The MSI Compass: When a Correct Signal Gets Read as a Rule

On the night of July 12, 2026, the Nexus inside the favoured side's base shattered on stage in Vancouver. I was sitting nearly ten thousand kilometres away, in a small apartment in Guangzhou, still holding a cup of coffee that had gone cold back in game three. A Korean team lifted the MSI trophy. Ten seconds later, the analyst desk changed the subject. On to Worlds. Nobody was allowed to celebrate for long. Nobody gave themselves permission to. I have an odd habit, and I know it is odd because a colleague once saw my spreadsheet and asked what was wrong with me. Every time MSI ends, I open an empty sheet and type three lines into it. The champion. The runner-up. The date the final ended. Then I close the laptop and wait until November. That sheet has never predicted anything perfectly. But every year it teaches me a new lesson about reading data the wrong way — and the most common wrong way, for me, is reading a correct signal as though it were a rule. In 2026 I had to sit with it longer than usual. For the first time in years, the signal coming out of MSI looked clean. Clean enough to want to believe. CONTEXT: A TOURNAMENT FAMOUS FOR BEING A BROKEN COMPASS MSI is League of Legends' mid-season international event, usually staged from late May to mid-July. Worlds closes in November. The gap between them is three to four months, and in that window the meta can be torn apart at least twice. MSI launched in 2026 with six teams, and for years it was shaped like a highly competitive but short invitational. Group stage, semifinal, final. A dozen or so games for the champion. In 2026 the event was not held at all. From 2026 to 2026 the format framework stayed the same. For almost a decade, analysts assumed one thing, and I assumed it too: winning MSI says very little about Worlds. History backed that assumption hard, and it backed it almost cruelly. EDG won MSI 2026 and exited in the Worlds quarterfinals that same year. SKT won MSI 2026 and fell to Samsung Galaxy in the Worlds final, in a match where I still remember the exact moment Faker slumped onto his keyboard. RNG won MSI 2026 after a near-perfect summer, then lost to G2 in the Worlds quarterfinals — a series I still have to pause a few times whenever I rewatch it. G2 won MSI 2026 and lost the Worlds final to FunPlus Phoenix. JDG won MSI 2026 and lost to T1 in the semifinals. Gen.G won MSI 2026 and also lost to T1 in the semifinals, on Chinese soil. Across that entire list, exactly one season saw the MSI champion go on to win Worlds: SKT in 2026. One out of eleven. That is the base rate anyone talking about MSI predicting Worlds has to face before opening their mouth. The summer of 2026 taught us one thing: the meta exists only to be broken. So when someone tells me MSI is becoming a reliable compass, I ask exactly one question back: reliable in what sense? CORE: THE EXPANDED FORMAT CHANGED WHAT MSI MEASURES Starting in 2026, MSI changed its format. A winners' and losers' bracket was introduced. More teams. More series. More games. This is the detail most analysis skips over. To me, it is the centre of everything. Under the old format, an MSI champion needed roughly thirteen to fifteen games before lifting the trophy. Under the expanded format, the champion plays more than twenty, and must move through a bracket structure that threatens to drop them into the lower bracket at any moment. That difference is not about broadcast hours. It is about the nature of the test. A short tournament measures individual skill inside one specific patch. A long tournament measures adaptability. And adaptability is the only thing that transfers from July to November. What the expanded MSI can predict and what it still cannot sit on two different layers: it filters the right contender pool, but it does not name the final winner. Look at the last three years, and look at them the way a data person looks rather than the way a storyteller looks. In 2026, JDG won MSI and BLG finished second. At Worlds, both reached the semifinals. T1 won. In 2026, Gen.G won MSI and BLG finished second. At Worlds, BLG reached the final, Gen.G reached the semifinals, T1 won. In 2026, Gen.G won MSI and T1 finished second. At Worlds, T1 won. Read conventionally, this table says MSI got the champion wrong two times out of three if you only look at first place. Read differently, it tells a completely different story: in all three years, the Worlds champion was inside the group of teams that went deep at MSI that same year. T1 won Worlds twice in a row after going deep at MSI. And in the third instance, 2026, T1 was the team that lost the MSI final and then lifted the Worlds trophy. This is where I have to stop and be very precise, because a lot of analysis is blurring this line: the MSI runner-up is a more interesting predictive datapoint than the MSI champion. The reason is simple, and it is uncomfortable for anyone who prefers solitary-hero narratives. An MSI champion enters Worlds as a target. The whole world has studied them for three months. Every draft they run has been dissected down to the last ban. Meanwhile, a team that finished second or third at MSI enters Worlds with an underexploited playbook, an equivalent level of tactical maturity, and at least one fresh wound to burn as fuel. Tracking matches across multiple seasons, I have seen this repeat often enough that it stops being coincidence. The team that peaks at MSI typically loses around a third of its surprise value by the time Worlds arrives. The runner-up and the third-place team keep theirs intact. In a tournament where every knockout series is a best-of-five, losing a third of your surprise value costs you one game. Losing one game costs you the season. That pressure never shows up on the scoreboard. It shows up in a teamfight where the stronger side hesitates on the fourth second instead of the third. One second. The entire season lives there. I have seen the same thing in another sport. Argentina in 2026 did not play football — they played a perfect disengage composition, and the whole world could only watch. The resemblance between Argentina 2026 and T1 2026 is not tactical. It is that both were collectives that had already lost, and turned the loss into structure. REGIONAL READ: WHY LPL AND LCK DOMINATE ABSOLUTELY Another detail worth pausing on: the entire contender pool MSI predicts correctly comes from just two regions, LPL and LCK. Short-term statistics do not make this obvious, but long-term data says it loudly. For years, Worlds champions have come only from China and Korea, with a single exception in Fnatic in 2026, at a tournament still in its infancy. In League of Legends, international competitiveness is a two-pole story, and every predictive analysis has to start there. Which means MSI can only ever be right inside a very narrow space. If you already know the Worlds champion will come from LPL or LCK, then MSI pointing at the correct region is not much of an achievement. It is a pre-loaded inference. I am not arguing that the gap between these two regions and the rest of the world is absolute. I am arguing that when you group the two strongest regions together and declare that the group always wins, you are correct but not necessarily useful. The real value of the expanded MSI sits somewhere else, and it is much narrower: it helps distinguish which team inside that group is actually ready. THE TURN BETWEEN JULY AND NOVEMBER There is one stretch nobody analyses enough: the roughly three months between the two events. It is a period where the standings do not change but everything inside them does. Teams gather in Korea or China for scrims. Patches arrive one after another. And a handful of organisations decide their problem is the people, not the system. I have always cared more about mid-season moves than pre-season moves, because they are more diagnostic. A team that swaps a player in July is publicly admitting it misdiagnosed in January. A team that swaps nobody is saying something similar in the opposite direction. In esports, mid-season transfers usually happen through buyout clauses whose value is never fully disclosed. That is why transfer data in this industry is far harder to analyse than in sports with public mechanisms. The structure accidentally protects weak deals, because nobody holds enough data to interrogate them. But you do not need the price to read the direction. You only need the timing. A team that changes its mid laner in late August has usually already abandoned the season. A team that hires an analyst coach in mid-September is usually genuinely trying to fix one specific bug. The gap between those two moves is enormous, and it has nothing to do with the reputation of the person signed. That is also why I refuse to use MSI to draw conclusions about Worlds. Too many variables squeeze in between, and most of them never appear on any stat sheet. THE CONTRARIAN ANGLE: THREE YEARS IS NOT A RULE Now comes the part I know will annoy some people. Three years. Three expanded MSI editions. That is the entire body of data the community is using to discuss a new trend in the discipline. Three datapoints do not make a trend. Three datapoints make three datapoints. I have written about seasons where every indicator pointed one way, then the knockout stage arrived and everything flipped inside forty minutes. In 2026, RNG entered Worlds as the best team on the planet according to nearly every ranking. They left in the quarterfinals. Anyone who used that year's MSI result to bet was wrong in the most expensive way available. Fate never plays favourites; it only rewards whoever knows how to read RNG. And in this case, that name is both a team and a pair of dice. There are three reasons I keep my distance from the MSI-is-a-compass reading. The first is patch distance. MSI happens mid-year. Worlds happens at year's end. In between, the publisher ships adjustments capable of erasing an entire playstyle. Any team whose MSI success was built on a specific champion or a specific tempo risks walking into Worlds with luggage that has expired. The second is survivorship bias. We remember the MSI champions who went on to shine at Worlds, and we forget the MSI champions who vanished. Collective memory is a selective filter with enormous bias. It keeps the good story and quietly deletes the rest. The third, and most important: the expanded format is a structural change, not a change in the nature of the discipline. It makes team filtering more accurate. It does not make the final outcome more inevitable. A better-designed tournament produces a more accurate contender pool. It does not produce a predetermined champion. A great coach is not the person who draws the meta, but the person brave enough to erase it. That applies to the people sitting at the tournament design table too. Every failure begins with a bug the team was too complacent to fix. With MSI, that bug is reading a short-term signal as though it were a long-term law. SIGNALS TO TRACK IN 2026 Three things go into my spreadsheet. First, the region of the 2026 MSI champion. If a team outside LPL and LCK wins, the data streak breaks, and every conclusion drawn from the last three years has to be rewritten from scratch. The probability is low, but it is the cleanest test of this hypothesis. Second, the finishing positions of the runner-up and third-place teams. If the MSI finalist again goes deep at Worlds, the wounded-fuel hypothesis gains another foothold. Third, the patch distance between the two events. If the coming mid-season adjustment is larger than usual, MSI's entire predictive value falls regardless of who wins. This is the most underrated variable in every debate I have followed. TAKEAWAY So what will MSI 2026 tell us about Worlds 2026? If you want a definitive answer, I will not give you one. If you want a correct reading, here it is: do not use MSI to guess who lifts the trophy. Use it to shorten the list. A list of four to six teams genuinely capable of playing into the final week of November. Inside that list, order does not matter. Only presence does. And if MSI 2026 repeats the pattern again, I still will not call it a rule. I will call it a datapoint that is growing, trustworthy enough to believe but not yet sacred enough to worship. Three years from now, when the count reaches six, I will reopen my spreadsheet and check again. That is the only way someone in this profession stays honest with himself: always leave the data the right to overrule you. The summer of 2026 is coming. Someone will lift the MSI trophy. Three months later, we will find out whether we read it right or wrong. The only thing I am certain of is that I will open another empty spreadsheet.