a tale of 2 players; or yet another pve scaling thread

Unknown

3 2*s are picked fairly at random for non heroic events to be boosted +40 levels. For example, Moonstone is boosted in Simulator Basic. Heroic events would only boost hero due to roster selection. In fact, we can just look at Simulator Basic to see why this is a problem. The boosted 2* (+40) are Moonstone, Storm, and Thor. The boosted 3* (+30/+90 on the rotating 3*) are IM40, Mystique, and Hulk. Since the +90 is rotated we'll just take the average and call that +50 levels per 3*. However, +50 levels on a 166 is a smaller portion of total levels (less than 1/3) compared to +40 levels on a 94 (about 40%), and since scaling is a multiplier the +40 levels on the 2* has a bigger impact. And then there's the character selection. 2*s are fewer in numbers and generally self contained, so there's a good chance the boosted character is someone you have leveled. For 3*, there are a lot of specialized characters (e.g. Falcon) and of course since there are way more of those it's far less likely you've all of them leveled. Any event immediately following a new character is going to feature a 3* that's practically useless (and equally good for anyone since you only have covers from the previous event). Looking back at this roster, Storm and Thor are immediately useful for 2*, and Moonstone is playable. The 3* counterparts are not nearly as big impact players compared to Storm, and that's if you have them maxed in the first place.

There's also the problem that since scaling cannot be undone, roster diversity suffers as you improve your characters. For example let's imagine we have the following rosters with 25 3* max covered with these levels:

1. Everyone at 94
2. 10 at 166, 15 at 94
3. Everyone at 166.

While you'd think the performance/diveristy of the roster should be increasing in the order listed, in this game the order from most to least effective would be 1 > 3 > 2. Note that while 1 & 3 are equally diverse, 1 likely has lower scaling as a % of roster, and for having everyone at a lower level this makes the +30 level buff a bigger increase compared to the one with everyone at 166. The first person also has less health which means shorter regen time. If we assume the two roster fought an identically difficult battle we can say both win while taking 50% damage, but 50% damage to the first roster is takes only about 60% the time to heal compared to the third roster, so the guy with everyone at 94 can more easily rotate guys around knowing it takes a shorter time before the guys injuried in the initial fight. And forget about the second roster. This guy is in a strictly inferior position since his enemies will be tuned against his 166s so the other 15 3*s are 94 is unlikely to do anything useful.

Scaling basically ensures only guys with level comparable to your highest guy can be considered as even a plausible contributor, but as you have guys at 166 or 270 this makes your roster less diverse because you're not going to be able to get everyone up to 166 and certainly not every 4* to 270 without a ton of resources. This isn't anything new though. It was always this way but X Force was overpowered enough to overcome this, but now he's not. On a particularly cheap node I can still have a decent shot at beating it with 3 of my 166s, but I have a lot less 166s to use compared to a guy who has everyone at 94 (or 100, or whatever), which costs significantly less resources than I do. In fact you can have a strictly stronger roster, like the roster 2 versus roster 1 example, and you'd still be screwed. How can a roster that has everyone in another roster at an equal or higher level be less diverse? But currently 25 guys at level 94 is considered as far more diverse than 10 guys at 166 and 15 guys at 94.

One very quick bandaid fix would be to make PvE boosts all percentage based too. Scaling is likely still screwy but at least you're not double penalized when a guy with a weak roster receives a much bigger relative boost to his roster for being weak. I'm guessing a static number is used because they want the 1/0/0 level 40 guy you won from last event feel like he's useful, but it's not like a level 130 1/0/0 is really all that powerful so I think people will live if you decrease the major buff to say 75% so that he's just a level 70 1/0/0 guy as opposed to level 130 1/0/0 guy.

fmftint

We've been discussing this internally but I'll share my numbers here as well. First and foremost I haven't been playing this competitively, I've posted less than 20k this event and have only played 2 rotations on the Latveria sub.
My T4 average is 166
March of the Bots24 (14%)
Identical Doom23 (13%)
Mystique's Trail34 (20%)
Brotherhood of Doom41 (24%)
Face to Face48 (28%)
Wrath of Doom167 (100%)
The Heart of Castle Doom183 (110%)
Lab Work191 (115%)
The Big Boom230 (138%)
Strange Allies75 (45%)
Latverian Brotherhood117 (70%)
Metal on Metal115 (70%)


Even though I've played less than you my % vs roster strength is comparable to your's

