devtrackers.gg → PhreakRiot
PhreakRiot

PhreakRiot



13 May

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by topdrogon

Yeah I have no idea how to account for any of that. There’s no proper mathematical way of doing so. That’s why pro play shouldn’t be on the chart at all.

BUT that doesn’t mean that the chart itself is useless. If done correctly, I actually think your analysis has the beginnings of actually statistically correct analysis that can provide meaningful insights for the pro scene as well as soloq in general.

The best way of approaching this IMO would be to keep soloq analysis as soloq, and then draw conclusions from soloq as to new possible strategies and in what contexts they could be good for pro.

From the conclusions you draw from soloq, you can then form hypotheses as to what new strategy (champ/build/rune) might be OP. You then can test these hypotheses in customs in-house and in scrims.

If a strategy passes all these tests and then doesn’t perform well on stage, it doesn’t necessarily mean you were wrong (low srs he sample size). But it could mean...

Read more

It makes perfect sense and I agree with virtually everything you wrote.

At the end of the day, it'd be great if teams could test everything. Certainly, it's wise for them to at least test promising candidates. But ultimately they just can't. There just isn't enough time to hypothesis test everything in League of Legends. There are 5.9 * 1021 possible team compositions. Good luck getting enough data on each one.

So by necessity you have to cut corners. You have to go more general. There are billions of possible drafts after the first six bans and a first pick Rumble. No team in the world has accounted for all the possible mid/jungle matchups by the time they made that pick. So I think to some degree a rigorous testing regimen is just not possible anyway, so don't try to hold anything to that standard.

This isn't to say you shouldn't practice anything. Of course your should. But you realistically can't VoD review every game of every champion you pla...

Read more

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by Kyriios188

Ah my bad then

It really felt like an answer to their previous stream where they talked a lot about your tweet and your usage of soloQ data since it came right after

Yeah I feel that. It's a popular topic, so it makes sense. It's just a reply to the topic in general as opposed to any one piece of content.

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by qo3s17

Hey man! Think your content is great and the fact that you're interacting with people on reddit is awesome.

Some other people have mentioned it, but I think your grouping of Masters+ as one tier is misleading, because the skill difference between 0 LP Master and 1k LP Challenger is very significant.

Looking on u.gg, looks like Rumble jungle's Masters+ winrate is 47.71%. When you break it down between Masters, GM, and Chall, his winrate goes from 45.74% in Masters to 46.58% in GM to 51.45% in Chall. So his winrate isn't even dropping in higher skill brackets, the way you group the data just makes it seem like it is.

If you're looking at just 11.10 data, there are currently 49 recorded challenger Rumble jungle games. If you do 14 days, you see a pretty flat ~-3% win rate compared to the average jungler.

I'm interested in the sample being large enough that I'm confident in not having huge swings in win rate because of 1 result (keep in mind 55% win rate in 49 games is just winning 2 more than you lost and that's just unacceptable for this). Second, once I'm confident in low variance, I care about the trend. If a trend doesn't exist when travelling from bronze to diamond, why do you think one suddenly shows up when travelling from masters to challenger? It's possible for one to exist, but shouldn't be a leading theory.

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by Disafae

Wouldn't you expect that as Jungle Rumble is played in higher and higher skill brackets that the people playing against Jungle Rumble would better know how to punish an off pick like that?

And pros don't know how to do that? Why would this relationship suddenly flip?

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by Cahecher

The issue with the underlying point is that your entire argument only holds up based on the global data that is scewed towards EUW and NA. If we look just at the Korean server, your argument falls flat, since Rumble jungle winrate is actually high and continues to go up with ranks. Considering the initial point was that western teams and western players in general are behind on the pick, it actually supports the argument of LS and IWD.

Another thing, even though mid Rumble in Korea has slightly higher winrates, the amount of games is significantly lower, i.e. on u.gg there ~560 mid Rumble games and ~2900 jungle Rumble games tracked in dia2+(cumulative 11.9 and 11.10). So it is kinda expected that a lower pick rate would inflate winrate, it's a rather common occurrence. However, if you still want to use it to say that Rumble is not a jungler because his winrate is somewhat lower than in mid, you'd still be wrong - if a champion has 52%+ wr in both roles, I'd say it's good ...

Read more

Ooh, that's a good point, but also unsupported by data.

