Warframe Revised: >100% Status Chance / Shotgun Megathread

[TheLexiConArtist]

1 hour ago, Andele3025 said:

DE decided to take per shot status net probability (as if measured by a regular rifle), multiply it by 3 then divide by pellet count flat instead of divide by pellet probability, this means, the more pellets a gun has above 3 (or its status per pellet times three divided by pellet count wasnt lower than 1), the more it got in practice nerfed when it comes to status chance compared to what was promised; you have over 5 pages total of people in this very thread explaining and complaining about it.

Very easy way to demo what the issue is. Get a Kohm with 4 60/60s, check its status output/spool 1, then spool 2 and spool 3 and finally fully spooled up, notice how because DE did it by pellet count, at spool 1 and even 2 shot kohm guarantees status per pellet and then proceeds to nerf itself as each part of the spool alters the base multishot instead of counting as modding/post calc multishot.

The Kohm is as much an exceptional case as the Exergis or Tigris are when discussing reloads. It's the only one that dynamically changes, so of course it's going to look weird. I don't trust Warframe's typical SpaghettiTech to put in a pure sliding scale that applies solely to the dynamic Kohms without exploding the game.

 

So I made a new chart to include all the shotguns in the list, allowing us to put this notion to rest unequivocally.

Using the base status chances per pellet given in the previous list as New/Old data series and working in 340% status chance using the old abomination of an equation as New/Old(3.4) data series.

New/Old gives us pretty much the chart we were looking at previously, although it's squashed up to the higher reaches now. We already know that has a minor negative or a positive depending on the Corinth and Strun Wraith outliers being normalised.

More interestingly is what we see in the New/Old 3.4  series. Looking at it immediately shows that there is a definite upward trend among most of the data we have, striking between Sobek and Basic Tigris up towards the Basic Strun. The Corinth, as before, has clearly benefitted the most from the change, and is a big outlier of a lower-pellet gun getting a great deal.

However, what we see in the under-par sector is populated precisely by the 100% status shotguns (minus the Exergis which we already established got an oddly good deal). The pellet count didn't matter. Only whether they could or could not hit 100% defined the nerf.

 

So from now, your factual arguments to debate are exclusively:

1) The Exergis was a 100% status shotgun and did not get reduced in potential like the other 100% status shotguns did

2) The Corinth was afforded significantly higher potential in the new system than other shotguns generally were.

 

Okay?

Edited March 12, 2020 by TheLexiConArtist
turns out I made an equal oopsie on both parts, cancelled out to the same charted outcome, lul

[Andele3025]

2 minutes ago, TheLexiConArtist said:

The Kohm is as much an exceptional case as the Exergis or Tigris are when discussing reloads. It's the only one that dynamically changes, so of course it's going to look weird. I don't trust Warframe's typical SpaghettiTech to put in a pure sliding scale that applies solely to the dynamic Kohms without exploding the game.

The Kohm is just a demonstration of how from implementation DE #*!%ed up the status calc change and nerfed shotguns based on pellet count.
At each level of spool up the kohm most likely uses a branch variant of its stat table (since tick based alterations of stats on all games tends to be done in a variant of that to not pointlessly take up memory re-calculating something that remains consistent), thus as DE did the stat changes, it applied to all branches of the kohms spool up to the point where it nerfs itself (or alternatively seen, where it gets buffed the lower its spoolup rank is). DE used per hit probability instead of status chance per pellet as the baseline to divide across pellets. Or mathematically, imagine that the 2.36~7 whatever the Kohms base status was (that was consistent across all levels of spool) is now getting a variable modifier based on current pellet count based on spool up instead of having been properly calced and than having it flat across levels of spool. Similiarly thus, all shotguns got nerfed based on pellet count compared to what was promised.

2 minutes ago, TheLexiConArtist said:

So from now, your factual arguments to debate are exclusively:

1) The Exergis was a 100% status shotgun and did not get reduced in potential like the other 100% status shotguns did

2) The Corinth was afforded significantly higher potential in the new system than other shotguns generally were.

