Hey folks, Dr. LoR here with an updated meta report. Patch 2.18 came with 27 changes, including 8 nerfs that hit some of the top decks in the meta, and 19 buffs to a variety of champions and other cards.
Riot have also stated their new live design philosophy – they will be scheduling balance patches every other month, in-between expansions. What we have now will be the meta until the “major PvE expansion” hits in November, followed by the set of major balance changes in January.
- The 76k matches of Plat+ data and 52k matches of Gold data are from October 20th to 25th, courtesy of Mobalytics Premium.
- I am relying on Mobalytics’s archetype algorithm, which uses champions and regions to define archetypes. This means that, for example, Bandle Tree Noxus is spread across multiple archetypes. I therefore combine some archetypes that are largely the same across different champion combinations. This mainly affects Dragons (Shyvana/Asol w/ Zoe/Jarvan), Bandle Tree Noxus (Fizz/Poppy, Fizz/Teemo/Poppy, Fizz/Lulu/Poppy, etc.), Shellfolk P&Z (Ezreal/Vi/Viktor), Poke City (TF/Gangplank/MF), Yordle Burn (Ziggs/Poppy/Teemo/Darius), Rally Elusives (Zed/Lulu/Poppy), and Targon’s Peak (Zoe/Braum/Tryndamere + Asol) archetypes.
- Mobalytics ranks data at the start of season are imperfect and come from optical recognition for Mobalytics deck tracker users. There are many ranked games that are left uncategorized (close to half). My Plat+ dataset has roughly 60% of the number of matches that other analysts have, who use last season’s Masters to generate this season’s Plat+ users. I think this shouldn’t bias results too much, but it does mean smaller numbers. Since there’s so little data in Diamond and Masters, I’ll only present them in the aggregate.
- I use Bayesian smoothing toward a win rate of 51.5% (which is the average win rate number in this sample of data). It is a technique that allows avoiding statistical flukes for archetypes with less data. If you want to learn more about the concept, check out this explainer.
- I use Herfindahl-Hirschman Index (HHI) to calculate meta diversity. The smaller HHI is – the more diverse is the meta. Read this if you would like to know more about the method.
- I include 95% confidence intervals and margin of error when talking about win rate stats. The interval is represented by two percentages, and we can be 95% sure that the ‘true’ win rate for the deck in question lands somewhere in-between these percentages. Margin of error is a single percentage number – we can be 95% sure that the true win rate of the decks is within +/- of their margins of error.
Most Popular Archetypes
Here are the play rates for the 20 most popular decks in Patch 2.18, sorted by rank. Each of these decks has at least a 1% play rate in Plat+. All of these decks together account for approximately 55% of the meta.
- The archetypes that were close to making the list: Pirates, Nightfall, Karma Lux, Scargrounds, and Teemo Sejuani BC.
- Dragons is the biggest news of the patch, having received 4 buffs, including 2 cards that weren’t really played in the deck but are considered staples now. It roared into the meta to claim the top spot, but this enthusiasm seems to be waning among Plat+ players (see the play rate over time graph below).
- Lurk was the darling of the last major expansion (before Akshan and Viego stole the spotlight), always lurking just out of the top few decks in the meta. It’s now reemerged as a strong counter to Dragons since its small cost units get quickly too large for Dragons to eat. In the past couple days, it’s actually taken over the crown (again, see play rate over time graph).
- Bandle Tree Noxus has many flavors, but most feature 2x Fizz and 2x Poppy with up to 2 other champs. It was one of the strongest decks to receive almost no nerfs last in the patch (just Aloof Travelers). It was also ‘hidden’ by the multiregion bug on Mobalytics last season, so people are perhaps just learning about it now.
- Draven Sion was the top deck of pre-patch on both WR and PR but was hit by a nerf to Draven, somewhat of a nerf to Lost Soul, and a minor nerf in Aloof Travelers (which most people have dropped completely). Its popularity dropped substantially but the deck is actually still quite strong. I expect its PR to climb again although people may be sick of its play patterns by now.
- Yordle Burn was hit by a small nerf to Stone Stackers and massive nerf to Tenor of Terror. Players still play the former and cut the latter for Lecturing Yordle, which means the deck kept most of its strength. Its popularity only dropped a bit and I suspect players who want to climb will pick it up again.
- Darkness was also hit by the nerf to Aloof Travelers but received a huge buff to Dess & Ada, which players have adopted as a 1x or 2x. It’s still a hard deck to play though and is weak into Dragons and Bandle Tree.
