296,416 Accounts and Four Penalty Tiers: How Riot Games Operates Its Anti-Boost Machine

In a VALORANT ranked match I watched through a friend's spectator mode late one night, an account climbed from Silver to Diamond in seventy-two hours. Not through a win streak, not through a stable headshot rate. The anomaly lay elsewhere: the person behind the keyboard made reads a Silver player simply cannot make — reading situations two beats ahead, pre-aiming corners before opponents appeared, choosing plant positions as if they had memorized every map across thousands of hours. I did not need to watch the whole match to recognize it as boosting. What I needed, and what most ranked players have been waiting for, is the number Riot Games just published: 296,416 accounts with rank manipulation behavior in VALORANT and League of Legends have been actioned by the Anti-Boost system.

That figure did not arrive in a grand press release. It arrived inside an explanation of the mechanism, alongside a four-tier penalty diagram and a list of behaviors classified as violations. To me, this is a more worthwhile read than any transfer headline. Because every day, millions of players enter the ranked queue on a silent assumption: that the rank they are climbing reflects their true skill. Boosting breaks that assumption. It does not shake a major tournament, but it shakes the thing beneath every major tournament: the belief that the ranked system means something.

Data needs no loudspeaker, but it shakes an empire.

I have followed esports since 2026, when I was a player and tournament organizer before moving into media. Over more than two decades, I learned one thing about how publishers handle problems: they publish a figure when the figure is large enough to prove they are working, and they stay silent when the figure shows the problem has outgrown their control. Riot's decision to publish 296,416 accounts is a two-way signal. On one hand, it says the system runs and catches people. On the other, it concedes that the scale of the boosting economy is large enough that the figure is no longer an exception.

In this article I am not retelling Riot's statement. I am dissecting the structure of the Anti-Boost system, pointing out where it is strong, where it leaks, and offering a verifiable prediction about where this fight goes in the next twelve months. This is a topic most of the community skims because it sounds administrative. But it is precisely in administrative topics that people see most clearly where real power sits.

What boosting is and why it deserves a dedicated system

Boosting is the act of a high-skill player logging into another person's account to play ranked matches on their behalf, helping the account owner climb to a higher tier without competing themselves. The performer is called a booster. The account owner pays, or trades some benefit, for a rank that is not theirs.

What makes boosting different from ordinary cheating is that it uses no illicit software. No firewall is breached, no malicious code runs in the background. The booster simply plays better than the account deserves. Detecting boosting therefore requires a different kind of intelligence: behavioral intelligence, not technical intelligence.

Riot Games, as publisher of both VALORANT and League of Legends, built Anti-Boost as a defensive layer operating at the behavioral level. It monitors unusual patterns in how an account plays — sudden shifts in skill, matches showing signs of being dictated by a player inconsistent with the account's history, pairs or groups that frequently appear together with overly favorable outcomes.

There is a legal and ethical boundary Riot draws very clearly, and this is the point I want to stress because it is often misread. Riot states that secondary accounts created and operated by the players themselves are normal activity. Having two or three accounts to play with friends at a lower tier is not considered a violation. Anti-Boost targets the intent to manipulate rank, not the existence of alt accounts. This is a notably narrow, intent-based standard, and it differs significantly from how many other publishers handle the same issue.

That difference matters because it sets the entire operating philosophy of the system. If you ban alt accounts, you hit a large population of legitimate players and trigger a backlash. If you target only manipulative intent, you keep legitimate players but you must accept that determining intent is far harder than determining behavior. This is the core trade-off Riot chose.

Four penalty tiers: the structure of a punishment machine

According to what Riot published, the Anti-Boost system operates on an escalating penalty ladder, tiered by severity. This is the part I consider most analytically valuable in everything the publisher disclosed, because it shows how they classify the danger level of each behavior.

Tier one applies to first-time detected manipulation. When the system determines an account has earned ranked points or rewards through manipulation, it cancels all points and rewards derived from that cheating, returns the account to its original rank, and applies a temporary suspension. Three actions happen at once: rollback, restoration, and short-term punishment.

Tier two applies to repeat offenses. Ban duration increases with each repeat violation. The very existence of this tier is a notable signal: if the recidivism rate were zero or negligible, Riot would not need an escalation mechanism. Designing one means they anticipate, or have observed, a non-trivial recidivism rate.

