Patch 26.22: Riot Deleted Mel's Reflect — and Shipped the Note Without a Single Win-Rate

The most important sentence in Riot's Patch 26.22 notes is not a number.

It is an admission, placed in the middle of the W description, exactly where a developer usually explains how strong the new effect will be. It says the reflect was Mel's signature characteristic, that the team tried to preserve it, and that they could not.

What sits immediately after that sentence is the actual story: no win-rate, no pick-rate, no ban-rate. Five abilities changed, a champion's identity rewritten, and not one line of arithmetic.

I opened my spreadsheet at 1:40 a.m., because I audit for a living and read patch notes out of habit. A spreadsheet is a confession — if you read the columns in order. The first column was empty. The second held Riot's stated intent. The third held my own uncertainty, which I do not hide, because since 2026 I have run a rule that every claim carries its sample size and its limitation in the same paragraph.

In three weeks this patch goes live. Until then, what we have is a document — and I rank documents above match reports.

Context: one ability that kept a champion alive in draft

Mel's draft value sat on her W. The mechanic was blunt: catch an incoming projectile and send it back. The opponent's own damage returned to them. Riot cited two examples in the notes themselves — Seraphine's ultimate reflecting back as a charm, and Jhin's fourth shot returning as a crit (source: Riot Games Patch 26.22 notes, LMHT Vietnamese localisation).

Those two examples are not accidental. They are precisely the projectile class Mel was drafted to punish. When your own burst can come back at you, you hesitate before committing — and making an opponent hesitate is the cheapest form of power in the game.

Riot has a long record of pruning mechanics that feel unjust to face. Old Poppy, old Yorick, and a list long enough that the logic is predictable. Patch 26.22 is the next instalment, except this time the cut landed on a new champion's identity rather than an old one's.

On September 2, 2026, a memo did what nine match reports couldn't — it ended a circuit overnight. Since then I read every official document the same way: first who wrote it, then what it does not say.

Method, published: I checked three things — the direction of all five ability changes (passive, Q, W, E, R), Riot's own rationale, and the timing (~three weeks from announcement to launch). I deliberately did not do three things: I did not invent a win-rate, I did not claim a tier for Mel mains, and I made no tournament claim, because the notes name no event and no calendar.

Core: five changes, and the direction behind each

Reading a patch note as a list is easy. The hard part is knowing which way each line pushes. I am splitting all five into "what it says" and "what it means", because the great dishonesty of balance notes is that they show you numbers and hide you the direction.

One — W: from theft to gift. It now destroys the incoming projectile and grants a shield plus movement speed to Mel and nearby allies. On the surface that is one tool swapped for another. In practice it inverts the champion's job. Reflect was a defensive-aggressive weapon: it attacked while defending. A shield and a movement-speed aura is an enabling tool: it does not attack, it makes other people's attacks possible. A champion who took something from the enemy's kit now gives something to her own. The number changed, and so did the direction.

Two — passive: less early damage and duration, more health and armour scaling. This looks like the most innocent line in the note. I think it is the most structural. When a champion's scaling tilts toward health and armour, the item build tilts with it, and a changed build changes the power-spike timing — which in mid lane moves the whole draft timeline. My confidence here is medium (55%), because Riot did not say it wants a tankier build. That is my read, not their statement. What would confirm it: if Mel's item distribution goes public and health-item share crosses 20%, Riot got exactly what it designed.

Three — Q: damage moved off the first explosion into subsequent hits, with a slower projectile. This is the change the community will file under "nerf" and the one that actually raises the skill floor. Guaranteed burst falls; the value of tracking and aim rises. In data language, this is a floor change, not a ceiling change — worse for low-mechanics players, roughly the same ceiling for good ones. That distinction matters, because flattening both into one colour is the most common error in patch coverage.

Four — E: shorter root at close range, longer root at max range. Not a straight nerf; a two-sided lever. It punishes the panic engage and rewards correct spacing. Anyone reading it as pure reduction is misreading Mel, and their complaint in the next patch will point in the wrong direction.

Five — R: nerfed. Least disputed, most direct. The only line in the note that can be read without interpretation.

