much greater value in E/I heavy fights. Not to the point that I would ignore mitigation, but the relative value is definitely skewed. Although I do see the benefits of using a hybridised figure for fights that use both damage types relatively evenly (up to about 70:30 split), but I'd personally calculate it on an encounter by encounter basis using the excellent spreadsheets you've provided.
Ideally, all theory-crafting would be done on a fight-by-fight basis (or even phase-by-phase). The problem is that I don't know anyone who remods their gear between fights. Even doing simple stuff like swapping relics is extremely rare. I'm the only person I know who re-gears mid-fight. Overall, people seem to like the "one-size-fits-all" gearing strategy. Makes sense from a big picture standpoint, but it does tend to miss some of the finer points of optimization.
Computer Programmer. Theory Crafter. Dilettante on The Ebon Hawk.
Tam (shadow tank) Tov-ren (commando healer) Aveo (retired sentinel) Nimri (ruffian scoundrel)
Averith (marksman sniper) Alish (lightning sorcerer) Aresham (vengeance jugg) Effek (pyro pt)

December 13, 2011 to January 30, 2017