Tier three applies to two more serious behaviors: buying, selling or transferring accounts, and intentional deranking. For these, the penalty can be a permanent ban. This is where I want to pause and analyze closely, because it reveals the economic logic behind Riot's decision.

Account trading is behavior with a clear commercial motive. Intentional deranking is too: it is often done to create a low-tier, high-skill account for further boosting or resale. Both behaviors sit in the deepest layer of the boosting economy, the directly profitable layer. Riot reserves its harshest penalty for the profitable layer, not the single execution layer. An individual booster may only be suspended. Someone running an account-trading chain faces a permanent sentence. This is the logic of a regulator targeting supply rather than only demand.

Tier four, and this is the most contentious part, extends liability to related parties. The booster's main account and teammates who frequently queue with them may also be actioned. This is called joint liability. It turns Anti-Boost from an individual punishment system into a network punishment system.

By design, this makes sense. If you want to break a boosting network, you cannot only stop the booster. You need to make queuing with them risky, raising the social cost of complicity or silence. But in enforcement, this is the system's highest-risk zone, and I will return to it in the counterpoint section.

How it works: prevention or remediation

A structural point most readers skip: Anti-Boost is a remediation system with rollback, not a purely preventive system. This shows in the fact that the central penalty of tier one is cancelling earned points and rewards, plus restoring the original rank. Both actions occur after manipulation has happened.

That means a lag exists between the moment of manipulation and the moment of remediation. During that lag, other players had to compete against a boosted account. They lost, dropped points, endured an unfair match. Restoring points for the boosted account does not restore the experience for those who met them.

Riot adds a match-level layer based on boosting signals, which shows awareness of the limits of the remediation model. But per their own description, match-level detection is still maturing, based on signs rather than direct proof. This is an important concession: the current system is not mature enough to prevent in advance, and is compensating with post-hoc detection speed.

Operationally, this is a reasonable choice. Absolute prevention in an open system like a global ranked queue is impossible. But it raises a question Riot has not answered: is the lag between manipulation and remediation narrowing over time, or widening as boosting methods grow more sophisticated.

The scale of the boosting economy

The figure of 296,416 accounts is Riot's own, pooled across VALORANT and League of Legends, not split by region and not split by title. This is a data weakness, and I want to show why.

VALORANT is a tactical shooter. League of Legends is a multiplayer online battle arena. Climbing pressure, queue structure, and boosting-market dynamics differ fundamentally between the two titles. Pooling them into one figure makes the announcement sound more impressive, but it hides distinct dynamics that should be analyzed independently.

If I want to know whether boosting in VALORANT is rising faster than in League of Legends, the pooled figure cannot answer. If I want to know whether the East Asian market has a higher violation rate than the Western market, the pooled figure cannot answer either. And most importantly, if I want to know whether the situation is getting worse or better, I need a comparable figure from a prior period. Riot provides only a cumulative figure.

A cumulative figure tells us the scale of the problem across the entire history of the system. It does not tell us about the trend. This is a basic distinction in data analysis that is sometimes overlooked in community discussions: a total is not a trend.

When Riot says it will expand Anti-Boost going forward, that is a statement about direction, not about results. And when their document expresses the expectation that these measures will help the ranked environment become fairer, that is a forward-looking expectation, correctly recorded as what it is, not a proven achievement.

Counterpoint: who the system protects and who it forgets

Most Anti-Boost content the community receives revolves around one message: Riot is tightening. But there are three areas where I believe the system is creating new risks while solving old ones, and they are rarely discussed.

First, the joint-liability clause for players who frequently queue with a booster. This is the highest false-positive risk zone of the entire system. Imagine an ordinary player who queues every night with a friend. That friend, for whatever reason, is also boosting another account. The ordinary player knows nothing. They keep queuing, keep winning, keep climbing. One day, the system actions the friend and sweeps up the ordinary player too, because they fit the frequent-pairing pattern.

Per Riot's own description, no independent appeal mechanism is mentioned, and no tolerance threshold is published. That means an accidentally swept player has no clear path to prove they did not know. In a system that publishes no threshold, clearing your name becomes harder than being convicted.

