Riot deletes Mel's reflect mechanic: when designers remove a skill instead of nerfing numbers

Minute 31 of a ranked game. Mel holds W. The enemy Seraphine fires her ultimate from the brush above, a violet ribbon sweeping across the entire formation. Mel's W goes up. The ribbon reverses. Four enemy players stand frozen in charm, and Mel's team takes Baron.

Three weeks later, another clip spreads faster. Jhin fires his fourth shot, the crit round leaves the barrel, Mel's W reflects it, and Jhin dies to his own bullet. Those are the two examples Riot Games cited in its announcement of the Mel W rework. They are also the only two.

No win-rate table. No ban rate. No frequency figure for how often that fateful reflect actually happens in an average game. I read the announcement three times looking for a single line of data, and each time I stopped at the same place: two clips, one design rationale, and a patch schedule.

In scouting meetings I have repeated one sentence for nine years: do not trust the standings, ask xG. The standings tell the past, the data tells the future. Here, Riot's balance team did precisely the opposite — they made a rework-level decision based on the memory of two beautiful moments, not on the frequency with which those moments actually occur.

That is what makes this rework one of the most interesting cases of the season. Not because it is strong or weak. Because of how it was decided.

Context: a skill stuck between two states

Mel was designed around a fairly clear idea: a mid-lane mage with the ability to deny incoming attacks. W sits at the centre of that idea. It does not block projectiles — it sends them back. That distinction matters far more than it does on a quick read of the tooltip.

A projectile-blocking ability is a defensive tool. A projectile-reflecting ability is a punishment tool. Mel players do not use W to survive a trade; they use it to turn that trade into a death sentence. Jhin fires his fourth shot and dies. Seraphine casts her ultimate and charms her own team. That is an elegant punishment mechanism, and an extremely frustrating experience.

Riot calls W Mel's "signature trait". They also state plainly that the design team tried to preserve the mechanic across multiple rounds of adjustments before giving up. That sentence deserves more attention than it usually gets.

A design team does not discard a signature mechanic after multiple tuning passes unless the mechanic is structurally untunable. Reflect is exactly that: it is binary. Either you catch the projectile or you do not. There is no such thing as a "40% reflect". You cannot reduce the multiplier of a mechanic whose outcome has only two states. To reduce its power you must shorten the reflect window — and past a certain threshold, the window becomes so short that the skill is meaningless. That is the dead end every designer of a binary mechanic knows.

The patch is numbered 26.22 and is scheduled to hit live servers roughly three weeks after the announcement. Those three weeks matter, and I will return to them at the end.

Core: five changes, one identity migration

W changes the most. The reflect mechanic is gone entirely. In its place, W destroys incoming projectiles, grants Mel a shield, and gives movement speed to nearby allies.

Read that line in isolation and many will assume it is a nerf. It is not. It is a career change.

Reflect turns W into a punishment tool. Destroy-plus-shield-plus-ally-movement-speed turns W into a peel tool. Those two functions sit at opposite ends of the champion taxonomy. Punishment tools belong to solo-lane mages — players who live by punishing mistakes. Peel tools belong to supports — players who live by protecting others. Riot did not weaken Mel. Riot moved Mel into a different niche and cut her value in the old one.

The passive points the same direction. Early-game damage and duration are down, but Mel gains health and armour scaling per level. Armour scaling on a mage passive is unusual, and it does not serve the goal of winning 1v1 trades. Armour helps you survive a dive. That is the stat line of someone standing behind the fight, not in front of it.

Q redistributes its damage structure: the front-loaded portion is trimmed, the follow-up portion is compensated. For players, that means Q is no longer a one-beat execution tool. It becomes a multi-beat pressure tool. Same ability, two different usage philosophies.

E is the change I find most elegant from a design standpoint. Root duration is reduced at close range and increased at maximum range. That is a very direct technical statement: standing close is taxed, standing far is rewarded. There is no other reading. Riot is pushing Mel to the back of the formation through the mechanic itself, not through advice.

R increases cooldown and reduces per-level scaling. Mel loses the dive-in ultimate combo, and she loses the R-initiate pattern. Her value must migrate to extended teamfights, where shield and ally movement speed are what create the differential.

Taken together, the picture is clear. Mel is being migrated from the damage-denial mage class into the utility mage class, with support as a genuine development path. Her power curve shifts from early-lane bully to late-game scaler.

There is a counter-intuitive point I want to stress, because it is routinely missed in champion-balance commentary. In professional drafts, a champion who can play two roles is worth more than a champion who is strong in one. The reason is purely structural: flex picks let a team delay revealing its role assignment until the opponent has already committed. Losing solo-lane power while gaining flex ambiguity can be a net-positive trade at the level of professional drafting.

I am not claiming Mel becomes a meta support. I am claiming that path is now open, and that it is the most plausible route if she is to keep a place in draft considerations.

The affected population is not a team, it is a mastery cohort

This is where I want to depart from the conventional read. The source article names no team, no player, no tournament. But it does contain one very specific affected group: Mel players who built their entire value on the skill of timing W.

What is that skill? It is seeing Seraphine begin her ultimate and deciding to hold W rather than spend it defensively. It is counting Jhin's fourth shot. It is reading the trajectory of an ability before it leaves the caster's hands. That is a high-level reactive skill, trainable, and measurable through reflect success rate.

Deleting reflect deletes that skill. And this is the fundamental difference between a numbers nerf and a mechanic deletion. With a numbers nerf, existing players can adapt — they keep the skill and simply have to play more precisely. With a mechanic deletion, the entire training investment evaporates. In my line of work we call that a skill-obsolescence event, and it is the class of change players react to most violently, even when the champion remains strong afterwards.

