/dev: End of Season Updates

Welcome back for a long overdue dive into the state of Competitive Gameplay on League. In this article we’ll be reviewing how well our changes have been performing from the 1st half of the year and what big changes are coming in preseason. We’ll also briefly touch on Clash and the future and wrap everything up with a sneak peek at Victorious.

Since it’s been a while, let’s quickly review our goals for the season and some of the core outcomes we expect.

Goals and Changes

This year we wanted to get into the core of our competitive systems and clean up some problems that have been painful for quite a while. Specifically, we’ve been working towards the following:

• Improve queue matchmaking quality without compromising queue time and availability.
  • Shipped: Autofill Parity - Reduced autofill disparity to <5%
  • Shipped: Premade disparity down from 54% to <5%
  • Shipped: New Matchmaking methodology—faster and more accurate identification of skill level in normal queues (coming to ranked in preseason)
  • Shipped: MMR Clamps quality of life change so players at the top and bottom .01% of the ladder aren’t perpetually stuck trying to find a match
• Improve transparency around ranked and matchmaking systems.
  • In Progress: Ranked informed matchmaking that aims to match only with those who are of a similar rank as you are
• Improve progression satisfaction and skill expression in our systems.
  • In Progress: Inter-divisional promotion removal
  • In Progress: Duo changes to Master +
  • In Progress: Structured Unlock of Master + tiers
• Make rewards more recognizable and relevant for time spent in League.
  • In Progress: Updates to Victorious thematic and content
• Players can play with, and find, others they want to play with.
  • Shipped: Flex restriction removal resulting in an increase the number of 5-stack teams by ~7% and total available flex matches between ~35% and 65% depending on region
  • Shipped: v1 of Clash Build-a-Team, allowing solo players to list themselves as free agents who can be recruited by friends and friends-of-friends

We’ve talked about improvements to matchmaking quality and ability to play with others more readily through autofill parity, premade parity, new matchmaking methodology, and flex restriction removals.

We’ve also been deep diving into Behavioral Systems, breaking those problem spaces out into a separate team that is focused completely on fundamental changes to how our systems work in the space.

Let's talk about preseason, where we’re shoring up the final Matchmaking changes for the year, diving into progression improvements, and revisiting rewards.

Coming in Preseason

We’re rapidly approaching the time of year when change blows heavily on the wind. Worlds looms large, season end is on the horizon, and the ever-looming preseason overhaul is coming in fast. Here are our core commitments over the next few months:

Division Promo Series Removal

Inter-division promotion games were initially implemented to provide compelling goals for players to tackle between tier. At best, they were a source of relief when you won and, at worst a source of major frustration when you were bounced out for the fifth time in a row. We wanted to revisit the original decision and solve for unnecessary frustration and climbing blocks between solo and flex queues divisions. As such, at the start of preseason, we’ll be removing the inter-division promotion series in both queues and making the following changes to avoid ranked inflation and rapid rubberbanding.

• On division promotion, rolling leftover LP into the next division
• Removing inter-division demotion protection for players who are at 0 LP

As this change is directly targeted at making the experience better for players at the divisional level, it will not extend to tier promotions (Silver to Gold, Gold to Platinum, etc.). We still believe that between tier promotion series are an important part of keeping the progression between tiers a meaningful and challenging endeavor. We’ll be really looking to you for feedback on the new ranked ladder experience while we monitor queue and ladder health independently.

Ranked Informed Matchmaking

Two of our core goals this year center around matchmaking quality and transparency of the ranked system. Though still primarily determined via MMR, we're layering on new guardrails that we’re calling “ranked informed matchmaking”. These guardrails will ensure teams are more balanced from an LP and division perspective in addition to being balanced from an MMR perspective.

We measure the success of this change by the difference in LP within your team and between the two teams. Our preliminary simulations are already showing a significant improvement in the following areas:

• No change to MMR difference between teams
• Within-team LP differential cut by 40% at all matchmaking percentiles
• Between-team LP differential cut by 40% at all matchmaking percentiles
• No more than +1:30 to queue times for .01% of players
• No more than +5 second to queue times for 50% of players

What this means on your side is that you should see very, very few games where any players in your lobby are more than a tier apart. This change, paired with the other matchmaking changes below, will bring match quality and visible rank closer together. We’re excited about these first tests and will be continuing with more tweaks with a plan to release everywhere before the first preseason patch.

Ranked Seeding

We spoke earlier this year about our Ranked Seeding explorations, highlighting the value associated with placing players more accurately when they are first placed in ranked queues: A less volatile experience for both new accounts entering ranked and veteran accounts when paired with these new ranked accounts.

During our initial exploration, we realized that our methodology was a bit too simplistic. While our first versions did show improvement to new player placement, we were still off against our accuracy target—these accounts were winning or losing more games than predicted after initial seeding, which signaled that they were drastically misplaced. Those tests took place shortly after our previous blog about three months ago.

Most of these misplaced accounts are what we identify as “Smurf accounts,” and they cover a large spectrum of skill levels. This makes it difficult to really pinpoint where these accounts should be placed and requires a deeper, more complex analysis. In order to really solve the problem the best way, and not just put a bandage over the symptoms, we decided to end the first Ranked Seeding experiment early and spend a few more months investing in a better version. We plan to deliver the second iteration of Ranked Seeding alongside other matchmaking solutions at the start of preseason.

