Three Hundred Thousand Bans, 0.2 Percent: Auditing the Ranked Ladder Where the Denominator Is Missing

There is an old habit in my notebook: when a big number arrives, I draw two empty boxes first. One for the window — how many days the figure actually covers. One for the denominator — who decided it. In 2026, sitting in a house in Boston and logging France versus Argentina's 4-3, twenty-three shots into a sixty-four-page notebook, both boxes were blank on page one. The scoreline pointed one way; my xG column said 2.7 against 1.9. The first xG notebook taught me that a match can be read twice. The number that surfaced after Riot Games brought Vanguard to League of Legends last September — roughly three hundred thousand ranked accounts actioned — is asking for that same second read.

The figure comes from Riot's own published article, which also names company spokesperson Phillip 'mirageofpenguins' Koskinas. The same piece states these accounts were actioned for rank manipulation, boosting and hitchhiking. Two titles sit inside that number at once — League of Legends and VALORANT. Both are Riot first-party games; both now sit under the same anti-cheat layer. This is not a champion or item patch story. It is a publisher-governance story.

Vanguard was built for VALORANT first, at kernel level — deep inside the operating system. Its arrival in League of Legends is a platform-level change; champions, items, maps and economy are untouched. In esports, the patch notes are the weather; the data is the climate. This is not a weather season; it is the composition of the atmosphere shifting. The meta being reshaped is not the tactical meta. It is the behavioural equilibrium of the ranked ladder: the boosting market, the smurf resale chain, and the expectations of ordinary players. We are in the regular season, with no tournament running a live signal, so the loudest data on the table this week is not a map-control table. It is a punishment table.

The vocabulary matters, because the whole argument rests on it. Boosting means paying a stronger player to log into your account and lift its rank. A hitchhiker is someone who queued on their own account alongside a boosted one and shared the result. Smurfing means using a secondary account — which Riot treats separately. Three behaviours, one door, but not the same lock.

That is where the first read and the second read split. I trust the model, but I audit the model before I trust the model. In Riot's model, three hundred thousand is cumulative — the total since September 2026. The monthly rate, the enforcement cadence, appears nowhere. And the 0.2 percent sitting beside it? Its denominator, the size of the player base, is not a Riot server log. It is an outside estimate.

This is where my notebook gets a red mark. In 2026, analysing all 83 Bundesliga matches after the restart, I learned that the easiest way to make a rate look large or small is to move the denominator. Home points fell from 1.54 to 1.32 per match, home win rate from 43.2 percent to 33.7 percent; without controlling team quality with a five-match rolling xG, those numbers could have told any story at all. Empty stadiums were a natural experiment; I just brought the spreadsheet. The same rule holds here. Zero point two percent means cheating is rare, or it means the lens is narrow. Telling them apart requires the denominator, and the denominator is currently pending verification.

The real signal is not the count but the structure of punishment. Riot says action against accounts tied to boosting will not stop at that account — it can extend across multiple accounts, including suspension of a main. That sentence cannot be written without a pre-existing account-linking capability — device fingerprinting, IP patterns, behavioural signals. The TPM 2.0 and hardware-verification plan is therefore not a new idea. It is an upgrade of existing infrastructure.

The second signal is escalating verification, and it is likely rank-differentiated: harder conditions at high elo, lighter ones below. Commercially that is coherent, because boosting is priced highest at the top of the ladder. The plan lists multi-factor authentication, TPM 2.0 chips and hardware-based verification. Two goals: make throwaway accounts hard to create, and close the route of returning on a fresh account after detection.

There is a natural experiment buried here that nobody has run. Vanguard has been inside VALORANT since 2026; it reached League of Legends in 2026. Same publisher, same anti-cheat layer, two titles — five years of experience on one side, a few months on the other. The comparison those two titles would produce side by side is exactly the split Riot has not published. Three hundred thousand spans both games with no breakdown. Which title carries how much cannot be answered, and that gap says more about the model than the headline does.

One more thing is missing from this data set: appeals. Players who queued on their own accounts alongside boosted ones have had ranked points revoked. How open the appeal door is for them, on what criteria an appeal succeeds, and how often decisions are reversed — none of it is public. Where the power to punish and the power to write the appeal rules sit in the same institution, transparency is the only counterweight. It is currently unwritten.

Which brings the contrarian question: is this three hundred thousand a clean-up, or a classification problem? Riot must separate punishable boosting from protected smurfing. Riot states smurfing can be legitimate — practicing champions, protecting a main's rank, playing with friends, privacy, one-champion accounts. This is not a black-and-white rule; it is a classification problem. Scale a small misclassification by three hundred thousand and community anger scales with it. The crowd was the variable we never put in the model — and it is absent from Riot's model too, which contains only a list of accounts.

The Morocco logic can be pulled in here as well. At Qatar 2026 I coded Morocco's 4-1-4-1 — PPDA of 14.2, xG allowed of 0.78 per match, a single own goal across their first five games. They won by refusing the expected tempo and holding patience with few resources. The boosting market is the inverted version, but the structure rhymes: thin capital, patience, probing for gaps in the defence. When verification tightens, that market does not accelerate; it swaps resources — older identities, rented identities, shared identities instead of new accounts. That is inference, not evidence, and the evidence is still pending.

The third counter-angle is technical and least discussed. TPM 2.0 is a hardware chip. Older laptops, older motherboards, and lower-income regions where upgrades are not trivial: this condition can bar the door not only to cheaters but to ordinary players. Rank-differentiated rollout softens the damage, but it is no solution for lower-income regions. The published article does not address this risk at all.

And one question stays open. Is this enforcement genuinely narrowing the road, or widening the race? The article itself concedes that blocking the route back after detection is a goal. The crackdown runs; the arms race runs with it. In that race, what counts as victory — cumulative bans, or the rate at which new accounts are created? That answer is not written yet.

The economics I know from an older scar. In 2026, consulting for the New England Revolution during the transfer window, I flagged Georges Mikautadze after the Euros: three goals, 0.68 xG per 90, 2.1 progressive carries per match. The deal collapsed at the medical when a prior knee issue surfaced. I had modelled output, not injury history. A transfer rumour is a hypothesis; a medical and a spreadsheet are evidence. Riot's figure stands in the same place. Three hundred thousand actions are evidence; the size, revenue and regional distribution of the boosting market are not published. A market nobody measures is a market whose win-loss ledger nobody can write.

One layer of the ranked ladder tends to vanish from this conversation. A large share of academy and tier-two scouting still runs on the ranked leaderboard. If the ladder fills with boosted accounts, the scouting signal itself is contaminated. Riot's enforcement therefore protects the pipeline indirectly — though tighter verification will add friction for amateur players who practice on alt accounts. That two-way effect is absent from the article.

Over the coming weeks I will track three things. First, the formal rollout dates and region-specific conditions for TPM 2.0 and MFA — if any region makes it mandatory, watch where the community reaction lands. Second, publication of the appeal policy, especially the reversal rate. Third, the next published ban figure — because that is the only way to tell whether Vanguard is an administrative instrument or merely a season of bans.

In my notebook, three hundred thousand is a big number today. Without a denominator it is exactly as incomplete as France's 2.7 xG was without Argentina's 1.9. Nobody is reading the ranked ladder twice right now. When someone does, the first question will be an ordinary one: the fraction sitting below zero point two percent — whose is it?