GR Tier vs GR Time: An Analysis

Update: I’ve now outlined a refined method that takes note of many of the issues listed in the two “edits”, directly below.

Edit: I’m going to write more on this later (right now I’ve got Turkey to cook), but Kikaha made a great point about incoming damage at both higher tier and higher time. Obviously, if the enemies kill you, the results will change. So this analysis, rather than producing a certainty that you could get a higher clear, actually just indicates that you have the damage to get the higher clear.

Edit 2: More wrinkles cropping up. Including:

  • Intermittent outgoing damage, i.e. CoE, which might significantly impact the time (increase by either more OR less than the “expected” 17%), especially in small-number tier increases.
  • Inconsistent outgoing damage, linked to either incoming damage (Squirt’s) or outgoing damage/kills (Archon stacks, Rampage stacks, etc), either of which can increase the time.
  • Scaling of Conduit time, depending on the amount of damage the player is doing.
  • Estimating “move time”, which should scale up very little, or not at all, as the tier increases.

1: Introduction

One thing we never seem to have come up with is a good way to assess leftover time in a given GR level, and compare that to overall GR level performance. For instance, we all know that clearing GR X under a certain amount of time means that it would be 100% possible for that player, with that same build/gear/paragon etc, to clear GR X+1. After all, we never see somebody clear 130 in 2 minutes, and think “there’s no way he could do 131!”

But, where is that line, exactly? 13 minutes? 10? 8?

As top GR clears have been compressed into 150, we’ve kind of lost track of how high some of these builds might really be able to go. So, let’s take a look at time in a GR, and the various factors that take up, or save, time.

2: Basic time:tier scaling

A Greater Rift is 15 minutes, or 900 seconds long, maximum, assuming you are going to achieve a success. In those 900 seconds, you have to do three things:

  1. Move around

  2. Kill enough enemies to fill the progression meter to 100%

  3. Kill the boss

Each of those activities takes a certain share of your time, and those shares are different for different builds. Frenzy Barb and Impale DH for instance, spend relatively more time on the 2nd element, and less on the 3rd, whereas with Leapquake Barb or Multishot DH, it’s the opposite.

But, we’re not really interested in direct build vs build comparisons at this point. What we’re looking at is one build, one player, one set of circumstances, essentially vs itself. Another way of looking at it: if we could replay a rift, exactly the same (like a challenge rift that also plays itself) what would happen if the difficulty were increased by 1 tier? Would we succeed? Or not?

Let’s suppose for a second that the time it takes to complete a GR just scaled up directly with monster health. Mobs get 17% tougher for each increase in GR level, so in this picture, it would take you 17% more time to complete the rift.

Take a look at the top clear worldwide in Era 14- a GR 150 clear in 4:37 (277 seconds) using Monk by the Chinese player fw6263. If the scaling were just a direct correlation between mob health and time, we would expect that he could do at least GR 157, since (277 seconds * 1.17^7) = 831 seconds, well under the 900 second timer.

But, of course, that isn’t exactly how the scaling works.

3: The difference-makers: Stricken + Conduit

There are two major factors that break down this 1:1 relationship between mob health increase and clear time: Bane of the Stricken and Conduit Pylons. Let’s look at each and assess how it affects the time spent.

Bane of the Stricken, as we all know, is mainly the “anti Rift Guardian” gem. It does have some other use, particularly for fast-attacking builds like Frenzy Barb or Generator Monk, but its main purpose is to help with killing Rift Guardians. Many builds, in fact, would find it essentially impossible to kill RGs at even a moderately high level without having this gem equipped.

Stricken alters the monster health:kill time ratio by increasing our damage for each attack we make. So each increase in monster health increases the time needed to complete the kill, but the more time we spend attacking, the greater our damage becomes, which then drives the time back down. Ultimately, Stricken causes the increase in time to more or less track the square root of the increase in mob health. So, if a boss’s health doubled, rather than taking twice as long, it would take only about 1.41 times as long.

Conduit Pylons also alter the relationship between monster health and time. But, rather than increasing in damage the more you hit enemies, like Stricken, it simply always kills enemies at the same rate, no matter the GR level, because its damage scales up at the same rate as the mobs’ health. So, given a group of trash and elites, the Conduit power will kill them in the same amount of time, whether you are playing GR 100, 120, or 150.

