Does MSI Predict Worlds? Six Out of Six and the Trap of a Tiny Sample

May 2026. I reopened the dataset on the last six MSI tournaments and calculated one simple number: if every MSI champion had an equal chance of winning Worlds, the probability that all six would win is 1/64. That figure is enough to make any analyst pause. But I didn't stop there. I placed another number next to it: six. Just six. And the story changed completely.

I've worked with sports forecasting models long enough to know that correct but scarce data is more dangerous than wrong data. A sample of six points on a chart is still an invitation to verify, not yet a rule. Data knows the story before we do, we just arrive late. And this time, the story the data is telling may be prettier than the truth.

MSI — the Mid-Season Invitational — is League of Legends' mid-season tournament, where regional champions meet on an international stage. For years, MSI was seen as a warm-up before Worlds, and the prevailing belief in the community was simple: winning MSI does not guarantee winning Worlds. History backed that belief. Many MSI champions have collapsed at Worlds, sometimes in the group stage.

But over the last three years, the picture has shifted. The organizers expanded MSI — more teams, more slots, more rounds — and according to recent summaries, the new format has aligned MSI more closely with Worlds outcomes. Six recent MSI tournaments, six champions, and all six went on to win Worlds. That's the point the analysis I read emphasizes.

What stands out: all of these champions came from two regions — the LPL (China) and the LCK (Korea). JDG, BLG, Gen.G, T1. Four names, six MSI titles. No team outside those two regions appears on the list.

I cross-checked a few sources and found minor differences in how each summary counts — whether MSI 2026 is included, whether a team that won MSI also won Worlds in the same year. But the big picture is consistent: six out of six. And that very consistency is what makes me suspicious.

A skewed number can retell an entire season. But a chain of six beautiful numbers in a row can retell a story that never happened.

Start with the format. Old MSI had fewer teams, fewer matches, and was sometimes seen as a light event compared to Worlds. Teams came, played a few games, and left. The pressure wasn't enough to forge a lasting tactical identity. Worlds, by contrast, is a long, brutal tournament where teams must adapt across multiple rounds, opponents, and patches.

When MSI expanded, that gap narrowed. More teams means more diverse opponents. More matches means more chances for a strategy to be exposed. More rounds means a team can't rely on a single identity to reach the final. Structurally, the new MSI became a miniature Worlds — close enough for its results to carry predictive weight.

But here's where I want to stop. Close enough to matter is not the same as accurate enough to predict. And there's another explanation for the six-out-of-six streak that few notice: both the LPL and LCK dominate both tournaments. If the strongest teams in the world are concentrated in two regions, then whoever wins MSI is almost certainly from the LPL or LCK. And whoever wins Worlds is too. A correlation appears — but it's the correlation of regional dominance, not of MSI's format.

In other words: MSI doesn't predict Worlds. The two strongest regions predict both. This is a subtle but crucial distinction. It's like noticing that every national math champion also won a provincial math contest — and concluding that winning provincials predicts winning nationals, when in reality the best students simply win both.

Four names stand out: JDG, BLG, Gen.G, T1. These are organizations with training systems, stable rosters, and elite coaching staffs. Over three years, they won MSI. Over the same period, they won Worlds. But the question I want to raise is: did they win Worlds because they won MSI, or did they win both simply because they were the best teams?

The second hypothesis seems simpler. And by Occam's razor, it has greater explanatory power. When you have a strong roster, two tournaments in the same year don't create a causal relationship — they merely reflect the same entity.

Still, I don't want to dismiss MSI's predictive value entirely. There's one thing the new MSI does better than the old one: it identifies the teams in the best form at the right moment. If you use MSI to filter a group of contenders — rather than a single team — accuracy improves considerably. The problem is that we usually look only at the champion, then use the Worlds result to confirm. That's a familiar cognitive bias.