- Rally Elusives, now mostly Zed Poppy, is possibly the hidden monster of the meta. Despite a partial nerf to Relentless Pursuit (which is also currently bugged to boot), it has strong matchups into the top 4 decks of the meta and is therefore crushing it on WR.
- In terms of other new meta decks, Lux has seen experimentation with Poppy, Karma, and other champs, but nothing has managed to gain popularity, yet. Targon’s Peak also gained popularity with buffs to Braum, Asol, and Faces of the Old Ones, but it still seems to be mostly a meme.
- Meta diversity is very healthy, with an HHI of 263 in Plat+, not quite as low as Patch 2.17, but still near historic lows (editor’s note: to remind the readers, in case of HHI – low means good). Yet, besides Dragons, a few Lux decks, and more Lurk, the meta feels awfully similar to the last two months. I’m hopeful that players can come up with some cool new ideas though and I’ll point some out later.
In the early days of this season, the meta has actually evolved quite a lot. The following graph plots the evolution of major decks since October 20th accounting for all of the previous season’s Masters players (which is essentially Plat+ rank for the current season), courtesy of Legna’s excellent stats site. As you can see, Dragons started off at 17% of the meta but has fallen to ~6% while Lurk started off from under 7% and has climbed to over 12%. Draven Sion, Rally Elusives, and Plunder are other major gainers.
Most Successful Archetypes
Below, you can see the win rates for the 20 highest play rate decks, again grouped by rank. I also included 95% confidence intervals (for the explanation of the concept, see Sources and Methodology section), which you can see as thin vertical lines on top of each bar.
- The decks are sorted by Plat+ WR. Just for fun, I grouped them by tier. This is purely based on WR and is not an evaluation of raw deck strength. The right pilot can probably play any of these decks to a Tier 1 win rate.
- Sadly, the best decks are almost exactly the same as before, with the one exception that Akshan Sivir Demacia is stronger than it used to be. In this case, I feel like Patch 2.18 has failed to move to needle.
- Tier 2 has Dragons and Poppy Lux as “new” decks. Below that are a mix of new and old decks, but the buffs to champions other than Asol haven’t really impacted the meta much.
- Sadly, Targon’s Peak and Predict are still very much meme tier decks.
- Balance Watch: Riot’s new live design philosophy no longer considers only the 55% WR and 15% PR as precise thresholds for balance changes, and now includes “more design philosophy stuff like is it going against intended design goals, does it create sharp matchups, etc” (source). It’s also very early in the new meta, so decks will continue to rise to counter the popular/strong decks as the meta evolves to balance itself. With new cards releasing in November and December, the next balance patch is coming in January 2022 (4 weeks after Magical Misadventures expansion).
Below is a spreadsheet of the top 45 decks in Plat+, sorted by win rate. The table is split into super-columns by rank. Under each super-column, you’ll find a sample size (match), the play rate (PR), win rate (WR), and the win rate adjusted using the Bayesian smoothing technique (bWR).
In addition, the data includes a margin of error (for example, we can be 95% sure that in Plat +, Rally Elusives have a ‘true’ win rate that lands somewhere close to 57.8%, +/- 1.6%). More data available leads to a smaller margin of error.
Most Meta-Defining Archetypes
In the chart below I’ve plotted win rates versus play rates of 25 decks in Platinum+, to demonstrate which archetypes can be seen as the most impactful in the current meta, and which decks can be labeled as ‘over-‘ or ‘underplayed’.
In the graph, you can see dots and vertical error bars (‘whiskers’) for each archetype. These ‘whiskers’ represent the 95% confidence intervals for win rates. The X-axis (play rate) is placed on a logarithmic scale to prevent too much lumping on the left.
- Most of the Tier 1 decks deserve their high playrate. Pirates is perhaps the only hidden gem among decks with reasonable PRs. Of course, these are all aggressive/midrange decks that are taken advantage of the soft early season meta as others experiment.
- Dragons and Lurk are somewhat overplayed. So are Darkness and Turbo Thralls.
- I also included 5 “bonus” decks with high WRs but low PRs (red dots on the left): MF Poppy, Vi Riven, TF GP Noxus, Ziggs Taliyah BC, and TF Swain. None of these are new decks, but they have probably deserve more exploration.
The meta is disappointingly similar to the pre-patch meta, but I hold out hope that some of the newer decks can find their footing.
Dragons have potential and some strong players have been having success with it at high levels. Lux decks tend to be harder to play as well and have different play patterns than other control decks, so it could be that players will take time to adapt to them as well.
My next full meta report will be 1 week after Patch 2.19 if there are balance changes – but more likely in Patch 2.20 after the PvE expansion. Follow me on Twitter for smaller meta updates and deck optimizations in the meantime!