Okay?

No, you are free to keep ignoring the point tho.

[TheLexiConArtist]

4 minutes ago, Andele3025 said:

The Kohm is just a demonstration of how from implementation DE #*!%ed up the status calc change and nerfed shotguns based on pellet count.

THEY

DID NOT

NERF

BASED

ON

PELLET

COUNT

It has been CONCLUSIVELY proven at this point.

Neither the status chance buffs compared to previous base values OR the actual proc-per-second performance in new system versus old supports this claim.

Unless you can provide hard mathematical evidence to contradict mine, you are simply deluding yourself.

[CopperBezel]

Yeah, shotguns always had lower per-pellet status chance if they had a larger number of pellets, the new status chances are formulated from an approximation of the old unmodded base pellet status chance, and the approximation becomes less accurate with higher SC, so that 30% base shotguns lost as much as 15% of the per-pellet chance they would have had relative to other shotguns in the new formulation, so you can say that high status chance shotguns got a small "nerf" or that all shotguns previously capable of 100% status builds got a huge nerf or even that high-pellet-count shotguns have had and still have bad status, but the relationship between high pellet count and underperforming is the one thing that has definitely not in any way changed, either in relation to the previous formulation or relatively among existing shotguns. 

Anyway, hey, have they nerfed Viral yet?

Edited March 12, 2020 by CopperBezel

[Andele3025]

37 minutes ago, TheLexiConArtist said:

It has been CONCLUSIVELY proven at this point.

Neither the status chance buffs compared to previous base values OR the actual proc-per-second performance in new system versus old supports this claim.

Unless you can provide hard mathematical evidence to contradict mine, you are simply deluding yourself.

Except they did nerf by pellet count and both the math and tables actually using in game stats instead of you making up numbers show so.

14 minutes ago, CopperBezel said:

but the relationship between high pellet count and underperforming is the one thing that has definitely not in any way changed, either in relation to the previous formulation or relatively among existing shotguns.

Except you are forgetting that because DE based prior status on per shot probability, when converting they multiplied it and then divided by raw pellet count instead of doing it by probability of each pellet from new status chance, the more pellets a gun had, the more it got nerfed proportional to its status chance. Its why the Tigris Prime per pellet status chance isnt 25% (thus 1-0.75 to the power of 8, being the 90% probability/3x30% that was promised per shot) which it would have been if they tripled the probability per shot and then calculated status each pellet should have to hit that, but its at 11% instead.

[nslay]

The P = S/N gaf from DE is a first order Taylor approximation of the right way to calculate per-pellet probability

P = 1-(1-S)^(1/N)

First order Taylor approximation

P ~ S/N (but only for S ~ 0).

There's some math floating around in one of these threads that I wrote. I show a 2nd and 3rd order Taylor approximation which would do a better job for even Strun Wraith.

But yes, owing to the way roots work in (0,1) (they make values in (0,1) larger) and then complementing the rooted quantity (i.e. 1 - root value) means that higher numbers of pellets are still more severely penalized regardless of the approximation used.

I think DE should use the mean value interpretation. It salvages their botched approximation and it's more intuitive for players to think in terms of expected number of pellets than the formula that requires a calculator.

[CopperBezel]

Not quite there @Andele3025.

The original status chance was 30% base for one pellet in a shot to proc. That meant in 8 individual rolls, there was a 30% chance of a single proc.

That means there was a 70% chance of no rolls. That means 70% represents eight trials, the probability of trial 1 x trial 2 x trial 3 and so on. It's the individual trial chance to the 8^th power.

Take 70% to the eighth root. That's a 95.6% chance that the individual pellet will not proc, or a 4.4% that it will.

A 4.4% chance times 3 is 13%. 

Now, 11.6% is in fact lower than 13%, but you'll find that fiddling with the number of pellets doesn't change this very dramatically. What makes the approximation used, total chance divided by pellet count, break down, is actually that diverges lower as the status chance increases. Do the same process for a 10% status shotgun, and it's much closer; do the same process for a 30% shotgun with 10 pellets, and it's almost identically wrong. 

