RDM Scortch

[Archwizard]

The actual ideal situation

... completely ignores that procs are given at a rate determined by RNG? And that reaching 80/80 doesn't eliminate the chance to gain more procs? And that even in the listed scenario, there's nothing keeping you from getting yet another proc, because the only way to guarantee proc prevention would be casting Impact in single-target? And that you don't need both procs to still be in a bad position for a Verfinisher, just the proc for the lower Mana value? That when your choice is losing mana to overcapping, losing it to overwriting an existing proc, or spending it on Reprise, only the latter turns all of the excess into potency? That this is exactly what I mean when I say the value of such a loss would be determined by RNG?

Obviously we're just talking in circles at this point, and it's a wasted effort to continue trying to convince the other. Again, please drop it.

[Singularity]

That when your choice is losing mana to overcapping, losing it to overwriting an existing proc, or spending it on Reprise

Using media tour numbers, the melee combo delivers 2270 potency for 125 mana (including refunded mana) over 4.08 GCDs (at base GCD). Using the distribution from the existing theorycraft (accounting for the buff to Jolt and removal of Impact), we can recalculate the average GCD potency to be 287.6, so the melee combo gains us ~1097 potency, or about 8.8 potency per mana.
Reprise gains ~47 potency (accounting for its 2.2s GCD) at a cost of 20 mana, for a net loss of ~129 potency. A proc is worth 20 potency and 3 mana (26.4 potency equivalent) over the Jolt it would replace, so even if you'd 100% of the time overwrite the proc (instead of 50%), it's always correct to do so vs. using Reprise and trying again, since the cost of doing so outweighs the gain of an extra proc.

Likewise we can calculate the cost of overcapping vs. using Reprise - that 129 potency loss is the equivalent of 14.7 mana, so if you're overcapping by 14 or less mana, that's also correct vs. using Reprise.

The upshot of this is that Reprise only has value in non-ideal situations. Since it costs more to use it than you gain from a proc (whether the proc is gained from avoiding an overwrite or from setting up your finisher to generate one), and more even than you lose to overcapping up to 14 mana, it's only a gain when you're losing GCDs to movement, or if you can use it to delay your melee combo such that the combo lands within buffs that more than compensate for the cost.

[Rongway]

Using media tour numbers, the melee combo delivers 2270 potency for 125 mana (including refunded mana) over 4.08 GCDs (at base GCD). Using the distribution from the existing theorycraft (accounting for the buff to Jolt and removal of Impact), we can recalculate the average GCD potency to be 287.6, so the melee combo gains us ~1097 potency, or about 8.8 potency per mana.
Reprise gains ~47 potency (accounting for its 2.2s GCD) at a cost of 20 mana, for a net loss of ~129 potency. A proc is worth 20 potency and 3 mana (26.4 potency equivalent) over the Jolt it would replace, so even if you'd 100% of the time overwrite the proc (instead of 50%), it's always correct to do so vs. using Reprise and trying again, since the cost of doing so outweighs the gain of an extra proc.

Likewise we can calculate the cost of overcapping vs. using Reprise - that 129 potency loss is the equivalent of 14.7 mana, so if you're overcapping by 14 or less mana, that's also correct vs. using Reprise.

The upshot of this is that Reprise only has value in non-ideal situations. Since it costs more to use it than you gain from a proc (whether the proc is gained from avoiding an overwrite or from setting up your finisher to generate one), and more even than you lose to overcapping up to 14 mana, it's only a gain when you're losing GCDs to movement, or if you can use it to delay your melee combo such that the combo lands within buffs that more than compensate for the cost.

We should include missed mana gains in the calculations. We've so far accounted for the potency of the spells we would have cast instead of performing a melee combo, but we haven't accounted for the mana those spells would have generated. When we account for this, the melee combo actually costs 163.21 mana, and E.Reprise actually costs 28.34 mana or 22.70 mana, depending.


If you start with 160 mana, you can perform a melee combo in the course of 10.08s, doing 2270 potency, and ending with 35 mana.

In a mana generation / spell casting phase, you can do 250(6) + 320(12.5) every 4.88s. This accounts for a Verslowspell's 50% chance to increase the potency of the next hard cast by 20(3). Then the expected potency and mana generation over 10.08s is 1177.38 potency and 38.21 mana.

So whereas starting with 160 mana and performing a melee combo would leave you at 35 mana, starting with 160 mana and casting spells for the same amount of time would leave you with 198.21 mana. The melee combo does 1092.62 potency over spellcasting, at the cost of 163.21 mana, yielding an effective 6.695 potency per mana.


Likewise, if you started with 20 mana and used E.Reprise, in 2.2s you would do 300 potency and end with 0 mana. In those same 2.2s, spellcasting would do on average 256.97 potency and leave you at 28.34 mana. Therefore, Reprise does 43.03 potency at the cost of 28.34 mana, or dealing 1.518 potency per mana. We could adjust these numbers slightly if we considered that Reprise should only ever take the place of a Jolt, in which case the spellcasting potency is 225.41 and the mana generation would have been 2.70, which would be a gain of 74.59 potency at the cost of 22.70 mana, or 3.286 potency per mana.

But for now let's go with the less optimistic calculation of 43.03 potency at the cost of 28.34 mana. At the melee combo rate of potency per mana, we expect an expenditure of 28.34 mana to deal 189.736 potency, which makes E.Reprise a loss of 146.71 potency.

If we convert this 146.71 potency back to mana, it's about 21.9. Therefore, going 21 mana over cap would still be preferable to using E.Reprise.

[Archwizard]

-snip-

This was just linked to me in another RDM thread, which I find very helpful for this particular discussion. Feel free to read it over and debate the math with its original author.