4: Modified time segments

So, now we can break down each rift into three sections:

  • “Regular Time”, which will scale up at the rate of mob health increase, or 17% per added GR level.

  • “Stricken Time”, which will scale up at the square root of the mob health increase.

  • “Conduit Time”, which will not scale up at all.

5: Applying modified time with “perfect knowledge”

In cases where we have essentially perfect knowledge of how a rift was played (i.e. video), we can now estimate quite well how this build/player/rift would play out if pushed up one or more GR tier.

Take, for example, this GR 146 clear, in 11:59 (719 seconds) by the Korean player Deal, using the Leapquake build (note that the video starts 5 seconds into the rift). From start - 7:59, he is in “Regular Time”, from 7:59 - 8:59, in “Conduit Time”, and from 8:59 - 11:59 in “Stricken Time”.

So, that is 8 minutes (rounded), or 480 seconds of “Regular Time”, 1 minute, or 60 seconds of “Conduit Time”, and 3 minutes, or 180 seconds of “Stricken Time”. Now let’s apply our time alterations for each segment, as if we had moved up 1 GR tier, to 147.

Regular: 480 * 1.17 = 561.6

Conduit: 60 *1 = 60

Stricken: 180 * √1.17 = 194.7

Total: 561.6 + 60 + 194.7 = 816.3

Obviously, he could have done 147, since 816 is well under 900! Let’s keep going. How about 148?

Regular: 561.6 * 1.17 = 657.1

Conduit: 60 * 1 = 60

Stricken: 194.7 * √1.17 = 210.6

Total: 657.1 + 60 + 210.6 = 927.7

Close, but not quite! We’re about 28 seconds over the time limit. That isn’t to say that with an even better rift this player couldn’t have cleared 148. But, what we’re after is complete certainty that a higher tier could be cleared, and to get that certainty, every element of the rift except the mob health has to remain the same.

At any rate, we can confidently perform these calculations for any clear in which we have video, or any other thorough record of the timing, aka perfect knowledge.

But, what if we don’t have perfect knowledge?

6: Applying modified time with “imperfect knowledge”

Most of the time, we aren’t going to have perfect knowledge of a clear. But, that’s not to say that we will have zero knowledge. We can usually check the in-game leaderboard to see the details of what build was used to achieve a clear. And often, especially in the case of clears high on the leaderboard, we can make some educated assumptions about our modified timings.

For instance, most clears that end up high on the leaderboard, for almost any class or build, make use of a Conduit Pylon. Yes, there’s no 100% guarantee of this, but it is far more likely than not. For some builds, the certainty is essentially, if not in actual fact, 100%. All 150 clears with Rend Barb, for instance, undoubtedly made use of Conduit, since it is nearly impossible to kill elites, especially yellows, without it. (And, every 150 for this build recorded on video has indeed used Conduit.)

So, for most high clears, we can factor in 60 seconds of Conduit time. We can also check the leaderboard to see if they are using Stricken (also near 100% certainty, but we can confirm). And, based on knowledge of the build, we can make a rough estimation of the time spent fighting the RG. This can vary significantly, of course, based on whether or not the player had the benefit of a Power Pylon vs the boss. In the Leapquake video we considered earlier, for instance, the time vs RG with a Power was 3:00. Without a Power, it would have been between 4:30 and 5:00- a significant difference. But, that still gives us a range to work with- 3:00 to 5:00.

So, let’s say we have a clear of 147, in 13:25 (805 seconds). We have no video. We do know, having looked at the in-game leaderboard with its build snapshot, that it was done with the Leapquake build, and that they used Stricken. We also know that Leapquake has almost as much trouble with elites as Rend, so the presence of a Conduit is a near 100% certainty. And we know that boss kills with this build, at high push, generally take between 3:00 and 5:00.

That means our timings are:

  • 60 seconds of Conduit time

  • 180 - 300 seconds of Stricken time

  • 445 - 565 seconds of Regular time (note that the lower Stricken time implies the higher Regular time, and vice-versa)

  • 805 seconds total

Scaling this up by 1 GR tier gives us:

  • 60 seconds of Conduit time

  • 194.7 - 324.5 seconds of Stricken time

  • 520.6 - 661.0 seconds of Regular time (note that the lower Stricken time implies the higher Regular time, and vice-versa)

  • 905.1 - 915.7 seconds total.