There's one detail in the analysis I think deserves emphasis: MSI champions usually don't win Worlds — that was the old belief. But in the last three years, that belief has been broken. This could be a genuine structural shift, or it could just be three lucky years for the LPL/LCK. We don't know. And that's the crux: we don't know.

Six data points. Three years. Two regions. One new format. Too many variables for too few observations.

If I had to make a conditional prediction, it would look like this: if a team outside the LPL/LCK wins MSI and then wins Worlds, the MSI-predicts-Worlds model gains significant support — because it breaks the LPL/LCK-only hypothesis. If an LPL/LCK team wins MSI but fails at Worlds, the model collapses — because it proves that even within the two strongest regions, MSI guarantees nothing.

And if an LPL/LCK team wins both as usual? The model isn't strengthened, but it isn't refuted either. It just keeps hovering in the gray zone of unverified hypotheses.

That's why I'm watching MSI 2026 with unusual curiosity. Not to see who wins, but to see whether the data structure changes.

One more thing I noticed during the analysis: the LPL/LCK teams dominate both results and adaptation. If you watch recent MSI matches, a common thread emerges: these teams can switch between multiple styles within a single tournament. They can play control, they can play aggressive, they can play late game. That's a skill teams outside the two regions often struggle to match.

But that adaptability is also another explanation for the MSI-Worlds correlation: if a team has enough skill to adapt and win MSI, it has enough skill to adapt and win Worlds. Once again, correlation does not equal causation — it's just a sign of a shared quality.

And as always, I return to the old question: what is the data telling us? Six out of six sounds very strong. But if you placed it next to a sample of, say, sixty MSI tournaments, six out of six would just be a small bright spot. The problem is we don't have sixty. We have six.

In statistics, there's a basic principle: you cannot infer a rule from too small a sample. And six is too small. Even assuming a 50% chance that an MSI champion wins Worlds — a fairly high number — the probability that all six win is about 1.6%. That sounds low, but it doesn't rule out luck. If the probability is 60%, the chance all six win is about 4.7%. At 70%, it's about 11.8%. In other words, a six-out-of-six streak isn't enough to distinguish a genuine rule from a lucky run with moderate probability.

This is why I find the way esports media presents such statistics frustrating. They give the number without the context. They say six out of six as if it were proof, when in reality it's just an observation. The noise of the crowd, it turns out, is also data. And in this case, the noise is louder than the signal.

But I don't want to end on pure skepticism. There's one positive thing worth noting: the new MSI format has made the tournament more serious. Teams no longer treat MSI as a vacation. They prepare, they study opponents, they invest. Structurally, that's good for esports. An MSI with predictive meaning is better than an MSI with none.

And if this correlation continues for a few more years — if MSI 2026 and 2027 also confirm the model — then eventually we'll have enough data to call it a rule. But we're not there yet. We're in year three of a trend.

In economics, there's a concept called survivorship bias. When you look only at the teams that succeeded, you easily conclude that their path was the only path to success. But there may be many other teams that won MSI and did not win Worlds. They exist in history, but they don't appear in the list of six recent winners. And precisely because they don't appear, we tend to forget them.

If you're a smaller team, this matters. It means you shouldn't read six out of six as a promise. It's just an observation about a specific group of teams, in a specific period, under a specific format. Change any one of those three variables, and the story can shift entirely.

I've followed League of Legends long enough to remember the years when MSI predicted nothing. I remember teams that won MSI and then collapsed at Worlds. I remember the old beliefs still etched into the community. And I remember the feeling when one of my forecasting models was wrong — it wasn't pleasant, but it was necessary.

So what will MSI 2026 tell us? Perhaps not a definitive answer. Perhaps just one more data point on the chart. But sometimes, one more data point is all we need to get closer to the truth.

Two million euros is not an answer, it's a question. Six out of six is the same. And the right question here is no longer Does MSI predict Worlds, but: how much data do we need to know for sure?

That's the question I'll carry into next season.