  • Korean Rumble's Mid minus Jungle win rate in bronze: +1.14.
  • Korean Rumble's Mid minus Jungle win rate in Master+: +3.12

Now I'll grant you that if you slice the data down far enough until you find data that supports you, you can eventually find data that starts to close the gap, but then you get a host of new issues. For example, there were under 500 games of Rumble mid in master+ Korean solo queue over the last two weeks. If we're going to debate 1-2% win rate shifts as indicative of champion performance, that's not enough samples to support those claims. You also have metagame issues that aren't relevant to pro play.

See, Korean solo queue is rife with early surrenders. Over the last two weeks in master+ Rumble jungle games, Korean solo queue had 23.25% of games end in 15-20 minute surrenders and another 33.23% end pre-25. Worldwide for the same skill bracket and time, it was 15....

Read more

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by topdrogon

You didn't use MMR as your independent variable. You used tier (Bronze - Masters+). The analysis SIGNIFICANTLY changes if you actually use MMR (more precisely MMR range) as the independent variable. I quickly made a visual where each cell = a standard unit of MMR range (100 LP). (Not sure if this is actually the case as obviously MMR and LP are not 1-to-1, but this is the best I can do as a non-Rioter).

https://i.imgur.com/g0VwYdL.jpg

You CANNOT group Masters+ when the range for Masters+ is LARGER than all of Iron - Diamond combined. Sure, you might have sample size issues at the highest of MMR ranges, but for popular champions the analysis would at least be somewhat correct if done in this way. I promise you that with enough games, the Lillia and Rumble plots you made would look parabolic and their winrates would get higher and higher as MMR increases from masters to GM and beyond. In that case, wouldn’t it actually be...

Read more

So, valid point on word choice.

However, keep in mind, pro play is only far to the right on things like coordination. Canyon may be rank 1 in EUW solo queue right now, but where should we put some of the other players? Cody from INF is 50-48 on his EUW account and is firmly in Master. If we're graphing how well champions play based on their hands and game knowledge... should he be to the right of challenger? If so, why? He's not even to the right of Master in solo queue. It's the same player. Yeah the coordination is different and I agree that one goes off the graph. But not hands. Not laning. Not positioning. Not target selection.

How about his teammates? His opponents? Should we give Rumble jungle the same lofty possibility that we give Wei to every team at MSI, or should teams maybe draft better?

TIP: How to deal 1035 true damage to objectives after smite changes!

Comment

Originally posted by ExceedingChunk

I would expect that true damage was a class at this point, and that things like this shouldn't be possible. But I guess the legacy code is too spaghetti to fix.

Also, it should be standard developer practice to create a unit test to check obvious stuff like this on all spells or items that increases or reduces damage. I'm really curious about how much of this is bad practice, and how much is just working on a system that was poorly coded and got too big to fix.

Bugs obviously do happen everywhere, but for things like true damage the system itself should prevent a single developer from "forgetting" or messing up.

I mean, it's very easy to just tell the ability "amplify magic and physical damage" instead of "amplify all damage."

After all, exceptions are useful. Ahri's Charm impacts the back half of Q. If true damage is hard coded to ignore all damage modifiers, you get way more issues than just making sure you build the skills or buffs correctly.

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by Easilytitled

I maybe stupid here, but I just feels like where you see the huge drop off in winrates from diamond to master because people are better at abusing inexperienced players and the margin of errors got significantly lower, which I feels like it will get better as times goes on.

Edit: I still feels like the total games played have to still be significantly lower than Rumble mid, as the pick just became recently popular in the Jungle

I have no doubt that Rumble junglers are relatively inexperienced. But unless I see an Azir curve, I'm not buying it.

It also means, regardless, that teams are still hard trolling their drafts. Even if you are in the camp of, "He's actually OP when played well, but teams aren't there yet" then why did virtually every team int all of their drafts by picking or banning him? How bad are you at understanding how your scrims went?

The end result is still the same: Draft better.

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by bigfish1992

I mean Lillia has been around 44-46% win rate jungler in D2+ as far back as u.gg allows (which is 11.6) and yet is still highly prioritized as a competitive jungler.

Trying to conflate solo queue winrate with competitive viability is not good analysis. If the argument was/is Rumble isn't good in solo queue and you could make arguments for why it's not sure that would be more reasonable as Lillia requires a good amount of coordination with her ultimate. Rumble is probably similar in his equalizer and fighting in spots where his ult is most impactful.

But Phreak wants to use this data to say it's a bad competitive pick when he also tweeted about Rumble having I think a 3-7 record after taking out RNGs "Free" wins (it's actually 4-7) without taking into account some of the free losses from teams like DWG beating Infinity on the last day. If we take the RNG wins into account Rumble is 7-7 in jungle, if we take out Group A games his record is 4-6 in jungle.