[nslay]

For high status shotguns, using the correct value over P ~ S/N can make as much as a 25% difference in likelihood ratio (for a 10 pellet shotgun). So even if the approximation looks similar, it does underestimate by quite a bit for the higher status shotguns.

But really... countless have complained about the "weird math" of the per-pellet chance. Let's forget about the per-shot and per-pellet relation and think in terms of expected number of pellet procs (S = N*P)! Then DE need only tune S to be the number of pellets it thinks your unmodded shotgun should proc on average. Then you get P = S / N which is what they use already!

[CopperBezel]

Yeah, it's higher-status shotguns that got underbuffed, not high-pellet-count ones. In the case of the Tigris Prime, by 15%.

Anyway, there's nothing weird about per-pellet math. It's exactly identical to all other weapon types in the game and DE shouldn't have expected players to learn a new formula for one class of guns in the first place. I haven't seen a single person misunderstand it or complain about it. At least half of players didn't understand the old math, and most of those players are angry about the new relationship, but that's not misunderstanding per-pellet math or finding it weird, that's being upset with the assigned status chances of the guns and the formula used to arbitrarily set them, which of course reflects part of the old math but nothing in particular of the new.

Procs per shot could work perfectly if DoTs didn't exist, but they do, and since your damage is divided among pellets, so is status-inflicted damage. So you have to know exactly what proportion of your shot is proccing, which is exactly the number that per-pellet status chance is. 

Edited March 12, 2020 by CopperBezel

[nslay]

1 minute ago, CopperBezel said:

Procs per shot could work perfectly if DoTs didn't exist, but they do, and since your damage is divided among pellets, so is status-inflicted damage. So you have to know exactly what proportion of your shot is proccing, which is exactly the number that per-pellet status chance is. 

That per-pellet chance only tells you how many pellets proc status. Not how many proc useful DoT status effects. They know that players will be forced to use 60/60 mods to reach an appreciable status chance... but these also change how many useful DoT effects the player can dish out in a shot.

And then, most shotguns don't naturally dish out useful DoT procs. Tigris Prime was one of the only (if not only) counter example in the old system... you could get half your pellets to proc Slash with the 4 60/60 mods owing the 4x weighting scheme and 100% per-shot status chance. But I see talking about Boar Prime... wha? Boar Prime is especially mostly proccing Impact in the old system. Same with Strun Wraith! The removal of 4x weighting does help these shotguns proc more useful status effects, but you still have to pollute their proc priorities with those 60/60 mods and you still need to surpass the awful innate damage values on most shotguns to get useful DoT effects.

So... I don't see the DoT effects being a problem now because, well, 60/60 mods and proc priorities will still gimp them like they did in the old system.

[Sevek7]

Hi, I have some feedback about the status changes. Several of the new statuses have a large chunk of bonus on the first proc, with smaller bonuses on subsequent procs. 

As it stands currently, crit weapons can basically get the majority of the bonus from status effects even with a low status chance, since the first proc gives such a big bonus. A crit-based automatic rifle therefore gets the best of both worlds (Even stradavar with 12% status chance will get 5 viral procs on very fast!). Status weapons can get the full bonus from status, but don't really gain a lot from crit. 

From my testing, this makes crit weapons better than status weapons 100% of the time. So I suggest we flip the calculation. Let's have the first proc give a small bonus, and have the bonuses increase on subsequent procs, until the 10th proc gives the big chunk. I think this might make status weapons compete with crit weapons.

[NuclearCoffeePot]

Gas damage needs to be flat out changes. I could care less about blast cause i rarely use it but gas is completely useless. It's gas dps needs a secondary effect. Right now ots only use is against infested in which case a slash viral heat build will do that better. 