I have seen a close analogue in football, when a midfielder valued for his passing-lane reading was pushed into a different role and lost his place entirely. People called it a tactical issue. It was an occupational identity issue.

Contrarian angle: the stated goal is only half met

Riot states the goal of the rework is to limit situations that are too frustrating when playing against Mel. I fully support that goal. The question is how far it is actually achieved.

The new W still destroys incoming projectiles. Which means the "my spell did nothing" situation survives intact. What has been removed is only the "my spell killed my own team" situation.

These two frustrations are different in kind. The first is structural frustration: a player feels their effort is nullified by a single button. The second is spectacular frustration: it produces a beautiful moment, spreads quickly, and leaves a strong impression. Riot removed the second and kept the first.

That makes sense from a communications standpoint. It is questionable from a balance standpoint.

And here I have to be blunt, knowing it will draw argument, just as it did when I questioned PPDA and was attacked for three weeks before FIFA published a report confirming the point. I was attacked for doubting PPDA. FIFA confirmed it. The lesson was not "I was right" — the lesson was that when a metric or a decision is presented without underlying data, demanding that data is professional conduct, not sabotage.

Here, the underlying data is missing at the single most important point. No frequency of reflect events per game. No Mel win rate by rank bracket. No ban rate. No PBE data. The Seraphine and Jhin examples are salience-selected, not frequency-selected. A Seraphine-ultimate reflect might happen once in twenty games, but it produces a clip watched two million times. Balancing on clips is balancing on reach, and reach does not correlate linearly with impact on player experience.

Let me be clear so I am not misread: I do not oppose removing reflect. That binary mechanic really was a design dead end, and I believe the team exhausted its options. What I oppose is presenting a decision of that magnitude as self-evident.

There is a second internal contradiction worth naming. Riot says the rework will make Mel "more balanced while still preserving her unique identity". But the unique identity referenced in that same announcement was the reflect mechanic. You cannot preserve an identity you have just deleted. You can only replace it with another and hope it proves appealing. The wording obscures that substitution, and the obscuring may cause players to misjudge the scale of the change before the patch ships.

A hidden consequence: the champions nobody named as winners

This rework has a group of winners listed in no announcement: champions who rely on projectiles and were previously punished by their own damage. Jhin with his fourth shot. Seraphine with her ultimate. Long-range artillery mages. Marksmen with special attack patterns.

None of them received a single stat buff. They simply lost a counter. In transfer analysis I always separate "buffed" from "benefited". Buffed means receiving added value. Benefited means losing the thing that used to kill you. The second category is harder to measure and is systematically undervalued by the market.

One more note on scope. Live-server balance changes are region-neutral by construction, because every region receives patch 26.22 simultaneously. Regional divergence comes only from adoption speed and stylistic preference. A region with a strong mid-lane control-mage tradition will absorb a utility-leaning Mel differently from a region that favours melee divers — the group that benefits most directly from reflect's removal. I offer this as a structural hypothesis, not a conclusion drawn from the source.

And I have to add one methodological note, because this is a trap I once set for myself. When moving from football to esports, it is very easy to import the entire metric set wholesale. xG does not exist in League of Legends. PPDA does not exist in League of Legends. But the logic behind them does: measure the quality of a chance rather than the outcome, measure pressure rather than possession. In League, the equivalent is damage per gold unit, vision control per minute, and kill participation by role. A pressure metric sounds frightening, but a team that runs out of gas at minute 75 is what is truly frightening. That principle holds in both disciplines, provided you localise it properly.

Why timing matters so much

Patch 26.22 lands in the late-season window, most likely after the World Championship patch lock and before next spring. If so, the competitive shock of a rework-level change is deferred almost entirely. Teams get a full preseason to rebuild their draft charts.

This is the point I want to underline: across years of tracking major changes, my data shows that mid-season changes cause far more disruption than preseason changes, because teams have no time to accumulate matchup data. People call that a natural experiment. I call it an opportunity to measure luck. A team that wins after a mid-season change is usually just the team that adapted faster over two weeks, not the team that understood the meta better.

Riot chose to ship the rework at the end of the patch cycle. That is itself a signal: internally, Riot judges Mel to be a live-server experience problem, not a professional-integrity problem. If they judged it professional, the patch would be pushed into a competitive window.

Signals to watch in the next cycle

I will not conclude whether this rework succeeds or fails. Nobody can, because not a single line of live-server data exists yet.

But I will state exactly what I will track, and how.

First, Mel's overall win rate is the most useless metric in this case. It blends two roles with different power curves into one figure. It must be split by role: mid-lane win rate and support win rate.

Second, the metric that actually matters is role share — the percentage of Mel games played in the support position. If that figure crosses one third, the rework has succeeded at identity migration, regardless of the overall win rate.

Third, I will wait for Mel's first professional appearance and see which role it comes from. My prediction: support, not mid. If I am wrong, I will rewrite this piece.

Fourth, and this is the most important over the long run: whether Riot publishes reflect-frequency data after the rework ships. If they do, it means they had collected it all along and simply chose not to disclose it. If they do not, it means the decision genuinely came from two clips.

Those two possibilities lead to very different conclusions about how one of the world's largest game studios makes design decisions.

A transfer fee is the number one party is willing to pay. True value is the number data does not have to negotiate. That holds for players, for champions, and for decisions made in a closed room in Los Angeles.

Mel now stands in an odd place: nobody knows yet whether she is stronger or weaker, but she is certainly a different champion. And if you are someone who spent two hundred games learning to count Jhin's fourth shot, the question facing you is not whether Mel is still strong. It is whether you still want to play her.