Sure it does. It gives you the cost to kill per mob. You have incorrectly assumed, or falsely claimed, that mob cost to kill is the same.
You spent such meticulous time poring over the loot, you forgot the most important bit.
Now you're trying to guess what f(x) is without even knowing x.
Why is it so difficult to share the most crucial part of the data? I can only assume it is because you and your colleague wish to purport a false perception of the loot profile and the importance of buffs in Version 2.0
This view is formed from the evidence that you have taken a passive agressive stance on sharing the most vital aspect of the data.
At this point in time, we don't even know what your TT% return is, because you refuse to state it.
Anyways I'm not going to push the point any further. Create another thread with properly referenced and logged data and I might take a look.
/Unsubscribe
Hello Immortal,
I somewhat understand what you are saying. Allow me to try and convince you that I have actually shared the cost by sharing gear information. Sorry I have not responded earlier, but have been somewhat tired due to a trip.
You will see that I presented my gear in both cases. (I also used a kill shot weapon to try to minimize over kill during the experiment). This was to try and determine the relative difference in cost to kill in the two parts of the experiment. I assume, and I think it is a good assumption, that the buffs increase dpp/decrease cost by the following formula
(%focused blows)*(1+increased crit damge%)+(2%)*(increased crit damage%).
It costs about 9% less to kill a caperons with the buffs. As stated in original post.
I also include dps change numbers (dps improves by ~20% with buffs), and established that regen is neglible compared to the effect of buffs on kill cost ( <1%).
For my experiment, I do not make any conclusions about absolute kill cost values, only relative. So I know cost to kill a caperon decreases by at least 9%, possibly 10% when you account for regen, when you use buffs. I do not assume that cost is directly associated with the loot return from a mob, as that is what I am seeking to measure in my experiment, is if I can see the impact of decreasing cost on size of mob loots. In fact, I do not even bother to try and capture kill cost. I assume my play style as an avatar is the same in both cases, and over many mobs the kill cost will reflect the change in cost due to buffs mainly. I assume kill cost decreases by the % i calculate above though. (I didn't post cost because I simply did not think it was necessary to track or important. Only the relative cost mattered, and I knew the effect of the buffs on cost well.) If someone wants to repeat the experiment and incorporate kill cost be my guest, but I do not think it will change the conclusions.
I draw the conclusion that buffs decrease cost, and this decrease in cost is reflected in decreased loots proportionally (approx same %).
For many of those who complain about sample size. I assume that the multiplier distribution is the same in both cases. (There will be x% of 1x, y% of 30x, z% of 300x as an example,etc) I also assume (obviously) that the percent of the smallest multipliers (those less than 1x) will substantially outweigh the percent chance of receiving large multiplier. Taking 500 or so of these samples in both cases, and neglecting the low % chance higher multipliers by neglecting the higher multis, you can get an excellent sense of the distribution and mean of the high % chance loots (those less than the expected kill cost of the mob). If you look on my spreadsheet you can see how I do this analysis. Since the percent chance of the small multiplier loots are so high, N = 500 is sufficient to get a pretty good sense of how the distribution looks like (you get lots of data points in the histo). Summing then serves to further remove probabilistic fluctuations (small meaningful differences add up, but unmeaningful fluctuations cancel). If someone wants to try N=1000 or N = 2000 please be my guest but I think N=500 on either side of the column is quite sufficient for this experiment.
Now if you want to assume the multiplier table %'s are somehow different in either case then yes, you would have to include all sample values, not being able to neglect outliers, you would indeed have to cycle many months to see the trends, with the same setup, for both cases.
As for tt return. I was at 101-2% tt return prior to 2.0 since tracking (~500k cycled), using majority of the time ares modified + EST. Since 2.0, I am using ares improved and not using EST (using arc spark + VI improved mainly). I am currently at 100-1% return (because i'm lucky probably lol), I have only cycled about 25k however.
Please feel free to reply with any further questions, I will try to answer as I can.
Update: Back down to 97.7% return. But at least I'm not counting conversion to shrapnel in that %
. Time to go shoot my adj maddox.