Five analysts published their Worlds 2026 Pickems predictions in the days before the lock - all eighteen Crystal Ball answers and a call on the Play-In. We put the five entries side by side, counted where they agree, and scored how strong each agreement actually is.
This is not a tip sheet. Where the panel is unanimous the question is close to free, and you should take the answer and move on. Where it splits three to two, nobody on this page knows more than you do - and those are the questions worth your thinking time.
If you do not follow the teams closely enough to have an opinion, this is the fastest way to fill a bracket that is not a coin flip: copy the unanimous answers, which cost nothing to agree with, and spend your guesses only where the experts themselves disagree.
- All three 100-point Crystal Ball questions are unanimous - 300 points on which five of five agree.
- The Crystal Ball locks 15 October, 11:00 PDT, in the same minute as the Play-In bracket.
- One question got five different answers from five analysts: highest win rate.
- The panel is weighted equally. Nobody has a published hit rate, so nobody gets a bigger vote.
- Swiss picks are open and lock 23 October, 17:00 UTC - eight advancing teams, three of them with a record bonus.
- The panel missed the Play-In: 76% said Cloud9, Karmine Corp won it.
Who is on the expert panel
Three languages, three roles: two former professional players, two full-time analysts and one caster. Each entry below links to the full breakdown it came from - the reasoning is theirs, the counting is ours.
- CaedrelEx-LEC pro, now the biggest English-language watch-along analystFull breakdown (YouTube, original source)
- JankosFormer LEC jungler, ten Worlds appearancesFull breakdown (YouTube, original source)
- TolkinGerman-language caster and analystFull breakdown (YouTube, original source)
- AkofenaSpanish-language analyst, cross-regional focusFull breakdown (YouTube, original source)
- StarkyRussian-language analyst, bootcamp and scrim sourcingFull breakdown (YouTube, original source)
We record each answer at one of three strengths: a main answer, a named alternative, or a option mentioned as a long shot. That is what the percentages further down are built from.
Crystal Ball predictions
Eighteen questions, 1,050 points, one deadline, and no second window. Here is the panel's answer to every one of them, with how much of the panel's conviction that answer holds.
Fill this part first. The Crystal Ball closes at 11:00 PDT on 15 October, before the first Play-In game, and unlike every bracket stage it never reopens - a missed Crystal Ball is 1,050 points you cannot win back later in the tournament.
What the consensus is worth
The three expensive questions - the Peyz and Gumayusi kill duel, Faker's career kill count and whether anyone picks Teemo - are worth 100 points each, and all three came back unanimous. Two of them are answered "no". That is 300 points, nearly a third of the Crystal Ball, on which five independent analysts said the same thing without hedging.
The pattern holds lower down. Of the eighteen Crystal Ball questions, seven came back unanimous, four more with only a shade of doubt, four with a clear lean and three with no panel answer at all. Read those last three as a map of where the game is actually decided: nobody gains an edge on a question everybody answers the same way.
One answer is worth singling out. Nautilus was named by every analyst on the board for highest average deaths - by two as their main answer and by the other three as the alternative behind a jungler. No other option anywhere on this page has that kind of coverage, which is how it leads its question on 64% while only two analysts actually picked it first.
Every Crystal Ball question, answered by five experts
Name = main pickName = named as an alternative
Which champion will have the highest presence (picked or banned) across all drafts?
- Vi76%CaedrelJankosTolkinAkofenaStarky
- Orianna28%JankosStarky
- Kalista20%Jankos
- Syndra20%Tolkin
- Cassiopeia11%JankosStarky
- Nocturne11%TolkinAkofena
Vi is on every single list - three analysts put her first and the other two name her behind their own pick. She is the engage jungler the eastern teams ban in game one of a Fearless series and keep banning. Jankos goes elsewhere: Kalista came through the recent patches untouched and is the safest blind pick on the bot lane. Tolkin calls Syndra his emotional choice.
Which champion will have the highest win rate (minimum five games played)?
- LeBlanc28%CaedrelStarky
- Akali20%Jankos
- Camille20%Akofena
- Gwen20%Starky
- Neeko20%Caedrel
- Yone20%Tolkin
- Cassiopeia8%Jankos
- Lissandra8%Starky
- Cho'Gath3%Starky
- Draven3%Starky
- Kled3%Starky
Five analysts, five different first answers - the widest split on the board. The question rewards a narrow counter-pick that only appears in favourable drafts, and there is no converging on one of those in advance. The five-game minimum matters too: Starky flags Kled, Draven and Cho'Gath as picks that might never reach it.
Which champion will have the highest average deaths per game (minimum five games)?
- Nautilus64%CaedrelJankosTolkinAkofenaStarky
- Jarvan IV20%Starky
- Rell20%Caedrel
- Vi20%Jankos
Every analyst named Nautilus - two as their main answer, three as the alternative behind a jungler. The engage support goes in first, gives up its crowd control and dies, and in an enchanter meta nobody is there to peel. Starky adds the caveat worth keeping: if bot lanes go ranged, Nautilus gets picked less and Jarvan IV inherits the dying.
Which pro will finish with the highest average KDA?
- Peyz48%TolkinAkofenaStarky
- Ruler40%JankosStarky
- Viper36%JankosAkofenaStarky
- Caliste20%Caedrel
- Chovy20%Starky
- Gumayusi8%Starky
- Hope8%Starky
Caedrel is playing the format rather than the player: a team that starts in the Play-In can run up a KDA against weak opposition and then be knocked out before the giants spoil the number. Everyone else backs a carry on a team they expect to go deep - the opposite bet, because late rounds are where KDA goes to die.
Which pro will finish with the highest average damage per game?
- Peyz100%CaedrelJankosTolkinAkofenaStarky
The cleanest answer on the board: five out of five with no hedging. His team funnels the game through the bot lane and damage share follows gold share. All five name the same fallbacks if it goes wrong, Viper and Ruler, but nobody actually picked them, so they are not scored here.
Which pro will finish with the highest average kill participation?
- Tarzan40%JankosTolkin
- Keria36%JankosAkofenaStarky
- Duro28%AkofenaStarky
- Histo20%Caedrel
- Delight8%Starky
- Kael8%Starky
- Labrov8%Tolkin
Junglers and roaming supports own this stat, and the panel splits on which kind. Keria edges ahead because three of the five name him somewhere, though only Starky picks him outright. Caedrel goes furthest from the field with Histo, on the grounds that nearly three quarters of his team's action runs through him.
Peyz vs Gumayusi: who will finish with more total kills?
- Peyz100%CaedrelJankosTolkinAkofenaStarky
A hundred points with no dissent. Starky calls it roughly 70-30 rather than a certainty: total kills depend on how far each team goes, so Gumayusi takes it only if his team outlasts the Korean favourites.
Will Faker reach 700 career Worlds kills this year? (616 entering, over ten appearances)
- No100%CaedrelJankosTolkinAkofenaStarky
Arithmetic, not opinion: 84 or more kills in a single tournament against a career average near 60, from a player on control mages whose team skips the Play-In entirely. It would take a run to the final and more than four kills a game along the way.
Will Teemo be picked at least once?
- No100%CaedrelJankosTolkinAkofenaStarky
The third hundred-point question and the third unanimous answer. Fearless Draft forces teams deeper into their champion pools than usual, which is the only argument on the other side - and none of the five took it.
How many unique champions will be picked?
- 100-10480%CaedrelJankosAkofenaStarky
- 110-11420%Tolkin
The four in the lower band anchor on recent Fearless Draft events, where the hundred mark has been the standard. Tolkin argues this year is different: six full best-of-five series in the Play-In instead of one drags more champions onto the board before the Swiss stage even starts.
Will more games finish before or after the 32 minute mark?
- After 32:0080%CaedrelTolkinAkofenaStarky
- Before 32:0020%Jankos
The four point at median game lengths across the major regions this season, which sit between 32:30 and 33:00 - just the wrong side of the line. Jankos bets on the Swiss stage instead: one-sided best-of-ones between seeded teams and the rest of the field end early.
Will we see any pro get at least 15 kills in a single game?
- Yes60%JankosTolkinAkofena
- No40%CaedrelStarky
A genuine coin flip, and the panel knows it. The yes camp needs one abnormal game from a carry like Peyz or Viper; the no camp points out that games at Worlds are tighter than domestic games and the recent record sits just under the line.
Will we see any pro reach a creep score of at least 450 in a single game?
- Yes100%CaedrelJankosTolkinAkofenaStarky
Unanimous, though Starky rates it closer to 60-40 than a certainty. Wave timers were sped up this season, so farm accumulates faster than the number suggests, and a tournament this long reliably produces the forty-minute games that get there.
What will be the role of the finals MVP winner?
- Mid60%TolkinAkofenaStarky
- ADC20%Jankos
- Top20%Caedrel
Mid is the historical default at Worlds finals and three analysts take it. The answer mostly encodes who each of them expects to win the thing: a mid pick is a bet on the Korean and Chinese favourites, Jankos is betting on a carry ADC, Caedrel on the top laners who decided recent finals.
How many different regions will be represented in the knockouts?
- 4100%CaedrelJankosTolkinAkofenaStarky
- 36%CaedrelStarky
All five answer four, but two of them flag that three is the mathematically safer answer: Korea and China would normally fill all eight seats between them. Four needs exactly what this panel expects anyway - one European and one North American team getting through.
Will a team from North America reach the knockouts?
- Yes100%CaedrelJankosTolkinAkofenaStarky
Unanimous, and it rests almost entirely on one team: Team Liquid are rated as the region's genuine contender this season.
Which team will place higher: HLE or T1?
- T188%CaedrelJankosTolkinAkofenaStarky
- HLE20%Starky
Four analysts take the pattern: T1 have not finished outside the top four at Worlds in years, whatever their domestic form says. Starky is the one who separates the two answers - by the season HLE are the logical and safer pick, but he admits the T1 record on this stage is what pulls everyone the other way.
Which team will win the shortest game by duration?
- Gen.G60%TolkinAkofenaStarky
- An LPL team20%Jankos
- Karmine Corp20%Caedrel
Gen.G punish mistakes faster than anyone and will meet weak opposition in the Swiss stage. The other two look at the opposite end of the field: a Play-In team that either snowballs or gets snowballed, or the Chinese teams who play the most aggressive early games. Note that Jankos named a region rather than a single team, which the official question does not allow.
Play-In: how the panel did
The first stage is settled, and the consensus lost. Karmine Corp won the Play-In - they beat Cloud9 3-2 on the opening day, watched FURIA knock Cloud9 out entirely, and then swept MVK Esports 3-0 for the Swiss seat.
Three of the five had Cloud9, and the other two had Karmine Corp while calling Cloud9 the safe answer. So the panel's answer came out at 76% for Cloud9 - a Strong rating - and it was wrong. Starky was the only analyst to call it end to end: Karmine Corp over Cloud9 in the opener and Karmine Corp through to the Swiss stage, on the strength of scrim reports from the Korean bootcamp rather than the season record. Tolkin had the right team for an avowedly emotional reason.
Worth keeping in mind as you read everything below: a Strong rating means five informed people leaned the same way, not that the answer is safe. This is the first test it has failed.
| Question | Pts | Panel answer | Agreement |
|---|---|---|---|
| Who wins the Play-In and takes the one Swiss stage seat? | 35 | Cloud9 | Missed |
Name = main pickName = named as an alternative
Who wins the Play-In and takes the one Swiss stage seat?
Panel missed: Karmine Corp
- Cloud976%CaedrelJankosTolkinAkofenaStarky
- Karmine Corp48%CaedrelTolkinStarky
Karmine Corp won the Play-In: 3-2 against Cloud9 on the opening day, then 3-0 against MVK Esports in the deciding series. Cloud9 were eliminated by FURIA before that final. Starky and Tolkin had it; the other three did not.
Swiss and knockout predictions
Swiss picks are open now and lock on 23 October at 17:00 UTC. They work differently from everything above: instead of answering questions you name the eight teams you expect to advance, and your first three picks earn a bonus if they come through on the record you gave them - 3-0, 3-1 or 3-2. The round 1 draw and the full sixteen-team field are on Worlds Pickems 2026.
The panel's Swiss picks go here as the analysts publish them, in the same format: one bar per team, showing how many of the five have it in the eight. In the meantime you can play the stage out yourself in the Worlds 2026 simulator - it draws each round the way Riot does and fills in the sheet as you go. Knockout picks open on 1 November and lock on 3 November, and get the same treatment.
How we score the consensus
Every analyst carries the same weight. That is a deliberate choice: none of the five publishes a verified record of past Pick'Em accuracy, so any weighting would be our opinion dressed up as arithmetic. If this page runs for a few events, we will have a record of our own to weight with - built in public, on this page.
Most analysts name more than one option per question: a pick they are making, then a couple of names they would accept instead. Those are not the same thing, so they do not count the same.
- Main pick - the answer they are actually entering. Full weight.
- Alternative - a name they would accept instead. 40% of the weight.
- Long shot - mentioned, with the risk spelled out. 15%.
The gap between the first two is deliberately wide. At a gentler ratio two people mentioning a name in passing would outscore one person actually picking it, and the page would end up recommending an answer nobody chose. Two alternatives still weigh less than one main pick.
On top of that there is a hard rule: the panel's answer must be somebody's main pick. Options are ranked by how many analysts picked them first, and the percentage only breaks ties. That is why the kill participation question is led by the name with two first choices rather than the one mentioned by three people.
In every breakdown the names under a bar show who named it, and a mint-coloured name means it was that analyst's main pick. A plain name means they offered it as an alternative. Five mint names on one bar is the strongest signal available on this page.
Each question then gets a label from the percentage of its leading option:
- Unanimous - 100%. All five picked that answer outright, with no hedging.
- Strong - 70% or more. Four first choices, or three backed by the other two as their alternative.
- Majority - 50% to 69%. Three first choices, with a live alternative behind them.
- Split - under 50%. The panel does not have an answer, only candidates.
We deliberately do not publish a separate "risk score". With an equally weighted panel it would be a mirror image of the consensus number and would tell you nothing twice.
Track record
One stage is scored so far, and it went against the panel.
| Stage | Panel answer | Result | Verdict |
|---|---|---|---|
| Play-In winner | Cloud9, 76% | Karmine Corp | Missed |
| Swiss stage | Picks open | Resolves 31 October | - |
| Crystal Ball, 18 questions | Locked 15 October | Resolves 14 November | - |
Individually: Starky and Tolkin had Karmine Corp, the other three had Cloud9. One stage is far too small a sample to rank anybody on, which is the point of keeping the table - it only becomes worth reading after several events, and it is the reason this page does not weight one analyst above another.
That record is the point of running this across a whole tournament rather than publishing one prediction page and walking away. Agreement is easy to measure in advance; being right is not.
Frequently asked questions
I do not follow the teams - what should I pick?
Take every answer marked unanimous or strong as it stands: those are questions where five people who watch this for a living landed in the same place, and your own guess cannot beat that. Then spend your attention on the handful marked split - those are genuinely open, and an uninformed guess there costs you no more than an informed one.
Should I just copy the consensus answers?
For the unanimous questions, there is little reason not to - five analysts with different biases landing on the same answer is about as much signal as this game offers. For the split questions, copying a 40% answer is not safer than picking your own, it only feels that way.
Are these the analysts' official entries?
They are the picks each analyst explained publicly in the breakdowns linked above. Where someone named a main answer and an alternative, we recorded both, which is why some questions show more options than analysts.
Why does the Play-In show one question instead of six?
Because that is what the analysts actually answered. The official bracket asks for six series picks, but all five discussed the stage as a single question - who comes out of it - and splitting their reasoning across six rows would mean writing picks they never made. When a source is thinner than the question, we show the thinner thing.
When does this page update?
When a stage opens and when it resolves. The Swiss bracket opens on 19 October, the knockout bracket on 1 November, and results land as each stage finishes. The Crystal Ball cannot be scored until the tournament ends on 14 November.
Where do I actually enter these picks?
On the official LoL Esports site, signed in with a Riot account. This page is an independent analysis - nothing entered here counts. If you have not filled a bracket before, how to play Pickems walks through it.