Unknown

fmftint wrote:

We've been discussing this internally but I'll share my numbers here as well. First and foremost I haven't been playing this competitively, I've posted less than 20k this event and have only played 2 rotations on the Latveria sub.
My T4 average is 166
March of the Bots24 (14%)
Identical Doom23 (13%)
Mystique's Trail34 (20%)
Brotherhood of Doom41 (24%)
Face to Face48 (28%)
Wrath of Doom167 (100%)
The Heart of Castle Doom183 (110%)
Lab Work191 (115%)
The Big Boom230 (138%)
Strange Allies75 (45%)
Latverian Brotherhood117 (70%)
Metal on Metal115 (70%)


Even though I've played less than you my % vs roster strength is comparable to your's

I'm currently #2 overall so I obviously played quite a bit. My T4 average is 216 for reference. Here's what I got for the harder nodes:

The Big Boom - 243
Lab Work - 207
The Heart of Castle Doom - 204
Wrath of Doom - 190
Metal on Metal - 158
Latverian Brotherhood - 134
Strange Allies - 121

One thing that immediately strikes me as odd is that my hardest node (The Big Boom) is only 13 levels higher, or barely 5% harder. However, for the essentials, Strange Allies is 46 levels higher, or more than 50% harder and that is indeed quite a difficult node. So we can already see how scaling doesn't make sense. My roster doesn't seem too much stronger than yours, but there's an incredible 50% jump in difficulty on my essentials compared to yours, while your highest node is only 5% lower than mine. This does confirm what I suspect that scaling is broken on the lower end. That is, our highest node look fairly reasonable (mine is 5% harder, and maybe my performance roster is indeed 5% better than you), but there's no way that 50% difference in the essentials, which is where a ton of points are at, can possibly make sense. This is also consistent with how every huge scaling event (Prodigal Sun, Gauntlet, Simulator Hard) tend to be dominated by strong rosters, since it sure looks pretty fair on the highest scaling nodes, at least compared to the lower scaling nodes.

Edit: decided to put all the data side by side:

The Big Boom (highest level node)

Davyx - 195
tintguy - 230 (+18% compared to Davyx)
Phantron - 243 (+25% compared to Davyx)

Metal on Metal (highest level not goon only essential node)
Davyx - 85
tintguy - 115 (+35% compared to Davyx)
Phantron - 158 (+86% compared to Davyx)

Look at the shocking increase of my level on the essential node that's worth a ton of point and actually very hard (Quicksilver is an absolute deadweight against Doom with a strong AP pump). There's no way this set of data can be fair within itself. I think I remember seeing Davyx's roster and +25% looks like a reasonable estimate in the difference between the roster strength, but +86% is definitely not. Those two numbers are so far apart that one of them has to be wrong.

Unknown

Interesting observation on the essential. For what it's worth, I left my QS at base level 40 (boosted to 130). Not sure if that factors into this.

I guess I'll take it as a compliment that you think my 137 roster is 80% as good as your 216 roster (level-wise a 173 roster +25% = 216)

The scaling does seem pretty weird, but I rather have an easier Big Boom than an easier Metal on Metal. Here's the percentage comparison:
Davyx, average top 4 powered up 137, Big Boom is 195, or +58 Lvs (42% higher than roster), Metal on Metal 85, or -52 Lvs (40% lower than roster)
Phantron, average top 4 216, Big Boom is 243, or +27 Lvs (13% higher than roster), Metal on Metal 158, or -58 Lvs (27% lower than roster)

cletus1985

Unless you guys are in the same bracket the differences in certain nodes could easily be from community scaling, if the guys trying to jump ahead of the pack are grinding down the high point essentials that would explain a much higher level in one bracket compared to the other.

I'm still sticking to my theory of the RNG hating certain people and taking over their scaling until someone proves me wrong though.

jackstar0

Just one random note, I just saw one of my nodes jump about 30 lvls and I realized that the last time I hit it the board just lined up for me. I had things fall into place, made good moves and had the match over quickly. I do not have XF as part of my playable roster. I've only got 2x 166s at this point, so my team consists of Patch166/Blade134/LDaken166. I burned through everything and other nodes did not jump nearly as much as Big Boom did (jumped from 170 to 0ver 200). Meanwhile Metal on Metal is a whopping 82.

So it might be a weird part of things... but maybe # of moves per match is also a factor? It'd be weird... but does seem like a thing they could track and use in their metrics.

