GR Tier vs GR Time: An Analysis

General Discussion
62

Ok, here’s the revised version of the method in the OP:

Each rift gets split up, into:

Regular time. Time spent fighting mobs. This scales up 17% per tier.

Move time. Time spent just moving around, without really fighting any mobs. Assessed either by video or at 60 seconds, or 20% of total time, whichever is higher. This does not scale up at all.

Stricken time, assuming this gem is used. Time spent fighting the boss. Assessed either by video or as an average amount of time for a boss fight, for the build in question. (Leapquake- 180-300 seconds. Frenzy- 60-90 seconds. Etc.)

Conduit time. Time with an active Conduit. Always 60 seconds, assuming FoT is used. Use if presence of a Conduit is known/suspected. This scales up at a rate of (Conduit time / (Conduit time + Regular time) * 17%). In a 5 minute rift with three minutes of Regular time, and a Conduit, this would be (60 / (180 + 60) * 17%), or 4.25%, per tier.

Using these time scalings, adjust the original total time upward to see how many additional tiers might be reachable.

If video is present, it can be inspected for instances where increased incoming damage might kill the player, increasing the time. If there is no video, the top 30% of the increased tier expectation is treated as a range. For instance, if we previously said “+7 tiers”, we now would say “+ 5-7 tiers”.

If Squirt’s is used in a way intended to maintain high uptime (using a shielding mechanism such as Gizzard, Fortress Ballista, etc) then video can be inspected for instances when the Squirt’s buff might drop due to higher incoming damage. If no video is present, another 30% of the “solid” potential becomes a range. For instance, our “+ 7” which became “+ 5-7” now becomes “+ 4-7 tiers”.

And that’s it. This may not provide us with a perfect, absolute answer, but I do think it will allow us to look at timing in a significantly clearer way than we have up until now.

Any interest in trying out this method, in whatever fashion you see fit?

1 Like
63

Yes, I can test a spectrum of rifts with various builds, starting point cited last:

  • LoN Bomb (stricken) 130
  • AoV Fist of the Heavens 125
  • S6 Impale 123
  • GoD6 Hungering Arrow 125
  • Invoker Thorns (stricken) 123

Maybe 10 levels each and I’ll run them up to a point where it becomes a push, with a 50% chance to fail on the last tier.

1 Like
64

That’s awesome!

I think the formula will work less well for content that is more on the “speed” end than the “pushing” end of things, but that doesn’t mean you shouldn’t do it- it’ll still be interesting to see how the results shake out.

Edit: actually yeah, some of the “speed” results should be very interesting. Because, let’s say you do a GR 110 or something, and run it through the formula, which predicts that your time will go up like so as the tier increases, and thus that you can go up +X tiers while staying under the 900 second timer. You’ll pretty much immediately get other info germane to that claim, as you run GR 111, 112, etc.

You wouldn’t expect those expanded results to completely track the formula, of course, since you’re not running the same rift over and over… but it’ll be very interesting to see, all the same.

65

Are you sure this part is right? If someone takes conduit immediately and kills everything (theoretically speaking) some strange result happens. Do you count move time with conduit on, but not fighting?

66

Definitely not! :slightly_smiling_face:

But, can you elaborate on your example a little bit?

Do you mean that, if Move time is greater than total time (like if the rift only takes 58 seconds) you end up dividing by zero?

If so, I guess that’s where using this analysis only on “leaderboard examples” really cleans up the input a lot, and you probably don’t have to worry about that kind of stuff.

Another answer would be that, in a 58 second long rift, basically everything is Move time, and Conduit time is kind of irrelevant. Your player-dealt damage is going to be so high in such an instance that the Conduit damage will be basically irrelevant.

Below a certain level of “tier vs player power” we can’t really get any info via this analysis. If you just run through a rift, instakilling everything, we don’t really get a good impression of how much damage you were doing. Did you overkill those mobs by a factor of 2, 20, 200, or 2000? There’s no way of knowing.

67

Let’s start from the beginning:

If conduit damage scales with mobs’ health then you need a formula to estimate the importance conduit plays regarding player’s damage, right?

68

Yes… sort of, more or less.

Don’t forget that on the one hand, there is really no way to know what share of the overall damage a Conduit might be doing (without video, anyway).