My solution-toxic gas is terrifying. With a single breath you can choke to death with your lungs melting and bleeding. So effectively in the real world gas is like max health damage that ignores armour and shields- so gas in warframe could potentially

A: be a damage type that has a damage increase on its ticks based on enemy level

B:Increases in damage the longer enemies are affected by it

C:could apply other status effects on its tic damage

[CopperBezel]

12 minutes ago, nslay said:

That per-pellet chance only tells you how many pellets proc status. Not how many proc useful DoT status effects. They know that players will be forced to use 60/60 mods to reach an appreciable status chance... but these also change how many useful DoT effects the player can dish out in a shot.

And then, most shotguns don't naturally dish out useful DoT procs. Tigris Prime was one of the only (if not only) counter example in the old system... you could get half your pellets to proc Slash with the 4 60/60 mods owing the 4x weighting scheme and 100% per-shot status chance. But I see talking about Boar Prime... wha? Boar Prime is especially mostly proccing Impact in the old system. Same with Strun Wraith! The removal of 4x weighting does help these shotguns proc more useful status effects, but you still have to pollute their proc priorities with those 60/60 mods and you still need to surpass the awful innate damage values on most shotguns to get useful DoT effects.

So... I don't see the DoT effects being a problem now because, well, 60/60 mods and proc priorities will still gimp them like they did in the old system.

Right, to know the chance of any given status effect, you still have to multiply by the proportion of your damage that's in that type. But that's true of any status effect, and at least removing the 4x on physical weighting means it's a straightforward proportion. It also means Slash is very hard to build for.

So I'm really thinking Toxin, Fire, Electric, and Gas if they fix it, not so much Slash, because pure-IPS builds were always-only-ever about Slash and now they're kinda dead I guess? Viral + Heat is a good combo right now, and you could possibly add Fanged Fusilade and replace that Scattering Inferno with Shotgun Savvy on a Tigris Prime for, er, 40% of the slash you'd have got out of an old-system 100% status build. 

The thing about the current layout is that it does account for everything somewhere. You could do more complicated things to more fully account for everything, but nothing is left out. It's not invisibly giving you lower DoT than you expect because of something happening on the backend. Meanwhile, if you really want to go more complicated, I feel like we'd have to abandon number of hits as the baseline for status in general and find a way for a sniper rifle to get as much use out of status as a shotgun or machine gun can.

Edited March 12, 2020 by CopperBezel

[nslay]

35 minutes ago, CopperBezel said:

Meanwhile, if you really want to go more complicated, I feel like we'd have to abandon number of hits as the baseline for status in general and find a way for a sniper rifle to get as much use out of status as a shotgun or machine gun can.

Well, that's the novelty of a shotgun! The sniper has a combo counter... that's the novelty of a sniper. The fully automatic rifle is well designed for crit and hybrid builds (which was not the case for most shotguns). That's the novelty of fully automatic rifles.

Status doesn't need to be (and shouldn't be IMO) equally effective among different weapon types. Status used to be shotguns' thing. In fact, making status/crit equally effective among different weapon types waters down the novelty of different weapon types considerably!

[CopperBezel]

Then let's leave the math as it is but buff some shotty status chance numbers so they can use DoTs effectively and at least get the full range of status effects to choose from! = ]

[nslay]

6 minutes ago, CopperBezel said:

Then let's leave the math as it is but buff some shotty status chance numbers so they can use DoTs effectively and at least get the full range of status effects to choose from! = ]

Yeah, agreed!

I do feel like there was a disconnect between DE and how players used shotguns. DE did buff shotguns for sub 100% status... But nobody used shotguns that way before! And I'm pretty sure the old DE picked status chances for a majority of shotguns so that 100% could be reached with 60/60 mods, Shotgun Savvy, Motus Setup and Nano Applicator! So why did they feel it was a good idea to do this to shotguns now?

[Andele3025]

3 hours ago, CopperBezel said:

Not quite there @Andele3025.

The original status chance was 30% base for one pellet in a shot to proc. That meant in 8 individual rolls, there was a 30% chance of a single proc.

I know

Quote

Take 70% to the eighth root. That's a 95.6% chance that the individual pellet will not proc, or a 4.4% that it will.