Food for thought...?

Spoit

cletus1985 wrote:

Unless you guys are in the same bracket the differences in certain nodes could easily be from community scaling, if the guys trying to jump ahead of the pack are grinding down the high point essentials that would explain a much higher level in one bracket compared to the other.

I'm still sticking to my theory of the RNG hating certain people and taking over their scaling until someone proves me wrong though.

He specifically said that they were in the same exact bracket

And yeah, I'm 90% sure placement doesn't matter for scaling. I still have about the same scaling, and haven't played PvE seriously since like, before KK's release

Unknown

Even if each node has its individual community scaling (which seems pretty unlikely if you look say how all the nodes in Simulator Hard are almost always going up in unison) you'd have to have a bracket that has an extremely different behavior to see that kind of difference between two nodes. I don't know if the difference between The Big Boom is fair but it's at least closer than the difference between Metal on Metal. Like I said this is quite consistent with the observed fact that all high scaling events tend to be dominated high roster, as our highest nodes look considerably closer compared to the essentials. My QS is at the base level 40 too, so there's no reason why there should be an extra large difference in the scaling on that node even if scaling accounted for these things.

Salgy

Phantron wrote:

The Big Boom (highest level node)

Davyx - 195
tintguy - 230 (+18% compared to Davyx)
Player A - 232
Phantron - 243 (+25% compared to Davyx)

Metal on Metal (highest level not goon only essential node)
Davyx - 85
tintguy - 115 (+35% compared to Davyx)
Player A - 115
Phantron - 158 (+86% compared to Davyx)

Thanks for all this... Even if it doesn't make the most sense to me...
I'll let you guys do the math! Player A has a roster slightly stronger than tint guy and has played this event just a little harder than him (45k total points)... Player B hit the top progression & took a break... So I don't have the info... He didn't even open up the Seed of Destruction node... Like phantron, player A's qs ideas left at level 40... Based on the average of the top 4, if scaling is based on top 4 character on your roster strength, Player A's scaling should be closer to right in between Phantrons & tint guys, not at tint guys level... No?

Unknown

I don't think it's that important to figure out what is responsible for scaling as opposed to discovering how it impacts you. If you look at my nodes, you'll find that the closer we get to the highest node (The Big Boom), the closer the levels become. For example if you look at The Heart of Castle Doom we have 139/183 (+32%)/204(+47%) , while for The Big Boom we have 195/230 (+18%)/243(+25%), while for Metal on Metal it's +35%/+86%. This shows a trend where the lower level the node is, the bigger the advantage a weaker roster has. This sounds unintuitive but is quite consistent with my observed behaviors. In any 'crazy scaling' event I generally have a much easier time to fend off the weaker rosters, while for anything that's pretty easy those turn out to be quite difficult to beat the weak roster. If events function similar to this then this makes total sense because the crazy scaling events put the nodes much closer between players so I'm not pressured by weaker rosters, but for the low level nodes, even if their difficulty turns out to be trivial for both (and Metal and Metal is definitely not trivial for me), the fact that my enemies have 86% more level and hence health alone will be more than enough as a deciding factor if they're indeed both trivial because I sure am not doing 86% more DPS during crunch time run. It seems to me scaling is unusually generous for a node in the essential branch for some. The difference between 85 (Davyx) and 158 (Phantron) is just far greater than what our roster can possibly explain, and since the essential tend to be the nodes worth the most points, having an edge here would indeed translate a significant edge overall which explains why 2*/transition rosters are doing much better than usual.

cletus1985

Phantron wrote:

Even if each node has its individual community scaling (which seems pretty unlikely if you look say how all the nodes in Simulator Hard are almost always going up in unison) you'd have to have a bracket that has an extremely different behavior to see that kind of difference between two nodes. I don't know if the difference between The Big Boom is fair but it's at least closer than the difference between Metal on Metal. Like I said this is quite consistent with the observed fact that all high scaling events tend to be dominated high roster, as our highest nodes look considerably closer compared to the essentials. My QS is at the base level 40 too, so there's no reason why there should be an extra large difference in the scaling on that node even if scaling accounted for these things.

The way goon nodes jump outrageously compared to regular nodes is why I've thought that individual nodes carry their own scaling (Although whether or not that's personal, community, or both is still a mystery). While all nodes seem to scale up consistently, goon nodes scale a lot quicker than most which implies to me that each node bases scaling somewhat separately.

