Riot Games' Anti-Boost: 296,416 Accounts Actioned and the Data Gap Behind It

At two in the morning Munich time, a VALORANT ranked account I was tracking through a stats tool played as if fitted with an engine. Its operator climbed from Platinum to Diamond in seven days, a 78 percent win rate, an average ACS inside the top one percent of that rank. Three weeks later, the account vanished from the public ladder. Not a permanent ban. The points were rolled back to where they started, seasonal rewards were revoked, and the registered owner received a timed suspension. I had seen this scene many times while organizing amateur tournaments before moving into data consulting for teams. What caught my attention was not the case itself but how it was handled: rollback instead of erasure, escalation instead of blanket bans. Those are the fingerprints of a system designed with data, not with reflex. Riot Games says it has actioned 296,416 accounts for rank manipulation across VALORANT and League of Legends. That number is being read as proof of a tightening crackdown. I read it as a single data point, and the gap between those two readings is the whole story. Boosting, in plain terms, is a highly skilled player logging into someone else's account to climb the ladder on the owner's behalf. It is a commercial transaction, not a prank. The buyer pays to own a rank their own skill cannot produce. Around it grows a gray economy of account brokers, rank-by-rank climbing services, and groups that deliberately lose matches to manufacture cheap accounts. For an ordinary player, boosting shows up as a very concrete experience: a teammate who plays like someone else, an opponent who plays like someone else, and a ladder that no longer reflects anyone's actual skill in the lobby. For a publisher, it is a trust problem. The ranked ladder is the core product of both titles. If it loses value, the entire chain from daily play to amateur scouting wobbles with it. Riot calls its enforcement system Anti-Boost. It does not target alt accounts in general. This is the point to read slowly: Riot states plainly that self-created, self-operated alt accounts are normal activity, and Anti-Boost targets only the intent to manipulate rank. The standard here is intent, not the existence of multiple accounts. The enforcement framework Riot describes runs on four tiers, each with its own logic. Tier one is rollback. When the system detects manipulation, all ranked points and rewards earned from that behavior are cancelled, the account is returned to its pre-intervention rank, and a temporary suspension follows. This is the most common tier and the one most often misread. It is not a heavy punishment. It is an accounting operation. Riot is writing an unrealized gain off the ladder's books. Tier two is escalation. Repeat offenses lengthen the ban. The very existence of this tier is a statement about recidivism rates. If nobody reoffended, the escalation rule would not need writing. Tier three is permanent bans, reserved for two clearly commercial behaviors: buying, selling or transferring accounts, and intentional deranking. This is the line Riot draws. Both the player who climbs for someone else and the player who buys the rank get hit. But whoever turns an account into tradeable goods gets removed from the game entirely. Tier four is joint liability. This is the most notable part. Riot states it may action both the booster's main account and teammates who frequently queue with them. Together these four tiers form a governance system graded by severity rather than a blanket ban. And that says a great deal about how Riot prices each type of behavior. Look at the economics underneath. Account trading creates the goods. Boosting creates the service. Deranking creates a cheap supply of goods. The three fit together into a complete supply chain. Riot strikes each link with a different tool: rollback to neutralize profit, escalation to raise the expected cost of reoffending, permanent bans to cut off account supply, and joint liability to break up organized operating groups. One technical detail worth noting: Anti-Boost is not a purely preventive system. It is a reactive system with rollback. That means there is always a lag between when manipulation happens and when it is neutralized. During that lag, other players have already endured distorted matches. This is a hidden cost no aggregate figure can capture. Based on my experience tracking ranked matches across multiple seasons, the signature of boosting is not in any single game. It is in the shape of the curve. A real account climbs in a sawtooth: win, lose, plateau, then spike. A boosted account climbs in a vertical line. That shape, not the score, is what detection models are aiming at. Riot also says it will expand the system, including detecting signs of boosting at the match level rather than relying only on account-level behavioral signals. That is a meaningful methodological shift: from reading the trace of an account to reading the trace of a match. The model is moving from the micro level to the macro level of the unit of analysis. The downstream effects of a clean ladder do not stop at daily play. The ladder is the input data layer for academies and amateur scouting. When rank is manipulated, the selection signal is noisy. An account that climbed to Radiant with money can take a trial slot that should have belonged to a real player. Anti-Boost, then, is not only a story about fairness inside the game. It is a story about who gets seen. This is where I have to stop and separate data from interpretation. The figure 296,416 is a cumulative total. It is given for both titles without a per-title split, without a regional split, and most importantly without a prior-period baseline. A single number does not make a trend. To claim Riot is tightening its grip, you need at least two data points in the same unit of measure. Here there is one. I am not saying Riot is wrong. I am saying the claim of tightening is the writer's inference, not the data's conclusion. That is the difference between a news item and an analysis. The second problem is the pooling. Lumping a tactical shooter together with a multiplayer online battle arena into a single figure hides two booster economies with different shapes. Rank-inflation pressure in a five-on-five shooter differs from that in a MOBA, where one individual can drag an entire match in ways that are nearly impossible in a shooter. A pooled number cannot answer which title is being manipulated more. The third is the source. 296,416 is a figure published by Riot itself, with no independent audit. This is not an accusation. It is a boundary condition for reading numbers. Every publisher reports its own enforcement data, and every self-reported figure needs to be read with context about who is doing the measuring. The fourth, and in my view the largest consequence, is the joint liability clause. The phrase about teammates who frequently queue together opens a risk zone the document does not quantify. What threshold counts as frequent? Does a duo queuing twenty matches together in a week fall inside it? If one of them engages a boosting service, does the other get actioned? The document describes no appeal mechanism. It also states no tolerance threshold. For a system defined by intent, that is a significant gap, because an intent standard is far harder to apply consistently than a hard line such as banning all alt accounts. The eye watches one match, the data watches a completely different one, and both are right. The player sees a potentially unfair punishment. Riot sees an organized operating network that needs breaking. Both pictures come from the same dataset; they differ only in where they place the assumption about intent. The other side of the equation will adapt too. When detection cost rises in one channel, money flows to harder-to-detect channels: organized deranking rings, off-platform communication, or forms of exchange that leave no account trace. Riot expands the system; the other side looks for a way around. It is a race in which both sides read the same scoreboard. There is a line I use often when reading reports like this: curses do not exist, only data we have not finished reading. The VALORANT and League of Legends ladders are not cursed. They are exploited by an economic system with clear incentives, and fought by an enforcement system with equally clear incentives. What to watch is not the next number. What to watch is the shape in which Riot publishes the next number. If next time they split the data by title and include a comparison period, we will have the one thing we currently lack entirely: a trend line. If they state a pairing threshold for the joint liability clause, we will know where the false-positive risk sits. If a loud appeal case surfaces, we will learn how much pressure an intent-based standard can take. I listen to the arena through a spreadsheet, because the roar of the crowd knows how to lie too. But spreadsheets also know how to stay silent, and the job of anyone reading data is to notice where the silence is. The war on boosting will not end with a ban. It will end, if it ends at all, when the expected cost of manipulation exceeds its expected return. Riot is pushing the cost up. The other side is looking for ways to push it down. And 296,416 is a photograph of one moment, not yet a film. The question I want to leave behind: if you had to pick a single metric to measure Anti-Boost's effectiveness over the next twelve months, would you choose the number of accounts actioned, or the recidivism rate?