Mel's W Rework: Reflect Out, Shield In — and One Blank Cell in the Patch Note

Seraphine's ultimate travelled toward Mel and came back to charm Seraphine's own team. Jhin's fourth shot landed on Mel's W with crit stacked, and the same crit landed back on Jhin's health bar. For several seasons in League of Legends this was a rehearsed joke: whoever picked Mel collected the fruits of the opponent's labour without paying for them.

Riot Games has removed that W in Patch 26.22. It no longer sends projectiles back. It destroys an incoming projectile and grants Mel and nearby allies a shield and movement speed. Alongside it, the passive, Q, E and R all received a numerical pass in one bundle.

The patch note that announced all this carries not a single win rate, pick rate or ban rate. I have the full change list and the stated reason for it. I do not have the measurement. I keep the ledger open until the data confesses — and here one column is blank. The story starts from that blank cell.

Context: what actually changed

Mel is a mid-lane mage, and her W was the moment a League draft flipped. The Vietnamese-language source names the ability Phan Quang Chiet Kinh — reflection. In-game language, it is the reflect.

What changed in 26.22 needs to be written carefully, because social media has flattened it into 'Mel nerf' and the case is not that simple.

The reflect is gone entirely from W. The new ability destroys an incoming projectile and grants Mel and nearby allies a shield plus movement speed. The passive loses early damage and duration but gains health and armour scaling. Q redistributes damage away from the first explosion into subsequent hits and slows the projectile. E now roots for a shorter duration at close range and a longer duration at maximum range. R numbers were cut.

Riot has publicly acknowledged that the reflect was Mel's signature characteristic, that they tried to preserve it, failed, and removed it anyway. That sentence is the heaviest line in the entire note and I will return to it.

One necessary note on the patch number. 26.22 is as stated in the source. Given Riot's season-based scheme it implies a 2026 patch, but this report has no independent confirmation of the number, so I am logging it as a timestamped entry rather than a claim. The launch window is given as roughly three weeks after announcement. The source is in Vietnamese — a localisation of global patch notes, not a regional competitive story. No team here, no tournament, no transfer, no money.

The only honest thing to do with a note like this is build a ledger: what is verified, what is inferred, and where my confidence is thin. That habit started for me in 2026, at a 32-team CS:GO LAN in an Ameerpet cafe that had no stat desk at all. I logged 42 maps by hand — round economy, opening-duel win rate, trade-kill ratio. Organisers used the sheet to reseed the bracket. At the Ameerpet desk a blank cell was always a question, never a placeholder left empty for convenience.

This patch note has three blank cells: Mel's previous win rate, her pick rate, her ban rate. Without them I cannot say how large the rework is. I can only say which direction it points.

Core: the reflect was a claim, not a number

Reading the reflect as a damage number is the most common mistake. It was not a number. It was a claim written against someone else's damage.

Mel was not building AP to make it work. The opponent was building items, completing a crit build, setting up an ultimate. And Mel collected the yield of all that work with one button press. The reflect's strength was never denominated in Mel's numbers. It was denominated in the opponent's. That is a clause where you invest nothing and collect a return on somebody else's asset.

The direct consequence: cutting Mel's own ratios does nothing to the reflect. You can trim Q damage, shorten the passive, reduce ultimate scaling — the reflect keeps working, because its price was written in the opponent's item build. The 'continuous adjustment rounds' referenced in the source failed for exactly this reason. They were posting entries to the wrong account.

So my first conclusion, stated plainly: Riot did not remove the reflect because it was too strong. They removed it because it could not be priced. A mechanic you cannot measure in your own numbers is a mechanic you cannot balance. You can only delete it. That distinction matters, because 'it was overpowered' turns this into an ordinary nerf, and it is not an ordinary nerf.

The new W is the direct answer to that problem. Shield value sits in Mel's AP ratios. Movement-speed duration sits in her skill level. Aura radius sits in her positioning. The new W is valued in Mel's own account, not the opponent's. An unpriceable asset was swapped for a priceable one. In design language that is hygiene. In ledger language it is a revaluation.

The second change tells you the direction. Reflect was a denial tool — it stopped the opponent doing what they wanted. Shield plus MS aura is an enabling tool — it gives an ally more room to work. A champion who won by breaking the enemy's plan is becoming a champion who runs her own team's plan. That is a role-identity shift, not a cosmetic one.

I read this through my portability-pricing habit. The real question about any asset is what it is worth outside the market that made it. Mel's price used to sit in the solo-queue carry market — a pick that wins lane alone and flips games. Her price is now moving to the coordinated-play utility market. Her carry ceiling has come down; her enabling floor has gone up. Those are different markets with different prices, and anyone who bought her at the old price is holding a loss.

The third change is the cleverest and the least discussed. E now roots briefly at close range and longer at maximum range. Calling it a nerf is half true. It is a two-sided lever. A player who holds precise distance is rewarded. A player who panics, steps in and roots gets a weaker hold. This is skill indexing, not punishment. The way PPDA tells you whether a football side is actually pressing, this E change tells you whether a Mel player understands spacing or is just pressing buttons. It is a filter inside the Mel player base, and that filter will outlast any number nerf.

