Bahadori Posted July 19, 2015 Share Posted July 19, 2015 (edited) Post has been pulled down for further testing please refer to the following update. Update: Hey guys. I'm going to be pulling this post down. It appears, when reviewing my stats in game, that I DID NOT have the amount of Alacrity required. This was my own fault, not sure how I messed it up but it appears the parses and live raid fights I did were with about 70% of the suggested Alacrity amount. I'll be redoing all my parses and HM fights and will post the updated results once I have all the numbers down. Should be about this time next week. Sorry for the mixup guys. Thank you for your understanding Edited July 20, 2015 by Bahadori Link to comment Share on other sites More sharing options...
zhaiyan Posted July 19, 2015 Share Posted July 19, 2015 nice result, thanks dude! besides, as to the dps difference, there seems to be multiple changing factors. I am just wondering will build A benefit more if some of the stat can be put on crits? To me the crit chance benefits the main filler massacre, and might create some difference. Just personal guessing tho, I am far from having a full set of 198 atm... Link to comment Share on other sites More sharing options...
Bahadori Posted July 20, 2015 Author Share Posted July 20, 2015 nice result, thanks dude! besides, as to the dps difference, there seems to be multiple changing factors. I am just wondering will build A benefit more if some of the stat can be put on crits? To me the crit chance benefits the main filler massacre, and might create some difference. Just personal guessing tho, I am far from having a full set of 198 atm... Possibly but to be honest Crit grants little return either + or - and then again it's completely RNG reliant so on a good parse or fight with some Crit Build A may pull out ahead. Link to comment Share on other sites More sharing options...
Arkhonos Posted July 20, 2015 Share Posted July 20, 2015 You would need quite literally hundreds of parses with the exact same rotation to make relatively confident conclusions about which stats are better for DPS for just 1 spec. 5 parses mean nothing because in SWTOR optimal stats influence DPS a few percentage points. In terms of contribution to variance in DPS, the following are all more important than stat optimization: Relic procs uptimeRelic procs timing (e.g. during your gore window)Overall crit rateCrit rate on hard-hitting abilitiesAPM / server lagIndividual damage rolls on hard hitting abilitiesSpec RNG (e.g. Lightning/TK) These are just a few factors that all are ultimately more important than stat optimization for explaining parse-to-parse variation in DPS (and I am almost definitely forgetting some other important things right now). So, to compare DPS across stat builds you would need enough parses to remove the influence of all those other factors while only changing your stats. In the end you would find a very minute difference between the builds. Needless to say, testing stat optimization in a raid is completely pointless because the amount of moving parts will always influence DPS significantly more than any stat optimization will. I'm not sure I totally trust the results from the thread you are referencing, but I can say with certainty that this testing is insufficient for drawing conclusions about those results because the sample size is too small. Link to comment Share on other sites More sharing options...
Bahadori Posted July 20, 2015 Author Share Posted July 20, 2015 You would need quite literally hundreds of parses with the exact same rotation to make relatively confident conclusions about which stats are better for DPS for just 1 spec. 5 parses mean nothing because in SWTOR optimal stats influence DPS a few percentage points. In terms of contribution to variance in DPS, the following are all more important than stat optimization: Relic procs uptimeRelic procs timing (e.g. during your gore window)Overall crit rateCrit rate on hard-hitting abilitiesAPM / server lagIndividual damage rolls on hard hitting abilitiesSpec RNG (e.g. Lightning/TK) These are just a few factors that all are ultimately more important than stat optimization for explaining parse-to-parse variation in DPS (and I am almost definitely forgetting some other important things right now). So, to compare DPS across stat builds you would need enough parses to remove the influence of all those other factors while only changing your stats. In the end you would find a very minute difference between the builds. Needless to say, testing stat optimization in a raid is completely pointless because the amount of moving parts will always influence DPS significantly more than any stat optimization will. I'm not sure I totally trust the results from the thread you are referencing, but I can say with certainty that this testing is insufficient for drawing conclusions about those results because the sample size is too small. Damn Loffi! LOL Minus 500 DKP! Literally though I said pretty much everything you just did in the "Interpretation and Opinion of Results" part of the post, you just went WAY more in detail than i did with it. I've had far to many people ask me my opinion on the build they posted in that optimized stat thread and this was my personal response to it. I put the gear in, did the same thing i do every week, and that is what's shown. No misrepresentation of information was given, i was very forward about the "best" way to do it legitimately, though i doubt any one person would have the time required to do it. Link to comment Share on other sites More sharing options...
Arkhonos Posted July 20, 2015 Share Posted July 20, 2015 Well, you did say the following: For the casual player or mid level player I would have a hard time saying one build is absolutely better than the other, 7.0% difference between the two builds is probably not a concern for players in those brackets. For endgame progression players I would personally suggest Build B over Build A and: However in the interest of time and weighing heavily on personal experience I felt this was sufficient enough to give some kind of personal opinion on Build A The issue with this is that it appears you are suggesting a stat build based off of your measured 7% DPS difference (which I would consider an extremely significant amount). However, as I pointed out this data is a drop in the bucket compared to what is necessary to make anything close to meaningful conclusions. You may as well just leave out the measured data as it is misleading. Non-mathematical observations such as it appeared that Build A when put under mechanics involving heavy swapping suffered significantly more than fights that imitated "dummy parsing" are more interesting because they are definitely relevant to players and do not require hundreds of parses worth of data to discuss empirically. Link to comment Share on other sites More sharing options...
