The VLR rating has received an update, and here is what you need to know.
As we approach the close of two years of VCT competition since the original rating was released, a modernization is in order. The game has changed significantly over the past two years, and the rating has been exposed to far more eyes across far more games than it could have been in development.
The core mechanics of the rating described in this explainer article, are still in place. However, it was determined that the rating skewed somewhat too heavily towards passivity relative to aggression. While it is important to capture the impact of supportive players that isn't reflected in their pure kill and death numbers, this update focuses around a small rebalancing of the scales against passive play.
The original rating was a combination of five factors: a kill contribution, death contribution, assist contribution, damage contribution, and survival contribution. The kill and death contributions were by far the most impactful, and they incorporated information about things like economic advantages, trading, and weighting the player differential for each fight.
Previously the fight weighting was based on the number of players alive on each team when a fight took place. For example, a kill in a 5v5 is worth more than a kill in a 5v2. However, turning a 4v5 into a 4v4 was worth less than the maximum weight because the player differential when the fight took place was greater than zero. It can be argued that turning a one player disadvantage into an equal situation has an equivalent impact to turning an equal situation into a one player advantage on average. Rating 2.0 makes that the case.
Even though the trading logic in the original rating was modified to not count against a player if it led to team success, and top duelists still had strong average ratings compared to many other agents, aggressive players were still frequently hampered by the trading modifier. In Rating 2.0, the effect of trading on a fight has been reduced.
The original rating was balanced on numbers from mid-2021 to mid-2022, but Valorant has balance changes which can significantly affect the rating. Right after the public release of the rating, patch 5.12 threw a spanner in the works with assist tails. Essentially, KAY/0, the controllers, and certain initiators suddenly saw a massive uptick in their assist totals. KAY/0 getting assists up to three seconds after his eight second suppression has expired or controllers getting assists two seconds after someone simply touched their smoke blew up a bunch of ratings in ways that didn't reflect their actual value. The overall weight of assists in Rating 2.0 has been reduced.
The survival contribution was intended to be a metric that rewarded players for saving valuable guns into the next round. This entire component was removed from the rating to focus it on the value a player provides within each round. The original weighting of this contribution was small, so removing it has a fairly minimal effect, but it is another minor chip in the armor of passive play.
The composite rating is roughly a bell curve, and it can be scaled to whatever mean and standard deviation is convenient. Although future balance changes will likely shift the distribution of ratings slightly, Rating 2.0 is set to have an average rating of 1.0 and a standard deviation of ⅓.
There are some fringe cases, where certain ratings shift dramatically, (see this game from kaajak where he originally had a 1.03 rating but now has a 1.20), but overall, the majority of ratings will not change by more than a couple percentage points.