The fourth sits in Q. Damage moved out of the first explosion into later hits, and the projectile slowed. Guaranteed burst is down; the value of tracking and aiming is up. That is a skill-expression tax. Anyone who wanted to do everything in one window and walk away now has to hold the target.

The fifth sits in the passive and I think it is the most under-read. Early damage and duration came down, health and armour scaling went up. That is not a number fix, it is a shove at the build path. When a champion's scaling bends toward health and armour, you are quietly pushing her onto an item route. A route that is off-meta, that you probably have not fully stress-tested, and that may need a hotfix within two patches.

Now the part I distrust most, and the part the source simply does not contain. No win rate, no pick rate, no ban rate. Nothing tells us how broken Mel was last patch or how often she was banned. Riot's argument is that playing against the reflect felt bad — and that argument is entirely legitimate, because they supplied two concrete examples: Seraphine's ultimate returning as a charm, and Jhin's fourth shot returning as a crit. Both examples tell you which champion class had the problem: projectile-dependent marksmen and poke mages. An example does not produce a magnitude.

Here I keep my confidence band visible. My confidence on intent is high — this is a considered design decision, not a panic hotfix. My confidence on outcome magnitude is low, because the source supplies no instrument to measure it. That Russia summer taught me to read xG as weather, not verdict. A patch note is weather too. It can tell you the direction; it cannot tell you the temperature.

And Shanghai showed me that a clean dashboard can still lie politely. This note is clean — tidy prose, two examples, one clear rationale. Exactly the way Forsaken's demo looked clean in 2026 until somebody built the outlier chart. A clean note and a correct note are not the same document.

One more thing that is always true of rework economics: the first pass almost never lands. A reworked champion usually enjoys an elevated pick rate and sometimes an elevated win rate — the rework honeymoon — and then needs one or two hotfixes once live data arrives. Five abilities changed at once. Bringing that many changes together means treating 26.22 as final would be premature. It is a first draft, published in public.

On the tournament side the source names no event and no calendar, so I will not claim anything. I will log one timing-based inference and label it as such: if 26.22 lands in a late-season or pre-season window, competitive disruption is limited. That is inferred from patch-numbering convention, not fact. And rework-level changes are usually kept off the active tournament patch until validated, so pro-play impact may lag live-server impact by a full patch cycle.

Contrarian: 'her identity is preserved' is the weakest line

One sentence keeps appearing in the coverage: that these changes preserve Mel's identity. In my reading it is the least supported claim in the whole case.

Why? Because Riot itself wrote that the reflect was her signature characteristic. Once you concede that the thing being removed was the identity, calling the result 'preserved' is a linguistic convenience, not an analysis. You cannot remove a signature and then claim the signature is intact. Something was added — shield, movement speed, aura — but adding is changing, not preserving.

The real story here is not Mel. The real story is that a publisher removed an ability it had itself labelled an identity, after failing to preserve it. That is publisher governance in its cleanest form: rule-maker, decision-maker and loss-bearer all on the same side. Between a player's attachment and an opponent's experience, Riot chose the opponent's experience. The choice is defensible. Presenting it as identity preservation is a different act.

The second thing the coverage skips: the headline is 'Mel will be easier to fight', and that headline runs on emotion, not data. No win rate, no pick rate, no ban rate. Under those conditions the probability of overstatement is high, because no claim can be challenged by a number that does not exist. The relief narrative will comfort the general player. For Mel mains it becomes an erasure narrative. That split repeats in every rework and it will repeat here.

The third point is the one that bothers me most. Removing the reflect removed one source of frustration; that is true. But what replaced it — an aura granting shields and movement speed to nearby allies — can generate its own frustration. Anything you cannot punish becomes irritating quickly. An enabler that makes a team faster and tankier will not be easier to break than a reflect. So total frustration going down is not yet demonstrated; only its address has changed.

And the fourth, offered with care because I have a parallel and not a document. The way Riot removed a signature mechanic is the same question every game faces where the publisher is the sole decision-maker. For Indian readers this is not an idle question, because our largest audiences sit inside games where the publisher's will is final. I am not predicting that. I am saying the ledger is worth reading.

Takeaway: what to watch in the next two patches

Every transfer leaves a paper trail, and I follow the numbers first. Three numbers are coming: Mel's win rate after launch, her ban rate, and the pick rate of champions who no longer fear the reflect.

If Mel's win rate collapses post-launch, the most likely next step is a partial return of the reflect or a re-tune — Riot has a history of walking back identity removals. If Seraphine, Jhin, Ezreal and Caitlyn pick rates climb, it confirms the reflect was a counter-pick tool and its removal reprices the draft. And if the passive's health-armour scaling produces an off-meta tank build, we will know Riot shipped a change it had not fully measured.

Until any of those signals arrive, I log only this: the intent is clear, the direction is clear, the magnitude is blank. And the next time a champion's signature ability is on the board, the question will not be how strong it was. The question will be whose account its price was written in.