I'm c...

Read more

Except she's not "still highly prioritized as a competitive jungler."

Her p/b at MSI is 14%. Her two picks are from players eliminated from the tournament. She lost both games.

Literally no one good at League of Legends is playing Lillia today.

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by Kyriios188

IMO Phreak shouldn't answer them so directly so quickly like this.

For a part of Riot to argue out in the open directly with two other big league personnalities feels very unprofessional to me

I'm not interested in arguing specifically against Dom or LS or anyone. The first draft of this video was two days ago unscripted and it was so rambly I left I unpublished and spent a couple hours yesterday writing a script. It so happened that they put out a new video in between those points in time.

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by catseye013

I've never had issues with Phreak being wrong, the issue is him being unable to admit it. Wasn't long ago he had that whole Zeal item Caitlyn argument about how building a Zeal item on Caitlyn was objectively bad because it's not as gold efficient as full AD Crit items, and when multiple pros added how Zeal items give movement speed and lets you stack headshots faster (which can't be measured in gold effiency) he literally said "i don't care, it's a bad build" while casting.

Rumble could literally go 100% wr 100% pb for the rest of MSI and Phreak would still argue he's not a good jungler and that he's right

Stacking Headshot is easily measured in gold efficiency: Headshots scale linearly with both AD and AS. It's a non-argument. Simply scaling DPS scales Headshots. To argue that AS is somehow a better Headshot scaler should make you trust them EVEN LESS.

"I need movement speed" is the exact same argument as "I need life steal first on Kog'Maw." I'm glad you feel that way. You're still wrong.

Meanwhile, I can not only prove out the DPS-per-gold, the exact same metric I used to predict new Phantom Dancer as overpowered (spoiler: it is), but also look at solo queue win rate data to see that in real-world applications, a second Zeal is still bad.

So, every single non-subjective form of proof is on my side. And on the other is... Opinion

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by azaza34

But isnt there something to be said for Rumble not really being analogpus to other league of legebds champs in the same eay that other picks are? IE experience on other champions does not translate well to rumble. At least for mid and toplaners, they have probably laned against rumble and are more familiar with the mechanics and playstyle than junglers necessarily would.

I agree that learning Rumble is more unique than, say, learning Cassiopeia. We are in total agreement here.

So, wouldn't learning Rumble be a byproduct of playing more games? If so, then why does Rumble jungle's performance drop as he's measured in higher-playing skill brackets?

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by AxeAndRod

Sorry for wall of text.

It's clear you still don't understand Phreak's argument at all.

If we give tiers for each soloQ level's amount of coordination I think it would be made clearer. (I'm making these up to show Phreak's point, but I hope we can all agree that coordination level goes up when you go up in rank)

Iron: 0/10 Coordination

Bronze: 1/10 Coordination

Silver: 2/10 Coordination

Gold: 3/10 Coordination

Platinum: 4/10 Coordination

Diamond: 6/10 Coordination

Master/Gm: 7/10 Coordination

Challenger: 8/10 Coordination

Now, you are right that Pro play is another level on top of this, for now we can just assume that Pro play is the highest level of coordination, AKA a 10/10.

So, in the video Phreak demonstrates that Rumble's SoloQ jungle relative win-rate goes down with increasing rank. This simply correlates with the idea that as players get better, coordination ge...

Read more

FWIW, I'd argue it's more like 0 / 0.3 / 0.6 / 1 / 2 / 3 / 4 / 10.

Just, things like Taliyah ganks only require a 3-4 to be good whereas Ryze ultimates need a 6 or something.

Very much an oversimplification still, but that's the general idea.

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by topdrogon

I completely disagree with your point about context of statistics. If you provide stats and use it for an argument, the onus is on YOU to provide the CORRECT context and interpretation of the stats and when it can be applied and when it cannot be applied. The onus is NOT on people to rectify your misuse of the statistics.

Example: Hecarim had a 48-49% win rate in plat+ soloq forever, but it turns out people were building wrong/using the wrong rune. Before people figured out phase rush made him broken, phreak’s ENTIRE argument could be used to make it seem like Hecarim is not a good pick for pro play in the jungle. This was categorically false, and people could have said “yes the soloq winrate for hecarim doesn’t matter because people are playing/building/runing him incorrectly” is absolutely a sound argument. Yes, that person needs to then say “this alternative playstyle, build path, rune choice is wholly superior”.

