Softcaps discussion

[Freehugs]

Yeah good work Style. That's enough to say that if the cap isn't 8%, it's very close to it.

[LagunaD]

(01-24-2012 08:11 AM)Freehugs Wrote:  Yeah good work Style. That's enough to say that if the cap isn't 8%, it's very close to it.

Within a few percent, probably.

Without the actual number of ranged, main-hand special attacks, you can't draw further conclusions, for the reasons I explained above. Not trying to knock the tester, it is just a mathematical fact.

We might be able to estimate the number of attacks with more information on the abilities used. Style only mentions Unload and the basic attack. If these were the only ranged abilities used in "about 6 minutes", Unload has a 15 second c/d, so it could be used 24 times. Each use makes 4 main-hand attacks. So that is 96 main-hand specials. Rail Shot may have also been used, which would add another 24 main-hand attacks in 6 minutes. So we are up to about 120 main-hand, ranged, special attacks. This does not account for pushback on Unload, which can be reduced, but not eliminated, by talents (assuming Style had aggro, which he didn't say).

There are Arsenal talents that allow the cooldown of Unload to be reset; I don't think they will double the number of uses, though, and we don't know whether Style is spec'ed that way and used the corresponding ability rotation. Power Shot is also Ranged, although Arsenal mercenaries would generally use Tracer Missile (which is Tech) instead. If he used Power Shot, that could increase the number of ranged attacks (which would be limited by heat, at that point).

It is reported that with 7.94% Accuracy (0.06% miss chance, if the Accuracy cap is 8%), one Unload attack missed. With 0.06% miss chance, 0.072 misses are expected in 120 ranged, main-hand special attacks. And Style saw 1 miss, which is extremely unlikely if his miss chance was only 0.06%.

Putting it another way: to expect 1 miss with a 0.06% miss chance, you would need to make about 1667 ranged, main-hand special attacks.

Ignoring talents and pushback, and assuming you do nothing but Unload, Rail Shot and Power Shot spam as filler (which is impossible in practice due to Heat, I think, but it gives us fairly optimistic numbers), you could make 12 ranged main-hand special attacks every 15 seconds, and it would take 2083 seconds (about 35 minutes) to expect 1 miss when your miss chance is 0.06%.

Now, maybe he just got unlucky with that one miss at 7.94% Accuracy. But on the other hand, maybe he just got lucky with no misses at 8.03%. This is why statistics are necessary.

In any case, more information about the abilities used, spec, and whether he had aggro, would help estimate the number of relevant attacks better.

[Style]

(01-24-2012 11:27 AM)LagunaD Wrote:  Style only mentions Unload and the basic attack.

... I used my full rotation, TM, HSM, RS, DFA, UL, RS, etc.

[LagunaD]

(01-24-2012 09:02 PM)Style Wrote:  

(01-24-2012 11:27 AM)LagunaD Wrote:  Style only mentions Unload and the basic attack.

... I used my full rotation, TM, HSM, RS, DFA, UL, RS, etc.

OK, thanks.

So the only ranged special attacks would be Unload and Rail Shot then?

Was the target being tanked by someone else, or attacking you?

[Style]

(01-24-2012 09:12 PM)LagunaD Wrote:  

(01-24-2012 09:02 PM)Style Wrote:  

(01-24-2012 11:27 AM)LagunaD Wrote:  Style only mentions Unload and the basic attack.

... I used my full rotation, TM, HSM, RS, DFA, UL, RS, etc.

OK, thanks.

So the only ranged special attacks would be Unload and Rail Shot then?

Was the target being tanked by someone else, or attacking you?

The tank. I took a few hits every now and then from random attacks.

[LagunaD]

Based on Style's information, here's a statistical estimate.

I assume the fights lasted 360 seconds, Rail Shot was used every 18s, and Unload was used every 10 seconds (thanks to the Barrage talent).

That makes 164 main-hand, ranged special attacks per trial.

I did a maximum likelihood fit combining the data from both runs. This uses all available information.

I also tried varying the number of attacks by +/-25%, to see how it would affect the results. The likelihood functions are shown below.

For 164 attacks, the best fit value for the cap is 8.25 +/- 0.31%. 8.85% is as likely as 8.00%.

If the number of attacks was 25% higher, the best fit is 8.18 +/- 0.24%. 8.60% is as likely as 8.00%.

If the number of attacks was 25% lower, the best fit is 8.35 +/- 0.41%. 9.25% is as likely as 8.00%.

The data suggests that the cap is probably closer to 8% than 9%, although uncertainty in the number of attacks makes it impossible to rule out values of 9% or more. The data can't distinguish between 8.0% and 8.5% at all - it is consistent with either value.



Edit:

There is an assumption that the cap should be a round percentage, but it may not be. If bosses have Defense Ratings, instead of Defense Chances, then the Rating could be a round number.

In particular, the parameter for Defense Rating is 27.5 points for 1% at Level 50. And 275 Defense Rating gives a defense chance of 8.6%, which is certainly allowed by the data from this test.

[Freehugs]

Death from Above is also a special attack.

[CaseyTheRetard]

(01-25-2012 06:47 AM)Freehugs Wrote:  Death from Above is also a special attack.

DFA isn't even Ranged - it's a Tech attack.

[Style]

(01-25-2012 07:16 AM)CaseyTheRetard Wrote:  

(01-25-2012 06:47 AM)Freehugs Wrote:  Death from Above is also a special attack.

DFA isn't even Ranged - it's a Tech attack.

still considered a special attack. even if it wasn't it's the same base accuracy anyway.

[CaseyTheRetard]

(01-25-2012 11:36 AM)Style Wrote:  still considered a special attack. even if it wasn't it's the same base accuracy anyway.

"Special attack" implies that it is a Ranged attack and subject to defense. It is in fact a Tech attack, and NOT subject to defense, so does not affect the analysis of boss defense taking place in this thread. Clear enough?