Apex Tier Changes (Master+)
Throughout the year, we’ve been reviewing the way the ladder works at the very top, and we have some concerns with consistent match quality and generally how players experience what is supposed to be a collection of the best individual players on each server. Regionally, there are vast differences with play rates and expectations for climbing. Some problems that keep coming up are integrity around duos vs non-duos, autofill by role, and disparity between ranks.

For preseason, we want to take a stab at really solidifying the statement that players at this level are the best individuals on the server, and we aim to make the playing field as balanced as possible.

Our first experimental change is to remove duo queue for all players at Master tier and above during preseason. For the majority of the ladder, we can effectively make matches to balance out premades across teams. However, duos at the top level make it more difficult to provide fair matches due to the extremely low availability of players at that skill level. We believe that these changes will improve overall match quality at the top of the ladder through less variance between team skill, reduction in autofill and secondary role positioning, and faster match times.

We understand that for a portion of Apex players, climbing with a buddy is their preferred way to play, but at the top of the ladder where folks tend to start knowing and recognizing each other across matches, there’s a big difference between getting that solo bot lane pair versus a professional bot lane. We still recognize the importance of players being able to play with friends. With lifted party restrictions and greater stability with between-team match quality, we hope to see more participation in Flex queue and Clash to satisfy these needs.

Start of Season Apex Unlock
Our second change to the Apex tier is focused on providing a meaningful experience in the form of Master + unlock progression. In previous years we would open up the Apex tier at a fixed point in the early season and everyone above a certain LP or ranking would automatically populate the positions with odd LP values. This experience was both jarring and confusing, and led to some awkward situations where players would spike to a higher ranking than they should be just by being at the cusp when the gates opened. This year, we want to take a different approach and unlock the Apex tier right at the start of the ranked season. This will allow for players to fight against each other for those first spots and only promote into the tier when they hit our Apex minimum LP requirement.

Our first values to be tested are:

• Unlocked immediately; Master

• 200 LP for Grand Master
• 500 LP for Challenger

Lastly, as discussed earlier in the year, we want to continue to take a look at ways we can improve the autofill experience without hurting your ability to actually find a match. At the top of the ladder, the most pronounced impact to games comes in the form of having an autofilled player come up against a Challenger player in their primary position. We don’t have an experiment scheduled for preseason given the other projects we've discussed so far, but wanted to let you know this is a problem space we're investigating. Ultimately, by trying to balance matchmaking filters in as many ways as we possibly can, we believe that matches right from the start will be more about who performs the best, with less emphasis on which team has the highest % chance of imbalance across the map.

We are going to be incredibly mindful of these changes and will be making changes based on player feedback and matchmaking quality metrics throughout the preseason.

End of Season

As the end of season rapidly approaches, we wanted to take a brief moment to talk about what’s coming in the form of rewards. Specifically, ranked seasons are defined by the Victorious Rewards we distribute at the end of the season. 2020 marks 10 years of ranked, and the Victorious theme hasn’t evolved as quickly as the competitive landscape. We believe there is an opportunity to provide players with a more "on-theme" in-game aesthetic that better represents what it means to endure the pain and fight for your rightful place on the ladder.

New to 2020, you'll also receive a chroma for every tier you reach above Gold to represent just how high you climbed during the season. Not only should you just feel good about reaching that first Victorious milestone, but you should also be able to represent the pinnacle of where you finished the season on the Rift. 

Clash End of Season

Adjacently, this November will mark the first end of season for Clash players. Starting this year, you'll receive Clash-specific end of season rewards and recognition for your teams' triumphs, based on Victory Points earned over the season. In this first version of Clash season end our goal is to lay the groundwork for future iterations.

We’ve outlined the following tiers and sub-tiers for rewards. Each reward milestone includes all the rewards earned in previous milestones.

• Contenders (players who participated and saw some success in Clash)
  • Contender Summoner Icon: 1 VP
  • Contender Clash Logo: 200 VP
  • Contender Clash Banner: 1000 VP
• Conquerors (players who saw some good success in multiple Clash events))
  • Conqueror Summoner Icon: 2000 VP
  • Conqueror Clash Logo: 3000 VP
  • Conqueror Clash Banner: 4000 VP
• Champions (players who were major Victory Point earners)
  • Champion Summoner Icon: 5000 VP
  • Champion Clash Logo: 6000 VP
  • Champion Clash Banner: 7000 VP

If you haven’t had a chance to participate in Clash, or aren’t at the Victory Point level you are happy with, don’t worry: You’ll have two more full weekends of Clash to earn VP with higher stakes during our Worlds Clash event. For this tournament, we’re targeting 16-team brackets with more Victory Points on the table.

Overall, in 2021 we’ll be revisiting progression in Clash and expanding on what we started this year. We really want to focus on making playing as a team a meaningful and longstanding tradition in League for years to come. Let us know how you feel about this first set of Clash end of season rewards so we can take your thoughts into consideration for future iterations!

The year has been full of challenges within and outside the boundaries of Runeterra. We hope that you join us in celebrating a tumultuous season and continue to voice your opinions and concerns. The changes we’ve made this year are a continued step forward into the best version of the most competitive game in the world and we’re committed to always strive for the best version of League. This'll be our last blog before the end of season and preseason, but we'll pick back up next year to keep you updated with how we're evolving your competitive League experiences.

As always, you can find us on the Rift, battling alongside you. Good luck and have fun.