Second, the intent-based standard. Targeting manipulative intent rather than the existence of alt accounts is correct in principle, but far harder to enforce transparently than a bright-line rule. When the standard is behavior, players know exactly where they stand. When the standard is intent, players only know they might be in a gray zone without realizing it. And when the standard is intent, the likelihood of community-perceived inconsistency rises.

Third, data independence. 296,416 is self-reported by Riot, with no independent audit. This does not mean the figure is wrong. It means we are evaluating the effectiveness of a system based on the report of the party operating that system. In any other field, people would demand a third source confirming the figure before treating it as evidence of effectiveness.

I am not against tradition, I am only giving tradition a new piece of evidence.

My point is not that Riot is hiding something. My point is that the system's structure creates three gray zones, and in all three, ordinary players bear the risk before the operator does. This is the nature of power relations in any governance system: the party that controls detection and adjudication is also the party that publishes the results.

The race between detection and adaptation

One of the points I consider most important in Riot's document is their admission that detection methods must keep improving. This is an indirect concession that current methods are imperfect, and that boosters are adapting faster than the system matures.

This is a classic arms-race model. Each time the system detects a new method, the operators of the boosting economy shift to another. They move from individual boosting to coordinated groups. They move from in-game communication to external channels. They move from direct play to organized deranking that creates low-tier, high-skill accounts.

Riot has an advantage in this race: they control data at the deepest layer. Every match, every behavior, every interaction passes through their servers. No platform has an equivalent information advantage. But information advantage does not automatically become enforcement advantage. It only does when the publisher has enough resources and enough will to keep updating detection models.

And here is where I offer a predictive observation. When a detection system grows stronger, it does not only catch more violators. It pushes uncaught violators toward harder-to-detect channels. The boosting economy will not disappear. It will relocate. The question for Riot is not whether they can erase boosting, but whether they can make the cost of participating in that economy exceed the benefit it delivers.

Anti-Boost as an investment in trust

At a macro level, I read Anti-Boost not as a punishment measure but as an investment in trust. And I believe this is the most important reading.

Both VALORANT and League of Legends operate on a model where the ranked queue is the foundation of the entire ecosystem. The best players come from the ranked queue. The next professional players are discovered there. Academies recruit from there. Major tournaments are built on that foundation. If the ranked queue loses meaning, the entire value chain sitting on it loses its base.

A rank tier has no intrinsic value. It has value only because everyone believes it reflects skill. When that belief erodes, the value of the rank collapses, and with it the value of the entire talent-discovery chain built on rank.

Algorithms do not tire, but fan hearts do.

This means a clean ranked queue is a necessary input for talent-discovery pipelines. If boosted accounts occupy high positions on the leaderboard, the signal scouts read becomes noisy. A high-tier account could be a genuinely talented player, or could be an account pushed up by someone else. That uncertainty devalues every signal from the ranked queue, including the accurate ones.

Riot's document does not speak directly to the link between Anti-Boost and talent-discovery pipelines. But it is a reasonable inference from ecosystem structure. And if true, it elevates Anti-Boost from a player-experience issue to an industrial-infrastructure issue.

The gray economy: where boosting meets betting and account trading

Boosting does not exist in isolation. It is a link in a chain of gray economic activities, including account trading, account transfers, and in some cases activities tied to betting.

Riot reserving its harshest penalty for account buying, selling and transferring shows they correctly identify the profitable layer of the chain. An individual booster earns a small amount from one client. An operator of an account-trading market earns from hundreds of transactions. If you want to weaken the boosting economy, you must target the layer with the highest leverage, not the layer with the most participants.

This also explains why permanent bans are reserved for tier three. It is a choice based on economic logic: raising the expected cost of the highest-profit behavior high enough to eliminate the incentive. If someone can profit from selling accounts while only facing a temporary suspension, the system creates insufficient deterrence. A permanent ban changes that equation.

But one thing must be said plainly: Riot's document provides no data on recidivism rates or on the real-world impact of penalties on market size. We know what the system does. We do not know how effective it is at shrinking the market.

Media narrative and the expectation gap

The current media narrative around Anti-Boost is a story about a publisher tightening. It is an easy story to tell, easy to agree with, and it repeats each time Riot announces a new enforcement wave.