Saying that soloq data is useful because of large...

Read more

Example: Hecarim had a 48-49% win rate in plat+ soloq forever, but it turns out people were building wrong/using the wrong rune. Before people figured out phase rush made him broken, phreak’s ENTIRE argument could be used to make it seem like Hecarim is not a good pick for pro play in the jungle. This was categorically false, and people could have said “yes the soloq winrate for hecarim doesn’t matter because people are playing/building/runing him incorrectly” is absolutely a sound argument. Yes, that person needs to then say “this alternative playstyle, build path, rune choice is wholly superior”.

This one is actually fairly easy to combat for the same reason that it's reasonable to use MMR trending toward pro: High MMR players have their ear to the ground. They pick up the new OP builds. For example, here is the purchase rate of Turbo Chemtank on Hecarim in patch 11.5, when his pro presence spiked to 97%

  • Bronze: 27.28%
  • Sil...
Read more

Rumble is not a jungler part 2 - Phreak

Comment

Originally posted by Easilytitled

I think it's normal that the winrates for Rumble jungle in higher elo is lower because the pick is newer than Rumble mid, and jungler have not grown accustomed to how he should be played; therefore got abused in higher elos

It's actually accounted for, though:

If Rumble jungle is just 5% worse than Rumble mid because players are bad at him, you'd see a flat win rate difference, but you don't. The difference grows with player skill, and it's bad for Rumble jungle.

See, players in higher MMR play more games. So higher-MMR players have more practice on a given pick. That means you would expect an uptick in new champ performance at higher ratings. But you don't.

Yes, it's imperfect. We're using MMR as a dummy variable to represent practice games. I'm aware that it's flawed. But you'd still intuitively expect that relationship to hold. Instead, we get the opposite.

For whatever reason, as players play more games and play higher-quality League of Legends, Rumble jungle is even worse of a pick and Rumble mid is even better.

You CAN Use Solo Queue Win Rates in Pro Play - Phreak

Comment

Originally posted by Wickd

Decided to fire this up on Reddit because the conversation is important

Thanks Wickd, you're a real one.

TIP: How to deal 1035 true damage to objectives after smite changes!

Comment

Originally posted by true_eudaimonia

If true damage isn't reduced by resistances, why is it increased by things like Abyssal Mask, Prowlers, etc.? Seems like it's only situationally "true"

On a technical level, it's just how the items/debuffs are coded.

I don't think it was really intentional; more of an oversight. I could be wrong.


11 May

Cloud9 vs. DetonatioN FocusMe / MSI 2021 - Group C / Post-Match Discussion

Comment

Originally posted by sameo15

The world was not ready for that pun.

I was honestly preparing to head into game 5 saying, "Alright, Kobe..." pronouncing it like the beef, but I forgot.


07 May

Riot need to decide whether true damage is affected by damage modifiers or not.

Comment

Originally posted by LittleBadWitch

Dashes and blinks are extremely consistent on this front.

They're anything but consistent in how they interact. Here is a list of dashes (as described in the ability) that are stopped by roots & snares:

Read more

FWIW Aurelion Sol and Taliyah are simple "they fall off in combat" and the root/stun interaction is superfluous.

Still though, valid point. I didn't know Akali/Camille/Rakan/Ziggs broke their dashes.


06 May

Riot need to decide whether true damage is affected by damage modifiers or not.

Comment

Originally posted by buwlerman

And what happens when a dash that knock-backs, stuns or knock-ups, meets another? Does the Poppy or the Camille get CC'd? Both?

Last time I checked this was very inconsistent.

There's also the weird interaction with Jarvan being able to use E-Q to dash while rooted, while champions with unstoppable abilities (Malphite, Vi) are not.

There are lots of these kinds of small inconsistencies in the game, and they add up in a way that makes the game harder to learn than it has to be. It's good that champions are unique, but sometimes it adds little to the character, and even when it does it should still be communicated clearly.

FWIW it's consistent: Abilities have hitboxes. Better hitbox wins. It's like fighting games: Something usually wins. Some things draw. But if you know an interaction once you know it every time.

Jarvan is quite special since his Q is just a normal ability. I replied to someone else about Caitlyn and how her dash is uniquely coded to not move if she's CCd when it's finished casting. Jarvan just isn't locked out of his Q while grounded/rooted, which certainly makes sense, but also makes him uniquely able to late-cast and get out. Definitely an exception, but EQ is an exceptional pair of abilities.