Riot Games and Anti-Boost: 296,416 Accounts, Four Penalty Tiers, and the Data Gap

Late last year, on an evening in Incheon while I was finishing notes for a regional tournament, I reopened Riot Games' enforcement documentation for the Anti-Boost system. I follow reports like this from two sides — as someone who once ran tournaments, and as someone who works with transfer data. The first thing that struck me was not the scale of the crackdown, but its structure.

The figure 296,416 accounts for rank manipulation across VALORANT and League of Legends is a cumulative total. It comes with no denominator. It is not split by region. It is not split by title. And it is self-reported by Riot, with no independent audit.

In my line of work there is an unwritten rule: whenever a number is published without a denominator, without a baseline, and without independent verification, that number is doing something else — not reporting, but signaling.

I once thought I was reading the map of a match; it turned out I was only looking at a mirror reflecting my own fear.

Context: The boosting market is not a glitch, it's an industry

Before entering the enforcement mechanism, it is worth reconstructing the economy the system is trying to block. Boosting is not a scattered phenomenon of a few lazy individuals climbing. It is a supply chain with clear tiers.

At the top are high-skill players. Riot, in its own description, calls them by a term neutral to the point of suspicion: "high-skill players." That phrase opens a gray zone. It may include professional or semi-professional players operating boosted accounts. Riot does not confirm this. If so, this is no longer a simple story of ladder climbing, but a story about disciplinary liability for people inside the professional system. But the data does not tell me. And I have to say plainly: it cannot be verified.

In the middle are the buyers. They pay for their account to be climbed to a desired rank. Reasons vary: badges, a sense of achievement, or qualifying for higher-level play. This is the demand side of the market.

At the bottom are intermediary services — where accounts are bought, sold, and transferred. Riot defines this as "buying, selling, or transferring accounts," and it is an activity that can lead to a permanent ban.

Then there is a less-noticed but systemic behavior: intentional deranking — deliberately lowering your own rank. It sounds irrational, but it has its own economic logic. Dropping rank helps high-skill players more easily create lopsided wins, serving other purposes — including boosting other accounts or producing easy-win matches to resell.

What caught my attention in this structure is not any single behavior, but how they fit together into a chain: buyer, seller, operator, and enabler. Once the whole chain exists, punishing a single link is not enough. Riot clearly understands this — and that is why its system does not target individuals, but behavior, and does not target isolated behavior, but the entire network of related parties.

The market does not move on news. It moves on the gap between two reports.

Core: Four penalty tiers and an expanded liability model

The Anti-Boost system Riot describes is not a single stick. It is a tiered enforcement ladder. Read closely, I see four levels.

Tier one, handling of detected violations. When the system records manipulation, ranked points and rewards obtained from cheating are cancelled. The account is returned to its original rank before manipulation. Temporary suspension follows. This is the basic reactive tier, more rollback than punishment.

Tier two, repeat offenses. Ban duration increases with each offense. This is a telling point: the very existence of an escalating mechanism is itself a statement about the recidivism rate. If recidivism were negligible, nobody would design a complex escalation ladder. The structure of the punishment itself implicitly admits that blocking the first offense is not enough to prevent the next.

Tier three, commercially motivated behaviors. Account buying/selling or transfer, along with intentional deranking, can lead to a permanent ban. This is the heaviest tier, reserved for behaviors seen as having a clearly economic motive rather than merely wanting to climb.

Tier four — and this is the part that made me stop longest — joint liability. Beyond the manipulated account, the system can act against the booster's main account, and against teammates who frequently queue with that person.

Alongside the penalty ladder, Riot draws a very clear definitional line: this is the most important distinction in the entire document. Riot separates two things that public opinion often conflates.

First, alt accounts created and operated by the player themselves. Riot states plainly: this is normal activity. A player having multiple accounts, playing on all of them themselves — including smurfing — is not actioned.

Second, behavior intended to manipulate rank. This is the true target of Anti-Boost.

The crux lies in the word "intent." Riot does not ban the existence of alt accounts. Riot targets the intent to manipulate — and the standard for guilt here is behavior, not the existence of an account. This is a deliberately narrow targeting standard.