Unknown

cletus1985 wrote:

Phantron wrote:

Even if each node has its individual community scaling (which seems pretty unlikely if you look say how all the nodes in Simulator Hard are almost always going up in unison) you'd have to have a bracket that has an extremely different behavior to see that kind of difference between two nodes. I don't know if the difference between The Big Boom is fair but it's at least closer than the difference between Metal on Metal. Like I said this is quite consistent with the observed fact that all high scaling events tend to be dominated high roster, as our highest nodes look considerably closer compared to the essentials. My QS is at the base level 40 too, so there's no reason why there should be an extra large difference in the scaling on that node even if scaling accounted for these things.

The way goon nodes jump outrageously compared to regular nodes is why I've thought that individual nodes carry their own scaling (Although whether or not that's personal, community, or both is still a mystery). While all nodes seem to scale up consistently, goon nodes scale a lot quicker than most which implies to me that each node bases scaling somewhat separately.

Scaling is a multiplier and goon only nodes tend to have much higher starting levels for being goons only. The Magneto + The Hood node in TaT has extra high levels for being short handed and it also scales much quicker than normal nodes because it started out that much higher. The Hulk/Thor solo nodes starts out even higher and scales to 395 faster than anything else in the game for the same reason.

cletus1985

Phantron wrote:

cletus1985 wrote:

Phantron wrote:

Even if each node has its individual community scaling (which seems pretty unlikely if you look say how all the nodes in Simulator Hard are almost always going up in unison) you'd have to have a bracket that has an extremely different behavior to see that kind of difference between two nodes. I don't know if the difference between The Big Boom is fair but it's at least closer than the difference between Metal on Metal. Like I said this is quite consistent with the observed fact that all high scaling events tend to be dominated high roster, as our highest nodes look considerably closer compared to the essentials. My QS is at the base level 40 too, so there's no reason why there should be an extra large difference in the scaling on that node even if scaling accounted for these things.

The way goon nodes jump outrageously compared to regular nodes is why I've thought that individual nodes carry their own scaling (Although whether or not that's personal, community, or both is still a mystery). While all nodes seem to scale up consistently, goon nodes scale a lot quicker than most which implies to me that each node bases scaling somewhat separately.

Scaling is a multiplier and goon only nodes tend to have much higher starting levels for being goons only. The Magneto + The Hood node in TaT has extra high levels for being short handed and it also scales much quicker than normal nodes because it started out that much higher. The Hulk/Thor solo nodes starts out even higher and scales to 395 faster than anything else in the game for the same reason.

I've also seen goon only nodes at roughly the same level as regular nodes starting out scale to 395 while the regular nodes scale up to levels in the 200's. This may be a rarer occurrence, but it has happened. With that said, I haven't put in the time or effort to truly try to break down scaling and it seems like you have so I won't argue any further. The scaling mystery continues and it is truly mystifying how they justify the difference in difficulty based on the results shown just in this little thread.

Zen808

Phantron, do you remember what level your Metal on Metal started at? Like was it always near to the 150's, or did it start relatively low (100ish) and climb?

My Node started at the mid or late 70's, and only climbed marginally throughout the course of the sub.

(For reference, my overall scaling is about the same as Davy's).

Unknown

808SpicyToro wrote:

Phantron, do you remember what level your Metal on Metal started at? Like was it always near to the 150's, or did it start relatively low (100ish) and climb?

My Node started at the mid or late 70's, and only climbed marginally throughout the course of the sub.

(For reference, my overall scaling is about the same as Davy's).

It definitely didn't start below 100. Probably around 120-130 range? It didn't go up by a lot so I was thinking I had pretty good scaling until all the other guys say it's below 100.

Xenoberyll

What's even more aggravating for me than the stupid scaling is the competition in your brackets that they pit you against. When you play a close to perfect PVE and win top spot the next PVE will be vs other near perfect players, forcing you to play even harder or lose out. It's stupid to hear your friend made top 5 with half your points when you're fighting to make top 10.

Stop to punish winning and improving please!!!