But the data foundation for that story is thinner than it appears. A single cumulative figure, with no baseline for comparison, is not enough to establish a trend. This is a publisher-led story, where Riot supplies the data, publishes the data, and interprets the data. There is no dissenting voice in the very source this document draws from. No false-positive case is raised. No debate about the appeal mechanism. That does not mean Riot is hiding the truth. It means the story we are reading is a retelling from one side, and a serious analysis must recognize that feature.

I believe a serious false-positive case, if it happens to a well-known player or a professional, would be the real test of the system's credibility. Until that test happens, every claim about the system's fairness remains forward-looking, not verified.

What to watch over the next twelve months

From the system's structure and the data gaps Riot's document leaves, I identify five signals worth tracking.

The first is the next published figure. If Riot publishes a new cumulative number next time, we will have the first chance to analyze a trend. If the new figure still does not split by title and by region, that is a signal that data-transparency policy has not been upgraded.

The second is a false-positive debate. If the community can organize a voice around a specific wrongful case, the credibility of the intent-based standard will be tested.

The third is any information about a pairing threshold in the joint-liability clause. If Riot publishes a tolerance threshold or an appeal mechanism, the over-reach risk decreases. If they stay silent, the risk remains.

The fourth is new violation categories appearing in the next document. The number of new categories is an indirect measure of how fast boosting-economy operators adapt.

The fifth is whether other publishers publish equivalent data. If a rival title publishes a comparable figure, we can position Riot's announced scale in a broader context.

What I believe and what I cannot yet prove

Here I want to separate what the data allows me to conclude from what I can only state as a judgment with a certain confidence level. This is the discipline I keep in every analysis: if a conclusion lacks evidence, it must be labeled as inference.

What I believe with high confidence: Riot is operating an Anti-Boost system based on an intent standard, with a four-tier escalating penalty ladder, focused on the profitable layer of the boosting economy. This is a sound design in principle and shows understanding of the problem's economic structure.

What I believe with medium confidence: the joint-liability clause is the system's largest risk point, and it can generate false positives without a tolerance threshold and a clear appeal mechanism.

What I believe with low confidence: a clean ranked queue has infrastructure value for talent-discovery pipelines, and improving it delivers indirect benefits to the entire scouting ecosystem.

What I cannot yet prove: whether boosting is getting worse or better. The cumulative figure Riot published cannot answer that, and anyone claiming otherwise is speaking beyond their data.

A note on how I read this kind of news

I grew up in an environment where following sports meant following numbers. Since my days as an editor, I have learned to distinguish three types of data: data describing state, data describing trend, and data describing intent. The third is the most easily swapped in media, because it sounds certain while actually being only a statement about the intent of the party issuing it.

When Riot says it is expanding the system, that is intent data. When Riot says 296,416 accounts have been actioned, that is state data. When an article says the situation is tightening, that is an inference from the two, and that inference is only correct if a baseline for comparison is added.

This distinction matters because it determines the quality of every subsequent analysis. If you mistake intent data for state data, you build conclusions on sand. If you mistake state data for trend data, you see trends that do not exist.

In the case of Anti-Boost, the state data is the figure 296,416. The intent data is the statement about expanding the system in the future. And the trend data is missing. That is why I stay cautious about the "Riot is tightening" narrative, even though I do not deny the system is active.

Looking ahead

The fight against boosting will not end with a decisive victory. It will continue as a long-term arms race, where each side learns from the other and adjusts its methods.

What Riot can control is the learning speed of its detection system. What Riot controls less is the learning speed of boosting-economy operators, and the social cost that expanded penalties impose on ordinary players.

If there is a lesson from this kind of topic, it is this: a strong governance system is judged not only by how many violators it catches, but also by how many innocent players it protects from itself. Riot has proven it can do the first. The second is still awaiting evidence.

And there is one thing I always believe in this work: players do not need a perfect system. They need a system that is honest about its own limits. If Riot publishes data split by title, by region, and by time baseline, it will move from the position of a publisher making claims to a publisher proving them. That step matters more than any future enforcement wave, because a verifiable figure is worth more than a hundred statements of intent.

A stadium can be empty of spectators, but history is never short of recorders. And in this case, the history of competitive-game governance is being recorded one figure at a time, by the very party operating it. Our task, as readers and analysts, is to make sure that record has more than one pen.