Both ends of this range are a bit over the 900 second timer. But if the original time were just a bit lower, around 13:00, we would assume that the clear probably could still have occurred at +1 GR tier.

And, for every clear, even when we have zero knowledge, the floor of this time:tier relationship is the direct scaling of monster health, i.e. 17% per level. Meaning that every clear ever accomplished that took place in less than 769 seconds (12:49) would still have been successful, 1 GR tier higher. And in less than 657 seconds (10:57), 2 tiers higher, etc.

7: Conclusion

With this information in hand, we can evaluate a lot of the GR 150 clears that have filled the leaderboards of late (not to mention any other clear, of any level, in the past). I lack the build-by-build knowledge to perform an “imperfect knowledge” analysis, at least above the “floor” level, for most builds outside the Barb class. But somebody has that knowledge, if not in whole then at least for individual builds, and I hope this analysis is helpful to them.

27 Likes

I wish I could give a post more than one like.

1 Like

Back in the days we made No Paragon Rankings. I still keep the site and the clears (you can also see the formula we used):
http://rankings.byethost33.com/
https://www.diablofans.com/forums/diablo-iii-general-forums/diablo-iii-general-discussion/141657-the-unofficial-top-100-rankings

The formula, although far from perfect, produced way better results than the regular rankings.

What you wrote can give you a good estimate about +1 GR tier in the 13:00 ~ 14:00 margin time. Below that however you can’t apply any precise formula. A guy making GR1 for 3 minutes probably won’t end at GR9.

Well, yes and no. What I’ve laid out here is really only for high pushing. You could use it for speeds, too, but if you do that, it isn’t going to tell you how high you can push, only how high you can “speed” in under 15 minutes (and of course, beyond maybe 5 or 6 minutes it’s really not speeds anymore). Furthermore, it really only shows how far you can be more or less mathematically certain that the next tier (or 2, or 3, whatever) would be doable.

So you’re right that the guy doing GR1 in 3 minutes (for whatever reason) probably can do quite a lot higher than GR 9. But, GR 9 may be as far as we can extend that mathematical certainty, based on that one GR 1 clear, even if common sense tells us he could go much higher.

I like the paragon-adjusted leaderboard, by the way. You should get it started up again.

1 Like

You gave an example with “GR 150 clear in 4:37”.

Yep. Knowing the minimum one can go will work.

I haven’t played a minute long of D3 post S5. I am coming back only if they cap Seasonal main stat and gem levels; and/or if they improve Challenge Rifts. In both cases the no paragon LB won’t be needed. Besides, right now (from your info about top clears) such LB won’t be of any use.

1 Like

2 minutes per GR best case scenario? Maybe 3. If I’m doing 9 minute GRs chances are there is room for 3 more levels before it gets hairy. But im sure you could write a formula that takes into consideration time vs power creep.

Haha, yeah! I guess that is where speeds and pushing come together somewhat. But, that is afaik still a pushing setup being used, and we could still evaluate the actual +GRs we might see (at least 7, maybe more) if such levels actually existed.

If somebody were using an actual speed build, say a speed setup for Rend, and did 130 in 4:37, then that wouldn’t tell you about how high they could push, since they would presumably use a pushing setup, rather than a speed setup.

Like I said, this method will work for what we usually call either pushing or speeds. But it can’t crisscross between them, because that implies a change of build, and this analysis presumes that everything, except the mob health and therefore the time, stays the same.

It doesn’t matter. You can only calculate the minimum again.

Unless I’m misunderstanding you, I think we are agreeing…

The map tileset is the biggest wild card. There is no league maps or whatever so it’s all RNG. Might be a big map, might be a small one, could be tunnels, lots of bridges. Some solo builds that have zero speed buffs would be in more trouble taking a wrong turn than even fighting tougher monsters.

Apache, I’m thinking you’re maybe not 100% understanding how this analysis works.

It doesn’t consider any possible changes in the conditions, other than the mob health (which by extension also increases the time). So, we’re not asking if they could go higher, if they had a different map, or different mobs, or a better boss, or more paragon, or better gear, or didn’t take that left turn at Albuquerque, or anything like that.

It just looks at whether they had enough time left that they could do exactly the same map but at a higher difficulty level.

So: this doesn’t tell you that a player could only go X tiers higher, no matter the conditions, just that they could for sure go X tiers higher, given the exact same rift, but with the mob health increased.