And on the other hand, we are not talking about adding or subtracting a Conduit. Either the player had one, and finished in the time we see, or didn’t have one, and finished in the time we see.

69

Okay, what’s Conduit’s damage at GR150 and the damage of the rank 1 dude you posted in OP? Are these 90:10 for conduit or 50:50 or?

70

My guess is that in that instance, his character was dealing a bit under 40% of the damage, and the Conduit, just over 60%.

At the end of the day, though, you’re just trying to decide whether to scale up that one minute by 17%, 0%, or somewhere in between. It doesn’t actually make a huge difference. If you just scaled that whole 4:37 rift at a flat 17%, you end up with a +7 GR mark, whereas if you assume one minute of Conduit, scaled at 6.5%/tier, (which is what I roughly calculated out for that clear), you get a + 8 GR mark. Even if you scale it at 0%, it’s still a +8 GR mark overall. Not really a big spread in the final result.

71

It may be more precise then to use the GR level into a formula since we’d have a good average estimation about player’s damage to conduit’s damage from one’s time and GR level.

Or simply use 10% as a fixed value for that 60 seconds with conduit.

72

I think that basing the Conduit scaling on GR level alone would run into a problem at both ends. At the lower end, I don’t think this analysis is very useful anyway. And, there’s nothing to distinguish, say, a tier where you do 1000x the damage of the Conduit (and kill the enemies instantly) from a tier where you do 10x the damage of the Conduit (and kill the enemies instantly).

And at the higher end, we know that different builds value a Conduit differently… a Rend Barb, for instance, basically requires a Conduit for a high clear. And in the case of a 150 Rend clear, the Conduit is probably doing close to 100% of the damage, as opposed to an Inna or Firebird clear, where the split is maybe 70/30 or 60/40.

I just put this together, maybe it’ll be helpful: a spreadsheet showing a 20 tier “slice” of GRs, comparing Player damage and Conduit damage. In GR X, the starting point, it’s assumed the player does about twice as much damage as a Conduit, so this is something like a solo XP/gear speed run.

Within the “Zone” where this analysis is at all useful, I think the most you’d ever scale up Conduit time is around 11% for one tier. And I think the vast majority of clears we (or I, anyway) would want to analyze would start around X+9, where the situation is the reverse of tier X, and Conduit does 2x the damage of the Player. From there, you’re only scaling up by about 5% for the first tier, and less for each additional tier.

1 Like
73

The very worst build is out of the way.

Invoker
GR Level Clear Time Pylons No. of Floors Notes
123 13:05 Ch,Po 6 End of GRift
124 9:21 Co,Ch,Po 5
125 10:42 Sp,Sh,Ch,Co 6 Many Blues
126 12:29 Po, Ch 7
127 13:00 Sp,Co 8 Many Blues
128 12:04 Sh,Ch,Sp 8
129 12:49 Sp,Ch,Po 5
130 11:29 Po,Co,Sh 4
131 18:16 Po,Ch,Sp 7 End of Grift, Many Blues
132 15:25 Sp,Co,Sh 8 End of Grift, Many Blues

This build is fundamentally flawed. Encountering the end of Greater Rift is frustrating. Really Invoker should synergize with Slash/Omnislash, you Slash until the trash is dead and then you finish off with Punish or vice/versa. – we should never run out of elites or be forced to kill 4-5 champ packs – so dumb.

1 Like
74

You know what, it would be easiest if the player using this tool finds the percent himself depending on conduit importance.

For example, have a table with all builds at Y and conduit importance at GR100-150 separated by 10 tiers for X. This way a player unfamiliar with a specific build could estimate conduit importance roughly.

Then when the player knows the conduit importance regarding his damage he can easily calculate the percent value for the 60 seconds increase with:
(100 - CI) * 0.17

CI is the percent of conduit importance. For example - 60% CI vs 40% player damage; or 100% when conduit deals all the damage; or 0% when the player deals all the damage.

75

Yeah, the more information, the better. Having video of the clear is best of all, but every individual morsel of info is a definite improvement over :man_shrugging: . And the player is generally going to have more info than a random onlooker, even if their info isn’t perfect. For cases where the user is not the player, though, I’d still like to have a decent way to make an estimate.

