So (and I hope I've remembered my stats correctly here). In those attempts I got 29 failures, or a 2.88% failure rate. So, I did some more runs on spider 20, a total of 56 runs in order to get the sample size of 1,008 debuff attempts. As far as I know from top end CB-teams and my own clan, 220 is believed to be sufficient to get at least so close to the max probability that the difference to 250 is not really observable any more. I think that the 250 recommended accuracy number has fairly generally been accepted as being too high these days. It should be fairly easy to build at least a fairly high number of observations, although with such a low probability being tested getting to a high confidence level will take a large sample as well. The nice thing with Draco and spider is that you actually get 18 data points on every run (2 debuffs on spider and each of the 8 spiderlings), and as there is no hit involved, it's only the acc/res test resulting in any failures. I've been running spider 20 with a high acc Draco, and my own estimation of the max probability has been around 95%, just from observing the runs. ![]() Nice work, and fits with at least my own anecdotal experience. If anyone wants to play around with the formula, it's here: If anyone has other data to share, please do so in comments. I took this data from my recent run of spider-20 with my debuffer missing about 6/176 runs. I estimate a 3.5% chance of failure no matter how high your acc is. ![]() 0,92 (at equal values, the advantage goes to attacker at 92% chance to land).-30,70 (at default values of 30 resistance and 0 accuracy, there's 30% chance to resist, and 70% chance to land debuff).-42,50 (at 42 more resistance than accuracy of attacker, defender has more than 50% chance of resisting).If anyone has a chance to test some values, let me know in comments! Over a high number of turns, you should be able to see the difference. As a point of reference, if you are off by -10 or +10, you'll see a big difference in landing chance with 34% and 68% chance to land. However if 168 ACC was enough to get 50% chance, then 230 would be optimal. Then from there, the optimal value for acc would be the boss' resistance + 20.Į.g., if the 50% chance happens with 188 ACC, then UNM CB has 230 resist, and the recommended value should be 250 (a possible value floated around). Once you find the accuracy that gives your debuff chance as close to 50% as possible, then we know the UNM boss will have that value + 42 of resistance. There are currently a number of people with unkillable teams, all it would take to test is to give one of your champions in the comp (who has a skill debuff in the rotation) a fixed amount of accuracy, record the run, and go back and see how many times it was resisted. So there is a way currently that we can check this easily (though time-consumming). This has huge implications in both PvP and PvE.įor example, if you're in high Plat arena, if you know an enemy has 500+ accuracy, there's no point in stacking resistance unless you're shooting for 460+ for it to even break even (50/50).įor clan boss, if we can accurately know the resistance of each CB, then we know the exactly what accuracy we need to be optimal. The optimal value seems to be to have an accuracy of 10-20 more than the enemy, or a resist of 80-90 or more of the enemy. Note a 90% chance to land debuff during resistance check means 10% resist chance and vice versa.Īs many have mentioned here before, stacking too much accuracy or resist for PvE content is pointless as you will never get your debuffs to land 100% (unless the skill specifically says cannot be resisted like Ghostborn's) or be able to resist debuffs 100%. y-axis is the % chance of accuracy winning over resistance and landing the debuff.x-axis is the value of (ACCURACY minus RESISTANCE).In it, it gave three datapoints for their formula, so I tried to create a graph that takes into account those data points.īefore I get into that, here's the graph in case anyone wants to skip the explanation / data / calculations: So the downtime got me digging around some youtube videos and I found one on accuracy vs resistance by RSL official account ( ). Formula gives a methodical way to find out an enemy's actual resistance through testing. Don't stack too much ACC / RES for the content you're at.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |