Threaded Mode | Linear Mode

**RuQu** · 02-08-2012, 12:23 AM

(02-07-2012 10:48 PM)LagunaD Wrote: Although I'm not familiar with Commando healing mechanics, I think I understand what your explanation. But I still don't see how more Alacrity can ever reduce your HPS, for the reasons given previously. The worst effect that resource constraints should ever have is making Alacrity HPS-neutral.

Let me try to explain my mental model.

Suppose that with 0 Alacrity, your highest HPS, ammo-neutral rotation heals for an amount H in time T. Thus, your 0-Alacrity HPS is H/T.

You also have a no-cost heal which heals for h in unit time.

With Alacrity q, you heal for H in time (1-q)T, for the same amount of ammo, and the no-cost ability heals for hqT in time qT.

The total time elapsed is still T, and your ammo consumption hasn't changed. Your HPS is now

(H + hqT)/T = H/T + qh

Thus, even in the worst case where you are resource-bound with 0 Alacrity, your HPS increases linearly with Alacrity, but the slope is determined by your filler, not your good stuff.

Maybe that is the point you are trying to make, but showing a decrease in HPS with Alacrity in your graphs suggests to me that the problem hasn't been formulated properly.

Your reasoning is perfectly sound in isolation, but breaks down during actual rotations.

We can burst heal for 10 seconds every time we cast Supercharge Cells (SCC). This removes the cooldown on Adv Probe (AP), which means we can follow it with a cheap Med Probe (MP). It also provides a 10% healing buff.

Now, assuming there there are always injured people out there in need of healing, you will cast AP/MP combos as long as you can inside that 10s window, and you will not waste any of those 10s on casting the free weak heal. Since the time of SCC is fixed, but the amount you can spend in it is not, increased alacrity can have you burst your way into a lower regen zone than a lower alacrity would have put you at. You then have to recover after the fact. While we cast Hammer Shot during regen, it actually doesn't return any Ammo, it is simply free so its EC is -1*regen. Casting HS is the same as standing around, but you can do some light healing while doing it.

In fact, this appears to be exactly what happens in the 25% case. An extra AP slips into the SCC cooldown, followed by an extra 1 Ammo MP, and this extra expenditure of 3 Ammo drops you into lower regen, thus slowing the rotation as a whole.

Seloun · 02-08-2012, 01:36 AM

(02-08-2012 12:23 AM)RuQu Wrote:
(02-07-2012 10:48 PM)LagunaD Wrote: Although I'm not familiar with Commando healing mechanics, I think I understand what your explanation. But I still don't see how more Alacrity can ever reduce your HPS, for the reasons given previously. The worst effect that resource constraints should ever have is making Alacrity HPS-neutral.
...
Thus, even in the worst case where you are resource-bound with 0 Alacrity, your HPS increases linearly with Alacrity, but the slope is determined by your filler, not your good stuff.

Maybe that is the point you are trying to make, but showing a decrease in HPS with Alacrity in your graphs suggests to me that the problem hasn't been formulated properly.

Your reasoning is perfectly sound in isolation, but breaks down during actual rotations.

We can burst heal for 10 seconds every time we cast Supercharge Cells (SCC). This removes the cooldown on Adv Probe (AP), which means we can follow it with a cheap Med Probe (MP). It also provides a 10% healing buff.

Now, assuming there there are always injured people out there in need of healing, you will cast AP/MP combos as long as you can inside that 10s window, and you will not waste any of those 10s on casting the free weak heal. Since the time of SCC is fixed, but the amount you can spend in it is not, increased alacrity can have you burst your way into a lower regen zone than a lower alacrity would have put you at. You then have to recover after the fact. While we cast Hammer Shot during regen, it actually doesn't return any Ammo, it is simply free so its EC is -1*regen. Casting HS is the same as standing around, but you can do some light healing while doing it.

In fact, this appears to be exactly what happens in the 25% case. An extra AP slips into the SCC cooldown, followed by an extra 1 Ammo MP, and this extra expenditure of 3 Ammo drops you into lower regen, thus slowing the rotation as a whole.

I think there's a mistake in reasoning here (in particular, a false premise). You say that you wouldn't waste any of the buff casting the free heal, but it's not a 'waste' if it actually increases your overall healing by keeping your regen high. Why is it a requirement to recover after the fact instead of simply keeping your regen up despite the buff, if you get better healing out of it?

The overall point is that there should be no reason why you cannot execute the same rotation when you have more alacrity than before, and if you execute the rotation exactly (by simply delaying where the new amount of alacrity reduces cast times) you should get the exactly the same amount of healing. E.g., in your 25% example, why can't you recover the original amount of healing (or better) by not using that extra AP? What prevents you from using the original rotation?

**RuQu** · 02-08-2012, 03:21 AM

(02-07-2012 10:48 PM)LagunaD Wrote: I think there's a mistake in reasoning here (in particular, a false premise). You say that you wouldn't waste any of the buff casting the free heal, but it's not a 'waste' if it actually increases your overall healing by keeping your regen high. Why is it a requirement to recover after the fact instead of simply keeping your regen up despite the buff, if you get better healing out of it?

The overall point is that there should be no reason why you cannot execute the same rotation when you have more alacrity than before, and if you execute the rotation exactly (by simply delaying where the new amount of alacrity reduces cast times) you should get the exactly the same amount of healing. E.g., in your 25% example, why can't you recover the original amount of healing (or better) by not using that extra AP? What prevents you from using the original rotation?

I think we are coming at this from two different approaches, and seeing different results because of it.

My efforts were not an optimization problem, looking for peak HPS and therefore never casting an ability if the HPS dropped as a result. As you suggest, you could make AP take 0.5 to cast, but then just wait 1 second every time before doing anything else. This would maintain the EC, so we should see exactly the same HPS, not lower.

My efforts were to approach it by simulating the decisions made by a player in game, with the information available to them. If a player has SCC active with its 10% buff and AP available, and someone is in need of healing, he will use it to save that person. SCC is our burst mode, and the player is used to doing a burst-recovery rotation, so getting in an extra AP seems like it should be beneficial, when, in fact it is not.

Now, you are correct, a wise/smart/informed player should know to not cast that extra AP if their ammo drops too low, however, I would say that if you have a stat with a level where good players get zero benefit (by standing around) and bad players get negative benefit (by not knowing to stand around instead of casting AP), that that is a pretty good indication of a point where the stat gets useless.

Looking at my calculator with current assumptions, Average Ammo over 300 casts is 9.64 with 0% alacrity, and 6.9 at 25% and 5.4 at 30%. Our regen thresholds are 3 and 8, so you can see the decrease is due to lower regen over the course of the fight.

Now, one change I could make is to only allow AP usage in SCC if the Ammo is high enough to not drop you below the 3 threshold. This naturally doesn't happen in 0% rotations, EC is low enough that it won't happen in SCC, but it does happen at high levels of Alacrity. Perhaps this adjustment, is where our methods meet, playing as a "smart player" who knows to stand around.

**CaseyTheRetard** · (This post was last modified: 02-08-2012 03:50 AM by CaseyTheRetard.)

(02-08-2012 03:21 AM)RuQu Wrote: My efforts were to approach it by simulating the decisions made by a player in game, with the information available to them. If a player has SCC active with its 10% buff and AP available, and someone is in need of healing, he will use it to save that person. SCC is our burst mode, and the player is used to doing a burst-recovery rotation, so getting in an extra AP seems like it should be beneficial, when, in fact it is not.

So you have demonstrated that a player choosing to optimize his burst healing is not going to achieve optimal sustained HPS. I think this comes as no surprise to anyone familiar with how the energy/heat/ammo systems work. There is no point at which Alacrity has zero benefit: as Laguna said, at the very least casting your energy-consuming heals quicker gives you more free time to cast Hammer Shots.

I don't see how being smart enough to not blow your bar while SCC is up is greatly more complicated than being smart enough to not blow your entire bar when it's down.

**RuQu** · 02-08-2012, 04:00 AM

(02-08-2012 03:47 AM)CaseyTheRetard Wrote:
(02-08-2012 03:21 AM)RuQu Wrote: My efforts were to approach it by simulating the decisions made by a player in game, with the information available to them. If a player has SCC active with its 10% buff and AP available, and someone is in need of healing, he will use it to save that person. SCC is our burst mode, and the player is used to doing a burst-recovery rotation, so getting in an extra AP seems like it should be beneficial, when, in fact it is not.

So you have demonstrated that a player choosing to optimize his burst healing is not going to achieve optimal sustained HPS. I think this comes as no surprise to anyone familiar with how the energy/heat/ammo systems work. There is no point at which Alacrity has zero benefit: as Laguna said, at the very least casting your energy-consuming heals quicker gives you more free time to cast Hammer Shots.

Burst and sustained do not exist in a vacuum. We frequently have to burst, then recover, then burst again.

I could link my original calculator which only assumed sustained healing, but that question is fairly trivial and non-interesting. A Smuggler can cast UWM/EM/EM in a 5 sec rotation all day long if the damage is light. Make it light enough and a Trooper can cast Hammer Shot on everyone and heal.

In that light sustained mode, Alacrity is worth 0, because if you were in a hurry to land that heal, you would be in burst mode. If there is a random spike, you can always cast Bacta Infusion once every 21 seconds, and again, if you are taking more damage than that, you aren't in sustained mode, you are in burst.

So, on it's face we can see that Alacrity is only of value if we do burst, but if we burst, we must recover, after which, we will likely be bursting again, until the boss is dead. This can appear in many forms (damage phase followed by mobility phase where damage only happens if people screw up, light boss damage with a huge damage ability every 30 seconds, etc). Exact details of how often we have to burst would require specifics of boss fights, and be less general results. People are unlikely to change out gear for individual bosses.

So yes, with limited resources burst cannot be sustained, but no, the result is non-trivial.

And "more time to cast Hammer Shot" is a massive waste of stat budget. Hammer shot has a coefficient of 1, and no static healing independent of your BonusHealing. Compare to MP with a coefficient of 2.72 and base healing (at 0 bonus) over over 900. You will always gain more by making the AP/MP/BI/TP heals in your rotation stronger than you ever could by fitting in an extra Hammer Shot. The only way Alacrity can benefit you is by getting more AP/MP combos into an SCC window (the only time it is spammable), and then only up until the point that it starts costing you Ammo from the increased EC.

**CaseyTheRetard** · 02-08-2012, 04:32 AM

(02-08-2012 04:00 AM)RuQu Wrote: So, on it's face we can see that Alacrity is only of value if we do burst, but if we burst, we must recover, after which, we will likely be bursting again, until the boss is dead.

Sustained HPS is only a suitable metric for evaluating sustained healing. If you are trying to evaluate a burst or mixed model with sustained HPS as your metric then you will have nonsensical results like Alacrity reducing effectiveness.

**RuQu** · 02-08-2012, 04:48 AM

(02-08-2012 04:32 AM)CaseyTheRetard Wrote:
(02-08-2012 04:00 AM)RuQu Wrote: So, on it's face we can see that Alacrity is only of value if we do burst, but if we burst, we must recover, after which, we will likely be bursting again, until the boss is dead.

Sustained HPS is only a suitable metric for evaluating sustained healing. If you are trying to evaluate a burst or mixed model with sustained HPS as your metric then you will have nonsensical results like Alacrity reducing effectiveness.

Please suggest a metric for describing performance in a burst-recover rotation.

To me, a burst-recover rotation is a sustained healing situation where the damage comes in waves, with peaks and troughs you can align your own burst healing to. Of course, you could hope that the burst damage is low enough to not kill the tank while you maintain constant sustained healing, and the tank health simply rises and falls instead of your healing effort, but the cries of the Smugglers and Agents on the forums indicate that that is not working out too well.

I also assume the worst case scenario where someone always needs healing. If you out-gear the fight, substitute in similar cost DPS abilities as desired, or put on a movie in the background, but an over-geared situation is non-interesting from a stat weight point of view.

Viewed as a sustained fight with non-constant damage, total HPS should be a valid metric. Of course, if you know of a better one, I'd be happy to implement that instead if it provides greater value to the Combat Medic community.

**CaseyTheRetard** · 02-08-2012, 05:31 AM

(02-08-2012 04:48 AM)RuQu Wrote: To me, a burst-recover rotation is a sustained healing situation where the damage comes in waves, with peaks and troughs you can align your own burst healing to.

Viewed as a sustained fight with non-constant damage, total HPS should be a valid metric.

I would look at modeling one peak - player does whatever is necessary to maintain some target HPS threshold - and one trough - player does as much healing as possible while recovering resources. If player state at the end of a full wave is the same as initial state, results from the model naturally extend to fights with an arbitrary number of waves.

Fraser · 02-08-2012, 05:36 AM

(02-08-2012 05:31 AM)CaseyTheRetard Wrote:
(02-08-2012 04:48 AM)RuQu Wrote: To me, a burst-recover rotation is a sustained healing situation where the damage comes in waves, with peaks and troughs you can align your own burst healing to.

Viewed as a sustained fight with non-constant damage, total HPS should be a valid metric.

I would look at modeling one peak - player does whatever is necessary to maintain some target HPS threshold - and one trough - player does as much healing as possible while recovering resources. If player state at the end of a full wave is the same as initial state, results from the model naturally extend to fights with an arbitrary number of waves.

Would you then compare one setup to the next with raw throughput, the hps of the wave, the comparative hps of the peak and trough or some amalgamation of them all?

Something I might add to that would be the sustain cycle you run between the end of a recovery period and the beginning of a burst cycle. Without knowing the effect a given stat has on all 3 analysis of the remainder is of little importance.

**RuQu** · 02-08-2012, 06:00 AM

And what if player state is not the same? In a 0 Alacrity rotation, you will have more Ammo when you reach 30 charges of CSC than you will in a high Alacrity state. Do you fire SCC off again when you have 30 charges, or at a fixed Ammo level? Based on that decision, you will have different results after the second burst-recover than after the first. How many do you run through before saying it is enough and simply extending through an arbitrary number? At this point, haven't we arrived at the exact state that my calculator is at, namely an arbitrary and variable number of peak-trough cycles based around an approximation of the logic used by a player in game with the final total HPS used as the metric of comparison?