Riot shaved two damage off Yunara's attacks. A third of her players left.

Near surface

Walk through it with Ping: greyscreenping.lol, the same report as a burrow you go down, one step at a time.

Bureau of Rift Statistics · Report No. 2 · The patch notes NA ranked solo queue, 9 September to 25 October 2026 · 202,725 games from 18,523 players drawn at random, Iron to Challenger · plus every Worlds from 2019 to 2026 · each number's margin of error and source is on the data page

Every two weeks

Every two weeks, Riot posts patch notes for League of Legends. If you play ranked and care about climbing, you read them. A patch note is a short letter from the designers, followed by a list of champions, the characters you pick before a game, five a side, spread across three lanes (top, mid and bottom) and the jungle between them. Each champion on the list gets one sentence on why, then the numbers, old to new. In patch 26.20, Riot wrote that Ashe is "one of the premier marksmen in pro play", and under that: attack damage per level, 3.5 ⇒ 3. That's a nerf, a change that makes a champion weaker. A buff makes one stronger. Riot often leads with who a change is for: pro play, high-level play, or all players.

Players read these notes as sailors check the wind. If the notes make a champion stronger, the first players to pick it up climb on it before everyone else catches on, and when the notes make a champion weaker, nobody wants to be the last one still playing it. Reading the notes keeps you upwind. It also means the notes determine what you play before you've played a single game on the new patch.

Crust

The Worlds patch

This October, Worlds is on: the world championship, where the best teams from every region meet. A team from the West, meaning North America or Europe, hasn't won it since 2011. Caedrel, a former pro who now runs one of the biggest League streams, spent the weeks in the Worlds lead-up asking why. In his interview series Will the West Win?, he talked with five Western players and coaches, about six hours in all, and I watched every episode. Three of them inspired this report.

Jojopyun plays mid for Movistar KOI, a European team, and a mid laner's job is to read the game a step ahead of the enemy. He described a habit he sees in players, what he calls placebo nerfs. In medicine, a placebo is a sugar pill that works because you believe in it. Jojopyun's definition runs the opposite way: Riot nerfs a champion, players drop it, and two weeks later they find the champion was fine. Caps has played mid for G2 Esports, Europe's most decorated team, since 2019, and Riot put him in its Hall of Legends this September. Caps explained where G2 finds its odd picks: in solo queue one-tricks, players who play one champion and nothing else, and in the patch notes. Spawn coaches Team Liquid and decides what his team practices, and he named Kog'Maw, a marksman built for the bottom lane, as a Western pick that might stand up to Korea and China. Between the three of them, I kept hearing the same idea: a series can turn on one pick the other team never practiced against, and the hunt for that pick starts in solo queue and the patch notes.

Watch: Jojopyun on placebo nerfs, Will the West Win? episode 3, from 52:45

Pros play every Worlds match on a single patch, and Riot named it in the notes for 26.20: this was the Worlds patch. Five champions got nerfs aimed at pro play. One of them was K'Sante, a tank with so many ways to dash and shield that players call him unkillable; I'd tested that reputation in my last report and found he dies about as often as any other tank. Riot's note said the nerf targeted his reliability as a blind pick, a champion a team can lock in first without knowing what the enemy will pick against him. At the Play-In, the first stage of Worlds, teams kept banning him and picking him, and he went 0 and 5. I thought to myself that night: "we might all be being a bit mind-controlled by the patch notes."

I set out to count. If Riot nerfs a champion, do players drop it faster than it gets worse? The search began in two places: solo queue, the ranked mode where most of us play, alone or with one friend, matched with strangers, and the Worlds stage, where Riot aimed the nerfs.

Crust

The count

I drew 18,523 players at random from Riot's ranked ladder, Iron to Challenger, and collected their ranked games from 9 September to 25 October: 202,725 games in all. I weighted the games so the sample adds up the way the real ladder plays, a pollster's methodology.

Pick rates rise and fall for reasons that have nothing to do with the notes. A big streamer picks up a champion, Worlds is on TV, the weekend brings out different players. A nerfed champion's picks falling proves nothing by itself, because the whole ladder might have shifted that week. To tell the two apart, I checked the champions Riot didn't change, over the same days. If their numbers shifted too, something besides the patch shifted everyone. If they stayed put while the nerfed champion dropped, the patch did it.

Yunara shows how that works. She's a marksman, released in July 2025. She starts weak and grows into a monster if the game runs long. In the two weeks before 26.20, she appeared in about one game in eight. In the two weeks after, one in twelve. Over those same days, the bottom-lane champions Riot left be won 50.3% of their games before the patch and 50.3% after. Nothing else in her lane moved, so the patch moved her. I ran that check for every champion Riot changed, once for how often players picked it and once for how often it won.

Mantle

Yunara, the placebo nerf

Yunara's nerf was small. Her attacks lost 2 damage per hit at every rank of her Q, her first ability, and her slow stopped stacking with other slows. Riot's reason: they called her the leading candidate for "handshake hyperscaler do nothing" (a real sentence, in real patch notes; Riot is Aware), and they didn't want that to become the Worlds meta. A hyperscaler is a champion who wins if the game drags on; a handshake game is one where both teams agree, without saying so, to farm in peace and fight late. The meta is the set of picks and strategies most players treat as strongest at the moment. Riot didn't want the world championship played that way.

Solo queue took that sentence to heart. A fifth of Yunara players left on patch day, before they could have learned anything from playing her, and over two weeks a third were gone. Her win rate fell about a point, and the range on that runs from a three-and-a-half-point drop to a one-point gain. Two damage per hit, plus a sentence calling her a problem, emptied a third of her games while her results moved about a point. That's Jojopyun's placebo nerf, in 202,725 games.

It was a nerf aimed at a meta-level, pro team habit, for one tournament, and it changed everything and nothing at all: a third of her players stopped picking her, while her win rate moved about a point. With two damage per hit, Riot steered what solo queue picked. Patch notes move players, and on Yunara they moved players much further than they moved the champion's in-game performance.

Jojopyun's placebo nerf has a second half: two weeks later, players find out the champion was fine and come back. The 26.20 nerfs had only two weeks before the next patch, so I checked the five nerfs from 26.19, which Riot then left alone for four weeks. Players didn't come back. Against untouched champions, picks of Nocturne, Poppy, Rumble, Ryze and Vi were down 9% in the first week and 19% in the fourth, and the drop deepened instead of recovering. Vi was the only one near her old numbers by week four. I added this check after I'd locked my plan, so read it as a strong lead rather than a tested result.

Outer core

A seventh of us walked away

Yunara was the extreme case. Riot nerfed Ambessa, Ashe, K'Sante and Yunara in 26.20, four champions solo queue plays enough to measure reliably. Take every game those four showed up in during the two weeks before the patch and call it 100. In the two weeks after, players put them into about 86. Out of every hundred games those champions did play, they won one or two fewer than before, and the true number could be anywhere from three and a half fewer to none. Fourteen games in a hundred went to other champions, while the champions themselves lost a game or so in a hundred. Players moved much further than the champions did.

That could have been an illusion. If the casual players left and the veterans stayed, the veterans would win more and hide a real drop in strength. League counts mastery points for every champion you play, and the more you play one, the more points you have. I sorted players by those points and checked who left. The share of experienced players on the nerfed champions moved by a tenth of a point.

Outside Yunara, you didn't leave on patch day. On day one I couldn't see any drop. A week in, picks were down 15%, and by the end of the second week, 18%. Players drifted off over two weeks, as word of the patch spread from game to game.

Aimed at the pros, felt by everyone

Riot often writes a nerf for one slice of players. K'Sante's nerf raised the mana cost on one ability and trimmed how much damage he blocks, changes that matter most to a pro team running planned fights for forty minutes and less to five strangers in a solo queue game. Riot's note said as much: the changes "should be skewed for elite audiences." That's the idea behind a targeted nerf: hit a champion where it's too strong without wrecking it for everyone else.

Solo queue players treated every nerf the same. I sorted the nerfs from 26.19 and 26.20 into two piles: the ones Riot said it aimed at pros, and the ones Riot aimed at everyone. Players dropped both piles by the same amount, about 13% of their games.

Then I checked whether solo queue had any reason to trust the label. If a nerf aimed at pros spared ordinary games, those champions should have kept their solo queue win rate. They didn't. The pro-aimed nerfs cost solo queue players about two points of win rate, and the nerfs for everyone about one and a third, with ranges that overlap. Riot aimed at the pros, and solo queue took the hit too. The players who ignored the label and assumed a nerf for the pros meant a nerf for them, too, were right.

Inner core

A skin moves you too

If a sentence in the notes can move a third of a champion's players, I wanted to know if anything moves us without changing the game at all. Riot sells skins, new outfits for champions that change how they look and nothing about how they play. I found 13 champions that got a new skin in 26.19 or 26.20 and no balance change. Players picked them a tenth more, and their win rate stayed where it was. Janna, a support (the role that protects teammates rather than chasing kills), got a new skin in 26.19 and gained almost two-thirds more players. A new outfit moved picks about two-thirds as far as a nerf did, in the other direction.

A new outfit moves picks because you choose from what you notice. That's the attention economy, and stock traders live in it too. On a Sunday in May 1998, the front page of the New York Times ran a story about two cancer drugs a small company called EntreMed was testing. The research had come out in a science journal the previous November; the story added nothing new. On Monday, EntreMed's stock opened at about $85, up from about $12 at Friday's close. The economists Robert Huberman and Tomer Regev later called it contagious speculation, and Brad Barber and Terrance Odean showed in 2008 that investors buy from the stocks that grab their attention and judge the value second.

Patch notes are League's front page. For skins and buffs, any press is good press: a mention alone brings players in. For a nerf, the press runs the other way. A scary headline sends players out even when the numbers underneath are small, as they were for Yunara.

Mantle, far side

The stage plays by its own notes

Then I went to the players Riot wrote the nerfs for. I read every Worlds from 2019 to 2026 on Leaguepedia, the esports wiki, 833 games in all, and every champion change Riot made in the three patches before each one, 414 changes, sorted by what Riot said it wanted.

Pros don't chase buffs. Across eight Worlds, teams picked or banned buffed champions about as often as champions Riot left alone; buffed champions came out ahead in four years and behind in the other four. This year, through 26 October, teams picked or banned a buffed champion in 3% of games on average. A buff in the notes doesn't buy a champion a place in a Worlds draft.

Pros draft the nerfed champions instead. Every year, teams picked or banned the champions Riot nerfed before Worlds in 28% to 43% of games, against 7% to 12% for the untouched ones. My read of the order: pros play a champion a lot, Riot nerfs it for being too strong on stage, and pros keep playing it because a weaker version is still their best option. Azir, a mid laner who commands sand soldiers, shows the pattern across years. Riot changed him before five of the eight Worlds, each time with a nerf in the package, and in 2025 teams still picked or banned him in 82% of games.

This year's five fit right in. While solo queue drifted away from them, teams at Worlds picked or banned Cassiopeia, a snake-bodied mage and the fifth champion nerfed in 26.20, in two of every three games and K'Sante in half. Solo queue read the notes and left; the pros read the notes and kept drafting.

Crust, far side

K'Sante, on both sides of the screen

K'Sante is where the two stories meet. Riot said the nerf targeted his reliability as a blind pick, so I measured that. A safe blind pick has no terrible matchup: whoever it meets in lane, it wins about as often. For each champion, I measured how far its win rate swings from one lane opponent to the next, once I'd removed the noise of small samples. A typical champion swings about two and a half points.

Before the nerf, K'Sante in the top lane fit the description. Even his worst matchup, Kayle, a winged fighter who grows stronger the longer a game runs, came out at 47.5% once I'd removed the noise of small samples. After the nerf his swing almost tripled, and my best estimate has Teemo, a small poisonous scout, beating him almost two games in three. Riot said the nerf would make him a riskier first pick, and the lane data shows it did.

His solo queue players stayed. They picked him as often after the nerf as before, and they lost about five points more of their games, the biggest drop of the four 26.20 nerfs I could measure. One thing may have kept them: the same patch released HEARTSTEEL K'Sante, a new skin, and a new skin alone moves picks about a tenth. I can't separate the two.

Some small nerfs do real damage. K'Sante's cost his players about one game in twenty, and his worst lane matchup collapsed. Yunara's didn't, at least not yet: her win rate moved about a point, a change no player could feel across a week of games. Same patch, same kind of note, two different outcomes, and nothing in the notes told you which was which. Yunara's players left and K'Sante's players stayed, and both guessed wrong.

The pros stayed too. After that 0 and 5 in the Play-In, teams kept picking him into the Main Event, where through 25 October he'd won 4 of 9. On both sides of the screen, K'Sante's players kept the faith. I counted too few K'Sante top games in solo queue for the full safe-pick test I'd planned, so treat these as a careful description of two patches rather than a final ruling.

Put to bed, woken up

It takes more than one patch to finish a champion. K'Sante's 26.20 nerf was the third small cut in a year and a half. Riot trimmed his passive, a mark that adds damage to his hits, and the armor-ignoring damage he deals during his ultimate, his biggest ability, in 25.05; his shield in 26.13; and his W, his second ability, in 26.20, with one small buff to that same W in between. Each change on its own looks like a rounding error. Together they're the straws on the camel's back, and 26.20 was the straw that broke his lane matchups.

Riot can trim a champion, but at Worlds the meta decides when it sleeps. Azir shows it. Teams picked or banned him in 72% of games in 2022. In 2024 they never touched him, 0%, after Riot buffed him and then nerfed him in the weeks before that Worlds. In 2025 Riot nerfed him twice, and teams picked or banned him in 82% of games anyway. This year Riot buffed him in 26.16, and he's back at 0%. Buffs didn't wake him and nerfs didn't put him down. Something other than his own line in the notes sent him to sleep in those years: the items, the systems, the mid laners around him. My data can't say which.

Yunara is the one case where Riot's aim and the stage line up. Teams picked or banned her in 88% of Worlds games in 2025, Riot nerfed her for 2026 to keep her style out of the Worlds meta, and she's down to 20%. Even there, one Worlds can't separate her nerf from a meta that moved on without her.

For a solo queue player, that's the case for patience. Riot will trim the champion you love, and the meta will turn again, with a buff or without one.

Far surface

Kog'Maw, the off-meta pick

Back to Spawn and his off-meta pick. Riot didn't touch Kog'Maw in 26.20, and solo queue players picked him 17% more. A champion whose picks rise with no buff makes you look at what changed around him. Kog'Maw's W, his second ability, Bio-Arcane Barrage, stretches his attack range and adds damage based on the target's maximum health, so it melts tanks. And the safe mid picks around him had been taking small nerfs: Riot trimmed Ryze in 26.19 and Cassiopeia in 26.20. If the reliable picks in a lane get a little worse, an odd pick gets a little better without anyone touching it.

The timing fits. On 26.18, solo queue played Kog'Maw mid in 9% of his games. On 26.19, the patch that nerfed Ryze, that share jumped to 16%, and it stayed near there on 26.20. I can't prove the nerfs sent players to him, but the move happened on the patch they shipped.

Was he always good? The few players who took him mid on 26.18 won six games in ten, over about 200 games, so the true rate is somewhere between 54% and 68%. Once the crowd arrived, his mid win rate fell to 53% on 26.19 and 45% on 26.20. The first players found a real edge, and the players who followed spread it thin while opponents got more chances to learn the matchup.

A champion rises, Riot trims it, the champions around it shift, and something few people were playing turns good without a single change to its kit. Then the crowd finds it, the edge wears off, and the wheel turns again. Is that not the circle of life in MOBAs?

Then on 16 October, MVK Esports, a team from Vietnam, put Kog'Maw in the mid lane on the Worlds stage and won two of three games with him. I counted where solo queue players took him, day by day. In the two weeks before MVK's games, solo queue already sent him mid in about one game in seven. On the two days MVK played him, that rose to one in five, then settled back to about one in six, inside the normal range for any given day. Solo queue had found mid Kog'Maw weeks before the pros put him on stage. The off-meta pick that wins a Worlds game might already be in your ranked games.

Submit or persist

Back to that night after the Play-In: solo queue leaves a nerfed champion, and the pros stay. I think pride, experience and comfort decide which one you do. You contextualize a nerf the way you do any bad news: you either submit to it or you persist against it. Most of solo queue submits and finds a new champion. A pro persists, at least at first, because a pro trusts their hands to beat a few points of damage. And in the moment you lock in, you want what feels safe, and a champion with a red arrow next to its name doesn't.

On stage, a pro keeps a champion until it stops winning, red arrow or not, which is why teams kept drafting this year's nerfed five. Solo queue reacts, and drops the champion before playing a game on the new patch. By my count, the pros' habit makes the better default. Across the four 26.20 nerfs, sticking with your champion cost about a game and a half in every hundred. The exception is a nerf that hits the one thing a champion does best, like K'Sante's reliability as a first pick, which cost his players a game in twenty. Read what a nerf targets, then decide like a pro, in your games and in the calls you make outside them.

A one-trick, someone who plays a single champion for years, persists by answering the patch notes with skill: their champion can be weak and they still know it better than anyone they face. Thebausffs built a career on one-tricking Sion, and reached Challenger in Korea in under a month. A pro can't live that way. In Fearless draft, which pro leagues use, once either team plays a champion in a series, neither team can play it again in that series, so a pro needs a deep champion pool and has to follow the notes to keep it current. A pro reads the notes to choose what to learn, then persists with what the team drafts. If you want steady climbs without bashing your head against a wall, you tend to be one or the other: the specialist who outlasts the notes, or the generalist who studies them.

I tried to be the specialist: Zilean support, one-tricked all the way to Challenger, or so I figured. No patch note ever scared me off him; I'm much too stubborn. It was a skill issue, which is how I ended up reporting on League, watching it and analyzing it: anything but playing it.

My data puts one limit on the one-trick theory. Sorted by mastery points, the most experienced third of players on each nerfed champion left at the same rate as everyone else. Whether they won more than newer players after the nerf, my mastery sample was too small to say: the two ranges overlap and both cross zero.

The limits

Teams picked all five 26.20 nerfs on the Worlds stage in the same two weeks that solo queue drifted away from them, so I can't separate reading the notes from watching the pros. Next Worlds, I'll look for a nerf the pros skip. My Worlds coding also leaves out the small mid-patch updates. Three questions I didn't get to: whether Yunara's nerf did what Riot wanted and cut her power in long games, where her players went, and whether one-tricks shrug off nerfs better than everyone else, which needs a bigger mastery sample. And I changed one test before I looked at any result: a simulation showed my sample was too small for a sharper measure I'd planned, so I used the side-by-side comparison you've read here. The method page has the details.

Next time you read the notes

• A nerf with a scary sentence and small numbers: give the champion a few games before you drop it. Yunara lost a third of her players over two damage per hit.
• A nerfed champion in a Worlds draft: look up its history. Odds are Riot nerfed it because pros already played it, and they'll keep playing it.
• K'Sante top, picked first: check whether the other team has a Teemo.

Hope for the one-trick

That leaves hope for the average player: stick it out with your one-trick. Most nerfs cost your champion a game or two in a hundred, and the skill you've built on it is worth more than that. The awe of watching pros comes from the opposite gift. They take their skill to every champion, so what you see on stage is skill alone, with no comfort pick to lean on. And when a pro does lock in their specialty, the champion they'd one-trick if they could, the arena leans in, because you might be about to see an even higher level.

The full method, the data and every number's source: the method and the data.

Filed by Census Officer Kaosi Unini. Another unnecessary statistic from the Bureau of Rift Statistics. The Rift belongs to Riot; the errors are the Bureau's own.

This article was created under Riot Games' "Legal Jibber Jabber" policy using assets owned by Riot Games. Riot Games does not endorse or sponsor this project.