While the math on the formula I gave can turn wonky in some extreme conditions, I think it’ll actually work pretty well for most cases. In that 4:37 Monk clear, for instance, I estimated a 60 second boss fight, and 60 seconds of move time.

So that would be (60 / (277 - 60 - 60) * 17%) = 6.5% scaling. (and 61.8% Conduit effectiveness). Seems about right, right? I mean, that falls between X+7 and X+8 on my chart, which means that the Conduit is doing around 2/3 of the damage. That seems pretty true to a mark of where “speeds” end and “pushing” begins, which makes sense with a total time of 4:37.

I think we’d need to think of Paragon/Augs as well… a P800 player with partially augged gear might have ~10k mainstat, compared to a P10000 player with around 60k mainstat. That’s roughly an 11.5 tier difference. The former player’s “high push” may well be the latter player’s “speed run”. And 60k mainstat isn’t even close to the limit, of course- there are players out there with over 18k Paragon, which would give them over 100k mainstat. So the level of Conduit Importance would vary quite widely, even when keeping the build + tier constant.

Still, I think that’s a pretty solid idea. You could do something like low/mid/high paragon splits for each build… or you could just use a “middle of the road” figure, like 5k paragon, which is the number Blizz claims to balance around.

76

Thanks for doing that! Before you proceed, do you want to talk through the results you got? Or do you just want to keep going?

77

In this particular case. I don’t like that formula however. A flat number would do better than it on average.

This. And the player could multiply a coefficient after, based on his actual paragon level. Or even have the whole thing in main stat entirely.

78

What flat number do you have in mind? 10% is obviously too high, especially since (this just occurred to me), to really do it right, you have to go tier-by-tier, reducing the scaling each time. For instance, per my chart, the first tier might scale by 10.15%, but the 2nd tier would only be 9.5%, the 3rd 8.8%, etc.

And, I think most instances where people would actually want to use this (i.e. pushing) would start around X+10, where Conduit does a bit less than 3x the damage of the player. From there, the scaling is around 5%, and headed down. So if we were going with a flat number, I’d go no higher than 5%.

dmkt, you got any opinion on this?

79

Either way – Invoker profits immensely from Conduit. You just pull it, regardless if there is an elite nearby. Killing trash is the profit from Conduit. It’s the most extreme build in that regard.

Progress from 3-packs blues up until GR129-130 is breakeven, it becomes a loss thereafter. Sometimes it is worth taking a loss just to get the progress for the next pylon spawn.

The RNG is huge, on that GR132 clear I passed 9 blue packs in a row… I’m around 60 seconds behind timer. I finally found a yellow… then a conduit, I was ahead by 2-3 minutes, it was again many blue packs on the last map… stuck at 96% 90 seconds left… the map ends, I have to kill a blue 5-pack, I get to the guardian with ~20 seconds, it only takes like 3-4 cycles to kill it.

That is Invoker life.

80

It really depends who will use it and with what build. The flat percent value will show itself from the average of all builds.

81

Frenzy Barb is very similar.

Have you got any opinion on the “Conduit scaling: formula vs flat scaling” debate?

Here’s another use of the formula, on the Leapquake clear I linked in the OP. In this instance it would be 60 / (720 - 144 -180) * 17% = 2.58%

That puts it on the table around X+16, where Conduit is doing about 6x the damage of the player. I can tell you from copious experience that in this instance, that is probably a pretty close estimate. Grabbing a Power Pylon, which gives you 5x damage for the same duration as a Conduit, wouldn’t let you kill the stuff you can kill with a Conduit. Since you still have 1x damage of your own, combined with 6x from the Conduit, that’s 7x compared to 5x with a Power, a 40% boost. That feels just about right in this instance.

I know you don’t like the formula. And, I’m not absolutely dead-set on using it. But, can you provide a counter-example? Not something dreamed up- video of an actual clear, performed by a competent player? A case where the formula produces a result that just seems ludicrous.

Let me take a step back for a minute and give a reminder that a lot of the details we’re discussing are really quite small in their effect. If you just scale that Leapquake clear at a flat 17%, entirely ignoring the effects of move time, Conduit, and Stricken, you end up with 842 seconds. If you perform the analysis as I’ve outlined it, it’s 794 seconds, only about a 6% difference (about 1/3 of a tier).