On July 30 the next Ranked Season begins and with it comes some updates to the sandbox, ranking system, and Ranked Slayer. Additionally, our backend matchmaking experiments have concluded and as promised we’ve got more detailed information on what we’ve been doing.

Let’s get right into it!

BALANCE UPDATES

There are quite a few changes coming to the balance of the Sandbox and you can read about all of them HERE. For this blog, we’ve pulled out the ones that directly impact Ranked.

Bandit EVO

After observing how the Bandit Evo has been performing since the start of the HCS season and listening to the community’s conversation around the weapon and how its fast time-to-kill was impacting the game, we have slowed the rate of fire slightly. We didn’t want to make drastic changes in the middle of the HCS season but rather nudging it in the right direction.

Changes:

Disruptor

The Disruptor was having a hard time finding its place on maps as an effective pick up like the other weapons on the map, so some changes have been made to beef it up while also removing elements that players found frustrating.

MA40 AR

The “bloom” of the reticle on the Assault Rifle was causing instances where players were missing shots but seemed like they seemed like they were not registering and overall the weapon at the max bloom felt inconsistent. This has been tightened a bit to give the weapon a more consistent feel.

M41 SPNKr

The SPNKr in some instances was feeling inconsistent with regards to damage and kills. The goal here wasn't to functionally change the way the SPNKr behaved but more so fine tune some cases to be more consistent and reliable for players.

Grenades

Grenade impulse tuned to allow grenade jumping in line with previous Halo titles.

SPAWN UPDATES

There have been changes to spawn points on the majority of the Ranked Arena map suite including Aquarius, Recharge, Solitude, Live Fire, Forbidden, Argyle, Empyrean, and Streets. The goals of these changes are to make the spawn system more readable and consistent, along with reducing instances where players are spawning in overly advantageous positions.

 AQUARIUS LAYOUT

In response to feedback on the location of the Overshield on Aquarius, the Overshield and Thruster have swapped locations. Overshield will now spawn top middle, Thruster will now spawn 'P1'. With the Overshield being a highly impactful item on the map, the intention is that this central location will be more fair for both teams to contest at all points in a match and not just the beginning.

RULESET UPDATES

Last year we added a requirement for players in Ranked Arena to search solo / duo after 1600 CSR. This was done in response to feedback that we’d been receiving around players not wanting to match against Parties of 3 or 4. While we have seen an overall improvement to match balance and quality, players at 1600 have been also feeling too restricted and want to play more matches with their friends. In the data, we see that the vast majority of players in Ranked Arena search solo / duo (upwards of 95%) so while we still want to make sure those players are enjoying their matches, we’re going to increase the solo / duo restriction to 1700 CSR.

And earlier this year we began experimenting with another change with deterministic CSR. Today, players above 1800 CSR go up +7 when they win, and down -7 when they lose. This has been met with positive feedback and our data shows players above that level are playing even more now! Starting on July 30, this feature will start at 1700 CSR. As usual, we’re going to be monitoring the impact based on what players are telling us and what we are observing in the data to continue to make informed decisions.

In summary, once players hit 1700 CSR they will begin searching solo/duo, and each match will reward or takeaway 7 CSR.

BACKEND EXPERIMENT OUTCOMES

At a very high level, we matchmake based on Skill and Latency and you can think of Time to Match as the currency we spend to find a match. If you prioritize finding better matches in terms of skill and latency, it means you could be spending more time finding a match. The reason I say, “could” is because there are obviously lots of other factors like time of day, region, local vs expanded options, party size, etc.

What we observed before the experiment in the matchmaking data for ranked and what we were hearing from players was that there was room to make more closely skilled matches. Additionally, find matches with more players in my region or part of my region so that I can I have a higher quality experience. And, we also noticed that there were many times where a lower quality match was being made when the matchmaker only took a few seconds to make the match. Based on this, we decided to experiment and push the match quality to be as good as it can be, without making it so that people could not find matches.

The first experiment was to increase the weighting of the Pairwise Predictability value. Pairwise Predictability will look at how each player stacks up against every other player in the lobby individually. This not only makes it so that the teams would be better balanced against one another, but we’re prioritizing every single person in the lobby being closely matched in skill, which we believe is more important for making a quality match. After observing the impact of this change, we did see an improvement in this area, particularly around the edges of the skill band so Bronze/Silver, and Diamond/Onyx.

Chart Guide:

Once we concluded that there was positive impact in terms of fair matches, we also looked at time to match and concluded that there was no widespread perceivable negative impact in terms of how long players were taking to find matches, as well as indicating data like how many players are canceling their searches.

After this, we decided to see if we could improve the latency, so we increased the weighting there a few weeks later after the first change had settled. Again, this means we’re telling the matchmaker that it should spend a little more time trying to find a match where players searching would have lower latency. We once again observed the results but there wasn’t conclusive data, which indicates that our change was too small or there were other factors that played into making this unclear like playlist changes across the game.

We once again increased Pairwise Predictability in hopes of further increasing the quality of matches.

At this point, we had started to see more feedback from the community, particularly from players in regions with less population and at skill bands where there are fewer players. We also saw feedback from players who reported searching during what we would consider to be “off peak” times like super late at night or early in the mornings. Typically, in the afternoons/evenings after work and school is when most players tend to jump on and play. From a match quality standpoint, it also wasn’t clear that there was a meaningful positive impact which led us to conclude that we pushed this a bit too far.

Finally, with Match Composer releasing we also added two new Ranked playlists in Ranked Slayer and Ranked Doubles. I’m happy to report that we collectively saw an increase in players (44% increase in the first week) and play time across ranked on a weekly basis. At the same time, a few percentage points of the player population for these new playlists were taken from Ranked Arena, which we believe is making time to match take too long and we’ve certainly seen an uptick in anecdotal reports.

Because of this, we’ve found and deployed another weighting update to find a middle ground from where this experiment started and where it last was so that we can improve the experience overall but still ensure that time to match isn’t unbearably long.

This latest update has been live for the last few weeks, and it seems to be a good working middle ground where we have improved the match quality from where it began without harming time to match significantly . At this point, we’re calling our experiment a success and will not be making further adjustments unless there are other conditions that change and we see changes in the feedback and what we can observe in our data.

RANKED SLAYER

We’ve been very pleased to see so many players jumping into Ranked Slayer and enjoying the game. We’re committed to making updates to the playlist and working to deliver updates we go!

We've made an update to the Ranked Slayer map pool along with a few Sandbox updates.

Map pool adjustment

Cliffside

Illusion

Forest

Once again thank you to everyone playing ranked in Halo Infinite and for the feedback you’ve given and continue to give us. The Ranked Team is committed to working on this experience day in and day out to improve the experience as much as we can. Best of luck to all teams competing at the HCS Atlanta Major this weekend!

Thanks,

Tashi