Unknown

I just finished my bracket. Overall, #1, #3, and #4 have no one higher than level 94 on their roster while I finished #2. With the data collected from this thread it also offered a lot of insight in how the final sub broke down. From my point of view, I grinded all the easy nodes down to 1, did about 4 passes of the essential nodes, and gambled on about 2 passes through the rest of the hard nodes. Since I can see that the margin of separation at #1 in the last sub is just 1K, and combined with the data collected in this thread I can say that just being able to efficiently grind down the essentials down to 1 would've been enough. Why didn't I do that? Because I didn't have enough time and at level 150 for Mystique/Doom, they were not really all that easy to begin with since Quicksilver is a deadweight and if you miss your alpha strike move that fight can easily get nasty. I imagine at level 80 or even lower (the 94 only rosters is even lower than Davyx's roster) it'd be pretty safe to grind those nodes. Note that the Doombots with Mystique only have around 2500 HP even at level 150, so if you half that, that's something you can pretty much blow away with just +match damage boosts and Mystique is quite easy without the doombots. From the final score it is quite clear that even the weaker rosters can't grind the hardest nodes down to 1, or their margin of victory would have been far bigger than 1K. However, looking at the data we also see that of the 4 'hard' nodes, the difference between the rosters gets bigger the easier they are, and since they're all worth about 400-600 points base, being able to do the easiest one a couple more times also could've easily accounted for the 1K final margin. I, of course, didn't have that choice because even the easiest node Doom was more than capable of one shotting almost anybody with Diabolica Plot so I might as well go for the node worth the most points and this is obviously not as good a strategy but I don't have a choice here.

Now the interesting question is why does scaling suddenly break down on the essential nodes and the easier nodes of the hard path? I wouldn't say the scaling on The Big Boom was fair, but it's certainly a lot more fair compared to the other nodes just by looking at the levels. One theory I have is that the game must assume that anything below level X must be pretty trivial for a top roster. For example my essentials are at 150, which is 80% higher than Davyx, but I can see the game saying, 'well he's got 270s how hard can some level 150s possibly be?'. Likewise the 3 nodes that aren't The Big Boom had a level breakdown of:

Lab Work - 207
The Heart of Castle Doom - 204
Wrath of Doom - 190

I can certainly see the game figuring '200 can't be that hard for this guy', except they're plenty hard enough. But even if they've an easy mix of enemies, speed matters when you're in crunch time and if my trivial nodes have 50% more HP to go through while my roster is not capable of doing 50% more damage than yours, then I'm obviously at a significant disadvantage here. In the final stretch run I can see that even if I continue to win all my games the 2*s would definitely close the gap, simply because each game I play for the same value of points as they do took longer due to enemies having a higher % health compared to my roster's extra firepower, and of course the extra levels also contributed to greater risk of losing, but even if there was never any risk of losing, it'd still have been difficult to fend off the charge. Now all this can be overcome, especially if I spent even more time, but why should the guy with the better roster be playing from a disadvantageous position where every fight takes longer for the same number of points? I'm aware that 2* need help but you don't handicap the most powerful rosters to let guys with inferior roster win. At best I can accept a completely even field, e.g. if my roster does 2000 DPS and your roster does 1000 DPS then my guys are twice the level as yours. I don't think that's even fair not to mention it gives no incentive to improve your roster, but at least that's an even playing field. Currently it's more like I can do twice the DPS but my enemies have 150% more levels to make up for that, and even if I never lost a game, I'm likely to fall behind since I'm only doing 200% damage against an enemy that has 250% the health.

Bowgentle

Phantron wrote:

Bla bla poor me.

TL;DR: You still finished 2nd, even though people with an all-95 roster probably have an easier time.

So your point is?
That they should make it even easier for you to finish T1?

Sounds to me that scaling is working as intended when you still manage to T2, but have to work harder for it than someone with a **** roster who can't get the covers in PVP that you can get.

Unknown

I guess he is saying that the his essential being 80% higher than mine while his roster is only 60% higher than mine is a huge disadvantage because that essential Metal on Metal has 800 points and he needed to spend more time on it.

Let's say he needed to spend 50% more time on clearing that node, it makes a 15 minute grind into a 22.5 minute grind. That extra 7.5 minutes that he needed cost him 800 x 7.5 / 480 = 12.5 pts. Since all three essentials are unfair like that, he's at a about a 35 pt disadvantage. Despite this disadvantage he still manages to place top 2 so I think he is doing rather well.

Unknown

what if every event have different equation for scaling??

like,
event A. 50%(past) 30%(roster size) 20%(lvl) 10%(CS)
event B 40%(past) 20%(roster size) 35%(lvl) 5%(CS)