I spent time comparing this approach to how other enforcement systems operate. Most systems try to draw a bright line — either allowed or not. Riot chooses the harder path: defining based on intent. By design, this protects legitimate players. In execution, it opens a question about transparency.

To me, a perfect system does not exist — only a system that is continuously verified.

The evidence chain: what the system was designed to fight

To read this system correctly, I have to reconstruct its real objective. If Anti-Boost were merely an anti-cheat tool, it would target offending accounts. But its design shows something else. It targets the relationships between accounts.

This is the most important point in this entire analysis. A punished account is an event. A punished network of accounts is a signal. When Riot extends scope to "the booster's main account" and "teammates who frequently queue with them," it is saying it understands the nature of the phenomenon: boosting is not the act of a single individual, but a network of relationships.

But this very point is the system's greatest structural weakness.

Imagine an ordinary player, completely unaware, who happens to queue with a booster across several consecutive matches. Under Riot's model, that player can be actioned. The document describes this mechanism but states no threshold.

The phrase "frequently queue with" — how many matches counts as frequent. There is no number. No tolerance threshold. No appeals process is described.

This is a gap that is not hypothetical but structural. When a standard is defined by relationships but comes with no quantitative threshold, its application depends on the system's judgment — and that system is operated by the very party that set the standard. No independent appeals body is mentioned.

I once watched a similar mechanism misfire in a different context. Not esports, but football — with VAR decisions judged as rigid and context-blind. The lesson I drew was not about whether the rule was right or wrong, but that a rule without a tolerance threshold will produce exceptions its own authors did not foresee.

This is when I think of what I regard as the biggest blind spot in this entire strategy: the system tightens at the relationship layer, but measures at the behavior layer. And between those two layers, a gray zone always exists. That gray zone is where error lives.

Cross-check: a total is not a trend

There is a detail in Riot's presentation I want to separate out, because it is usually skimmed over.

The document says measures are being tightened. This is a claim about a trend. But the only data given is the 296,416-account figure — a cumulative total, with no baseline, no comparison to a prior period.

A total does not reveal a trend. To prove "tightening," you need at least two data points over time, under the same measurement definition. We do not have that.

I emphasize this not to diminish Riot's effort, but to keep analytical discipline. If I accept a total as evidence of a trend, I am doing what I criticize in others: reading beyond the limits of the data.

And there is another notable point. The 296,416 figure pools VALORANT and League of Legends. A tactical shooter and a MOBA have very different boosting economies. Rank-manipulation pressure in these two ecosystems is not the same. But the data does not split them. Pooling two different titles into a single figure destroys the ability to analyze per title — precisely the ability readers need most.

This is when I think of a familiar image: Germany's offside trap was not broken by speed, but by one link slower than all my predictions.

In this case, that slow link is the data layer. A system this sophisticated at enforcement, yet reporting results this crude — that I did not expect.

Counterintuitive point: an intent-based standard is harder to make transparent

Here I have to break from the conventional read.

Intuitively, an intent-based standard — targeting only manipulation, not the existence of alt accounts — sounds fairer than a hard ban. And it is indeed fairer to legitimate players.

But there is a downside public opinion often ignores. An intent-based standard is harder to enforce transparently than a bright-line standard.

Compare two ways of assigning guilt. First: an absolute ban on alt accounts. This is rigid but transparent — players know exactly where they stand. Second — the path Riot chose — distinguishing legitimate alt accounts from manipulation. This is fair to the majority, but places the entire burden of determining "intent" on an automated system. And intent is harder to measure than behavior.

The result. Such a system risks being perceived as inconsistent. Two cases identical in behavior can receive different outcomes, because the measurement signals differ. And when players do not understand why they were actioned, trust in the system — the very thing Anti-Boost is trying to protect — erodes.

This is the paradox: a system designed to protect fairness can create a sense of unfairness, simply because its mechanism is hard to explain transparently.

I have no evidence of a wave of complaints. The document reflects no community reaction. But the very absence of criticism in a document provided by a single source is itself a signal: this is a summary of the publisher's viewpoint, not balanced reporting.