A 4.4% chance times 3 is 13%.

And while true, the point is  thats not what DE stated they would do, they said they would multiply the shotgun status three times. Buffing per pellet status chance doesnt do that in any calculation, it never even gets close to that unless the shotgun is very low on pellet count (or wasnt a status shotgun at all but for those it was DEs intent to make them actually able to proc at least a bit as stated during the stream).

 

Edited March 12, 2020 by Andele3025

[(PSN)Irmao2Vigilante]

Hello,

On 2020-03-12 at 8:17 PM, nslay said:

For high status shotguns, using the correct value over P ~ S/N can make as much as a 25% difference in likelihood ratio (for a 10 pellet shotgun). So even if the approximation looks similar, it does underestimate by quite a bit for the higher status shotguns.

But really... countless have complained about the "weird math" of the per-pellet chance. Let's forget about the per-shot and per-pellet relation and think in terms of expected number of pellet procs (S = N*P)! Then DE need only tune S to be the number of pellets it thinks your unmodded shotgun should proc on average. Then you get P = S / N which is what they use already!

An important fact is expected value are linear, whereas variability isn't. This fact was not taken into account before. As consequence the status shotgun mains were severely nerfed. Strun Wraith could achieve 100% status probability per pellet using only 3 of the 60/60 mods. Now, it has 12% chance meaning that using all 4 60/60 mods one will get less than 50% probability per pellet. So, it is a fact - it was nerfed - why? because the equation they design to transition between the old system and the new one was designed as being linear equation addressing a non linear problem.

There are many good procedures that could have been applied. For instance, adjust  shotguns that achieved 100% status chance with 3 or 4 mods so it gets closer to this number (lets say 98 ~100%) with the same amount of mods and then, using one of them to redefine the other shotguns status probability proportionally. This would altered the game without drastically nerfing things, preserved part of the dynamic without allowing things to go out of hands.

However, i imagine there was too many things to be done for this update, and little to no time was devoted to reason about how probabilities work. They said themselves, they "pick up the shotgun box and shook it".

Hopefully, they also said they would fix if things were too broken - well things are broken. So DE Team, please, will you easy our concerns?  please tell us if you intend to fix this soon?

  🥺  please, tell us this time you actually sat down and worked the numbers properly. and Pretty please,  🥺  tell us that you prepared things in advance so a hotfix will work for consoles as well as for pc so we don't have to wait months with bugged shotguns in warframe - (the bugged build is in cert now - Oh God )

 

 

 

[CopperBezel]

Oh, I don't think that was overlooked or accidental at all. Smashing the radical function curve down into a straight line so that status chances other than 100% on shotguns meant something was most of the point. They've just waited so long to do it that all of the actual shotguns in the game are balanced around the expectation that this is how things work, so it completely pulls the rug out from under high-status shotguns as a set.

[nslay]

5 hours ago, CopperBezel said:

Then let's leave the math as it is but buff some shotty status chance numbers so they can use DoTs effectively and at least get the full range of status effects to choose from! = ]

You know, my previous suggestion wasn't an actual change in the math. It was merely a new interpretation of the same math they coded... mainly to reconcile the strange seemingly wrong thing DE did (but you know, they might have the same idea... just didn't communicate it!). But also the interpretation lends itself well to overstatus and I think balancing because it's easier to think about (for developers and players alike).

Think of these interpretations in terms of shotgun models. Every shotgun produces N pellets that each independently proc status with probability P. So a shotgun shot is represented by the Binomial distribution. Now in the old system, DE parameterized the shotgun in terms of N and the "per-shot" status chance S. Here S is the probability one or more pellets procs status of the N pellets. So you have this shotgun model:

Shotgun(N,S): P = 1 - (1-S)^(1/N)

But you don't have to parameterize the shotgun like this. You could parameterize in terms of N and mu, where mu represents the average number of pellets that proc status.