2 Likes

Yep, exactly. The setup (pushing, speed or whatever) doesn’t make any difference or give you more information. It’s not the setup that defines the activity as “pushing”, it’s the finishing time. That’s why a no paragon LB won’t be relevant atm.

It’s sad Blizzard don’t care about competition in D3 at all.

1 Like

I think we’re on the same page. I was just trying to make it clear to whoever else is reading that you could use this for whatever build, but you can’t get a clear with one build (call it a “speed” build), use this analysis on it, and expect that to have any bearing on a different build (call it a “push” build).

MathsBarb™
Strongest build in the game.

1 Like

Not exactly. Only 2 stages in a push:

  • Progression Time
  • RG Time

An experienced player knows exactly the proportion to split these times. RG time, you take the slowest kill time and compare against progression making ability, then filter out unrealistic RG, aka fishing for RG. Stricken time is not as static as you mentioned, stricken effect on single target vs rg with adds have different efficiency.

In a push tier, there’s progression map or main map and follows by a follow up unless its a 100% floor (summoning mobs or illusion enslave nightmare). Your choice on the main map depends on the build’s characteristics. Festering/Battle Field best contender as main map, but on extreme pushing, pretty much only festering is selected as main map. This is because total no of elites are different depending on map/tile set ranging from 0/1 on small maps to 7-8 on big maps. BF despite been open, has only 5 elites + pylons vs festering 6 + pylons. You have to determine do you want a 60% main map or 80% main map.

Generically speaking, most builds pushing 145-150 is about similar timings, even though you are looking 1.17^5 =2.19 times more health. In some situations, 140-150 clear is around the same timings. Consistency in killing elites = consistency in GR clear times.

There are more factors reflect LB time. Player habits and individual preference the biggest factor. Some attempts hard core fishings, some attempts anything looking “doable” My self for example, the inna/fb wiz was 1 key each, didnt care about map/mobs. Crusader on the other hand only attempted 4 corner festering and Grotesque summoner, all other keys auto ripped. But these things are not reflective on LB, although stat selection (provided not entirely copycat from Rank 1 clears) somewhat reflect the players preference.

Progression time splits between maintaining progression and burst time. Condy is a big burst, sometimes power > condy in burst efficiency. Recently the inception of squirts making shield pylon you biggest burst. 2 min shield = 2 min of 100% or 2x SQ = 4x during duration, higher than power of 250% or 3.5x during progression. Top clears with SQ these days, shield pylon has increasing impact. Take my LoD sader in s24 for example, same 6 packs around 30% progression, the cost time (killer time) from lowest to highest is shield > power > condy. When you controlling progression time, its important to convert progression into time and factor in the cost time, especially on bad/tight floors. Progression made from kills can be further called (my accounting backgrounds) “Net Progression” after deducting cost time. THUD has advantage but studying /memorising http://warpath.eu/mp.html has big advantage. Even for THUD users, the difference in “pro level” is whether you can accurate calculate progression time from looking at mobs other than looking at the plugin. If 100% time = 15 min or 900 secs, then 1% = 9 secs. 5% = 45 secs. If a 5% yellow packs takes 4 cycles of 16 sec or 64 secs to kill. You just lost 19 secs of lead from that setup.

There’s a lot more to write. What are you looking for Rage? Do you want a thorough summary on GR push strategy or ways to determine the builds push capacity?

2 Likes

I think there’s a bit of a language barrier here…

Please don’t be offended, but basically nothing you wrote has any bearing on what I’m talking about. This isn’t a discussion of “how to push GRs better” or anything like that. Don’t get me wrong, I’m always interested to discuss that stuff. But that’s not at all what this thread is about.

Exactly. Hence

I dont think I fully gets what your purpose is… what is your purpose?

In brief, it’s a way of determining how much higher a particular clear could have been, at a minimum based only on the information you can extract from that clear.

Roughly 1 min per tier… so if you scored 10 mins on a 145… similar scenarios you can clr 150.

But its not always the case, as i said some builds 140-150 are the same timings… 12 min on a 140 dont mean your limit is 14.xx 145… it can well be 12.xx 150s… builds such as FB MI… posion scythe etc.

You’re still not understanding me. I’m chalking this one up to language barrier, and calling it a day. No offense intended, you obviously have a lot of know-how when it comes to pushing GRs.