And I must be honest about my own limits: there is a human question the data cannot answer. When a player is wrongly punished for happening to queue with a booster, what did they do wrong. The system does not answer that. And perhaps it was not designed to.

The race between detection and evasion

There is a dynamic dimension I always look at first in any enforcement system: the gap between the detector and the evader.

Riot says it is expanding Anti-Boost's scope, and developing match-level detection — recognizing "signs of boosting" within the structure of the match itself. This is a technical step forward. It moves from account-based detection to in-match behavioral-pattern detection.

But the very admission that this capability is being developed is an implicit confession: current methods are not enough. If current methods were sufficient, a new detection tier would not be needed.

Here I think of a principle I learned after years of working with data. Whenever a defense system relies on pattern recognition, the attacker — those with an economic motive — will adapt. This is the nature of an arms race. Each time the system recognizes a new pattern, boosting services adjust their behavior to avoid matching it.

These adjustments can be very hard to detect. For example, an organized coordinated-derank model — many accounts lowering rank to a complex rhythm — would be harder to recognize than isolated deranking. Or moving communication off-platform, making in-game signals cleaner.

The document does not address these scenarios. But in my experience, any enforcement system will face them over time. When a gray market is large enough, it produces professional evaders. And those professionals always lag by one step — but have the incentive to move faster.

This is why I assess the economic motive — the financial resources behind boosting services — as the factor determining the pace of the race. A free system must face a profitable one. In the long run, this is not a balanced match. But luckily for Riot, it controls the platform — an advantage the other side lacks.

Every transfer is a murder case. The culprit is expectation; the weapon is timing.

And in this case, I will rephrase: every enforcement system is a race. The detector holds the law; the evader holds the money. The question is not who wins, but who wins for how long.

Signals for the next cycle: what to watch

Here I leave off reading what exists and move to what is worth watching. This is how I work: not predicting, but indexing the signals to come.

First, the next enforcement disclosure. This is the most important signal, because it is the second data point — the one we lack to be able to speak of a trend. When Riot publishes a new figure, I will compare measurement definitions before comparing numbers. If definitions change, all trend conclusions must be treated with caution.

Second, an appeals mechanism and tolerance threshold for joint-liability cases. If Riot publishes a specific threshold for "frequently queue with," false-positive risk will fall. If not, that risk will accumulate with enforcement scale.

Third, a noted false-positive case. This is the most watchable signal, because it will test the correctness of the intent-based standard. Such a case — if it occurs — will put the entire expanded-liability model on public display.

Fourth, comparison with other publishers. If a rival title publishes comparable data, we can begin to place the 296,416 figure in context. Until then, it remains a number standing alone.

And fifth, how boosting services adapt. This is the hardest signal to observe, because it lives in the gray market with no public data. But it will determine the real pace of the race.

Takeaway: what the data cannot answer

Ending here, I want to return to the opening question, but on a different layer.

The question is not whether Riot is handling boosting. Clearly it is. Nor is the question whether the system is sophisticated. Clearly it is.

The real question, the one the data cannot answer, is this: can a system based on intent, operated automatically, with no clear tolerance threshold and no independent appeals body, sustain player trust in the long run.

That is not a technical question. It is a question about trust. And trust is not in the enforcement data.

There is one thing I learned from my wrong predictions. When a complex system is presented with simple numbers, the most important thing is not the number printed, but the numbers left out. Here, the numbers left out are the denominator, the baseline, the regional split — and above all, the number of wrongly punished people.

The 296,416 figure is a statement. It says Riot is active. It does not say the system is perfect.

K League 2026 taught me that: the pioneer does not fail because he looks far, but because he looks far yet counts one data column short. In this case, the undercounted data column is not a coding error. It is the number of legitimate players swept into the system because they happened to stand near a signal.

And if there is one thing I will watch in the coming months, it is that column.

The applause in an empty stand is not noise; it is a signal from a future we have not been brave enough to index.

Tightening is the easy thing to claim. Protecting the right people is the hard thing to prove. And in a system where the adjudicator is also the legislator, proving that hard thing must become the next priority — if Riot truly wants to protect what Anti-Boost is building: the belief that a rank on the ladder means something.