NewShotgun(N,mu): mu = P*N

Since the Binomial distribution B(N,P) has a mean of P*N then it follows that mu = P*N. You can even use this model to characterize strange guns like Kohm that spool up without any loss of generality. It's just that you calculate your P differently compared to the original model. But this doesn't change the math... we're still sampling the Binomial distribution!

Well so what? Well, instead of Shotgun(N,S) where I have to think about S as the probability that one or more pellets proc status (which tells me little without a calculator), I can now simply think of the average number of pellets that proc status in NewShotgun(N,mu). That's nice and easy for me! For example, if mu = 3, then on average 3 of my pellets proc status (it's still random though!).

And NewShotgun(N,mu) works with DE's strange new change.. because mu = 3*S in this current patch! Here S is the base "per-shot" status chance from the old system.

Now why did DE pick mu = 3*S? That seems nonsensical to me. It ruins most shotguns because they tend to have low crit (why do you need any crit when you could fully leverage 100% per-pellet proc chance?). Only a handful can effectively leverage crit to workaround the nerf. And of course, it ruins the novelty aspect of shotguns as the defacto weapon for dishing out status!

NewShotgun(N,mu) also works with overstatus with a tweak in interpretation. If per-pellet chance P is no longer bounded by 1, mu represents the average number of status procs (rather than pellets). Don't misunderstand me... you still use mu = N*P... but for example, if P = 2, then each and every pellet is proccing two status effects.

Anyway, NewShotgun(N,mu) reconciles DE's weird pellet chance calculation as well as overstatus. And it doesn't change any math that is not currently already in the game. This is merely a different interpretation of the same math.

Now the question is... can DE pick mu that makes more sense (3*S doesn't make any sense to me). Mu should be picked to keep the original use and theme of the shotgun as much as possible (and that ain't mu = 3*S!).

 

Edited March 13, 2020 by nslay
s/at least/one or more/g

[CopperBezel]

Well, 3*S just made a pretty line that stayed above most of the shotgun status curve before it went vertical. Again, I think the correct value of mu would have been calculated from the formula of "hand-set for this weapon based on balance considerations". 

But yeah, I don't think "expected status events per shot" would have fit into the arsenal text line. = P They added pellet count in the same update, so mu is there, you just have to multiply that line by the status chance.

[nslay]

8 minutes ago, CopperBezel said:

Well, 3*S just made a pretty line that stayed above most of the shotgun status curve before it went vertical. Again, I think the correct value of mu would have been calculated from the formula of "hand-set for this weapon based on balance considerations". 

But yeah, I don't think "expected status events per shot" would have fit into the arsenal text line. = P They added pellet count in the same update, so mu is there, you just have to multiply that line by the status chance.

Right, I mean the UI is not necessarily showing the parameters of NewShotgun(N, mu) even though that might be how it's working now. The old UI could have been changed to show the same status/projectile information just as well. Instead, they showed the "per-shot" status which wasn't the most informative.

[CopperBezel]

The old UI didn't even show pellet number. Or sum the total damage for convenience. I contest the idea that it could have been changed to show anything well.

[Drago55577]

47 minutes ago, CopperBezel said:

Oh, I don't think that was overlooked or accidental at all. Smashing the radical function curve down into a straight line so that status chances other than 100% on shotguns meant something was most of the point. They've just waited so long to do it that all of the actual shotguns in the game are balanced around the expectation that this is how things work, so it completely pulls the rug out from under high-status shotguns as a set.

It still should have used 100 status shotguns as the basis for the linear equation. 

[Nox_Terminus]

• Akbronco's status chance is currently not modified beyond dividing its status chance by pellets.
• Fusilai's alt-fire, which uses shotgun status calculation received boost to its status chance.
• Arca Plasmor, despite being nerfed twice, received no status improvements. Unlike most hybrids the Arca Plasmor cannot headshot, fires relatively slowly, has a limited range, and a long reload. It also has a fairly low crit multiplier despite being a hybrid.
• Astilla, which is already an okay shotgun, received no status improvements.
• Lastly why were these status changes not at least listed in the patch notes? It would've made it much easier to see what actually changed statistically.