Add them up. Damage is down across passive, Q and R. Utility is up: shield, movement-speed aura, health and armour scaling. The net result moves Mel from burst counter-pick to sustain/enable mid. That is a role-identity shift, not a numbers pass — and role shifts always show up on the draft board, outside the champion herself.

Who gains, who pays

The gainers are the champions who used to choke on her reflect: Seraphine, Jhin, and on the same line Ezreal and Caitlyn-class poke-crit marksmen. They get breathing room, because their own damage no longer boomerangs. The irony is that the patch does not protect the champions Mel beat; it protects the champions Mel was picked against. Supply has been added to the counter-pick market.

The losers are the people who played her for exactly that reason. And here I owe one human thread, because my first drafts always read like audits and lives do not reconcile. For that Mel one-trick, this is not "one of four abilities". Her entire counter-pick value rested on one of four abilities — 25% by count, 100% by worth. That is the one figure in this story that is genuinely available to us rather than estimated. Somewhere between the GPS vest and the pivot table, the player became a pattern — and a pattern should not mean the person disappears.

A frame borrowed from football, because the metric language is the same

A team that plays a sweeper-keeper does not merely save one goal; it changes the opponent's choice of pass. The striker stops playing the ball behind the line, because he knows someone is there. Mel's reflect did that. It did not stop hits, it stopped the decision to hit. Remove it and you have not just deleted a defensive tool — you have deleted a constraint from your opponent's thinking. So my first estimate is that the projectile-class pick-rate bump will come from the constraint lifting, not from Mel being weak. Estimate only. 60%.

And one thing this debate should say out loud: modern football is erasing the touchline winger in the name of the inverted one, and calling uniformity an improvement. The problem was never that some things were different. The problem is when everything fits one mould and nobody asks who built the mould.

The contrarian angle: what the note does not contain

One: the frustration claim has no denominator. The notes say many players found facing Mel frustrating. I believe it. But frustration is not a metric. Ban rate is a metric, pick rate is a metric, and neither appears. That leaves a claim with feeling on one side and nothing on the other — the most dangerous combination, because it might be correct. A champion that beats you once and stays in memory is not automatically a statistical problem. Memory and statistics are different units.

Two: "her identity is preserved" is the weakest sentence in the note. Riot itself called reflect her signature characteristic and then removed it. Call a thing your signature and delete it, and "identity preserved" stops being an audit and becomes editorial framing. I put this at 90% confidence as overstated. The honest sentence would read: her identity is changing, and we hope the new one is accepted. Riot was clear about its intent and evasive about its outcome, and that gap is where community anger enters.

Three: rework churn. Changing five abilities at once is rare, and the historical base rate says the first pass is almost never the last. I put 65% on a follow-up tuning pass for Mel inside two patches. The reason is in the note itself — the reference to "continuous adjustment rounds" implies earlier attempts that failed to stabilise her. Removing a mechanic is usually the last card played, not the first.

Four, and my biggest audit concern: timing. Patch 26.22 lands roughly three weeks after announcement. If that window sits on top of a competitive block, we get a tournament-patch versus practice-patch gap — pros preparing a champion that will not exist on the server. I am calling that a risk, not an event, because no event, tier or calendar is named. State the uncertainty; do not fill it.

And one estimate I will label as an estimate: this rework was driven by solo-queue data, not pro play, because no team, player or tournament appears anywhere in the notes. Confidence 60%. The document that would prove it is one we will not get: champion-level ban-rate history across the last several patches. If it ever goes public, this entire argument can be rewritten with data, and I will withdraw my estimate happily.

What to watch next

Do not watch the win-rate. Watch the ban-rate. Riot's stated goal was reducing the opponent's frustration, and frustration shows up first in bans, later in picks, and last in win-rate — because win-rate passes through a learning curve, an item build and a matchup spread before it settles.

I am recording two signals publicly now so I cannot pretend later that I said something else. If ban rate falls while pick rate holds, Riot was right and Mel has landed in her new role. If pick rate collapses, the story flips from "fixed" to "gutted" inside three weeks, and that flip will be driven by sentiment rather than Riot's data. The attempt to preserve the reflect failed; Riot wrote that itself. The open question is whether the next champion built on a single mechanic gets the same treatment — and whether we will once again write "her identity is preserved" to make ourselves feel better about it.