WayOfTheWarriorx Posted July 20, 2015 Share Posted July 20, 2015 (edited) You would need quite literally hundreds of parses with the exact same rotation to make relatively confident conclusions about which stats are better for DPS for just 1 spec. 5 parses mean nothing because in SWTOR optimal stats influence DPS a few percentage points. In terms of contribution to variance in DPS, the following are all more important than stat optimization: Relic procs uptimeRelic procs timing (e.g. during your gore window)Overall crit rateCrit rate on hard-hitting abilitiesAPM / server lagIndividual damage rolls on hard hitting abilitiesSpec RNG (e.g. Lightning/TK) These are just a few factors that all are ultimately more important than stat optimization for explaining parse-to-parse variation in DPS (and I am almost definitely forgetting some other important things right now). So, to compare DPS across stat builds you would need enough parses to remove the influence of all those other factors while only changing your stats. In the end you would find a very minute difference between the builds. Needless to say, testing stat optimization in a raid is completely pointless because the amount of moving parts will always influence DPS significantly more than any stat optimization will. I'm not sure I totally trust the results from the thread you are referencing, but I can say with certainty that this testing is insufficient for drawing conclusions about those results because the sample size is too small. He used the term "personal opinion". From that it should be fairly obvious that he is in no way trying to give the impression of certainty. He gave the results of 2 different builds of 5 parses each and the percentage difference of the 2 builds used. It wouldn't matter if one parsed each build 10000 times. For the variables you sited in your post would exist every single time no matter the build and thus one could never be assured of outcome no matter how many times they used the builds. ---------------- I would like to add one non scientific finding of a similar note. All the variables you sited are absolutely the case. I would however argue that requiring such an expansive number of parses as you suggest is not always necessary to come to a finding one can feel a good deal secure with. That said.... I have recently been experimenting with this alacrity build that Hayete used in this test. (My numbers stat wise weren't exact as I do not have the 198 (Revanite) alacrity pieces, nor am 100% BiS). Stats were very close though. While I cannot reach the numbers Hayete can achieve (in any build) overall, there is absolutely no question in my mind that it is a straight dps gain. I have been able to achieve numbers that I previously never could and have done so many times now. I parse everyday for at least an hour (I have been doing this for about 6 months) so the numbers involved have long since been established with regard to what number ranges I can achieve on a dummy parse. The numbers I have been able to achieve with the alacrity build (similar to the one suggested), have noticeably improved. My rotation has not been altered. While I sop pose one could state that as a HM end game progression raider (not as successful a one as Hayete) that experience gained in that venue could account for some of a dps increase, the number's changed considerably the very second I switched to the alacrity build and have stayed in that new, higher range. My latest highest parse is over 100 more DPS than my previous highest parse pre-alacrity build and I have been able to achieve this new post alacrity build parse high numerous times. Is this scientific? No, of course not. But I can't imagine coming to any other conclusion other than the alacrity build is the basis of the DPS increase. I have not yet used the alacrity build in HM Ravagers or ToS as I am not yet 100% acclimated to the change. I'd like to get some more practice in with it before moving the the build into the HM OPs. There are so many variables involved in DPS numbers. I couldn't agree with you more on that. I consider myself a decent player with a fairly good grasp on the use of a Carnage Marauder. Yet, I cannot come close to achieving the numbers Hayete can. I cannot blame the rotation on that. I am not quite as well geared as Hayete (I want that 204 main hand!!!), but half my gear is Revanite and the rest is by no means sub-par and naturally I have the 6 piece set bonus. My main-hand damage bonus is between 2406-2421 (I play around with augs sometimes) and I run zero crit. I come in 1st or 2nd DPS wise in my raid group equally, this being prior to the alacrity build. (Though our Merc usually does the barrels and runs the bombs during those fights). You are obviously intelligent and there is no one's opinion I hold in higher regard with respect to the Carnage spec than Hayete's. But as far as I am concerned, the alacrity build is a straight up DPS increase. Not a scientific finding, but my firm and honest belief. Edited July 20, 2015 by WayOfTheWarriorx Link to comment Share on other sites More sharing options...
Arkhonos Posted July 20, 2015 Share Posted July 20, 2015 It wouldn't matter if one parsed each build 10000 times. For the variables you sited in your post would exist every single time no matter the build and thus one could never be assured of outcome no matter how many times they used the builds. This is untrue. As you increase the amount of trials in an experiment, the likelihood that you reach the "true" average increases. So, if you did two sets of 1000 parses you would expect the average DPS of each set to be extremely close to each other, despite the variability between individual parses. Now, if you did 2 sets of 1000 parses and only changed your stat optimization between them, you would expect there to be a greater difference between the two sets. This would be explained mostly by the influence of stats instead of by the other factors I listed. Link to comment Share on other sites More sharing options...
Bahadori Posted July 20, 2015 Author Share Posted July 20, 2015 Update: Hey guys. I'm going to be pulling this post down. It appears, when reviewing my stats in game, that I DID NOT have the amount of Alacrity required. This was my own fault, not sure how I messed it up but it appears the parses and live raid fights I did were with about 70% of the suggested Alacrity amount. I'll be redoing all my parses and HM fights and will post the updated results once I have all the numbers down. Should be about this time next week. Sorry for the mixup guys. Thank you for your understanding Link to comment Share on other sites More sharing options...
Recommended Posts