MSI 6/6 and Worlds: Six Rows of Data Being Read as a Law

The infographic appeared on social media within two hours of the Worlds final. Two columns, MSI on the left, Worlds on the right, six rows, six green checkmarks, and a one-line caption: "MSI is now Worlds' dress rehearsal." I was in Busan. I reopened the brackets from the last three MSI editions, and the first thing I did was count where those six rows came from.

I am not a person who distrusts everything. I am a person who once spent a full week re-measuring a metric simply because it looked too good. My job is valuing players for a club, and in that job, a beautiful data pattern is always the most suspicious thing before it becomes the most trustworthy thing.

Here is what I found: the six rows are real, but they do not say what the crowd thinks they say.

Context: MSI changed shape

MSI is the Mid-Season Invitational, run by Riot since 2026. For nearly a decade it was a small, short event, often six to twelve teams, sometimes just a round robin and a final. Its status among hardcore fans was clear: a place to see big teams experiment, not a place to learn who would win Worlds.

That shifted in 2026. The MSI format expanded, more teams attended, and the bracket structure moved closer to the rhythm of a real championship. In parallel, Worlds 2026 replaced the group stage with a Swiss stage before the knockout bracket.

Both changes happened at almost the same time. Most analysis ignores this: when you compare MSI and Worlds results across years, you are comparing two tournaments that changed format simultaneously. Any correlation you find is contaminated by the change itself.

I met this exact kind of noise once, in another sport. In 2026, as a student, I analysed South Korea's 2-0 win over Germany in Kazan. Germany's PPDA was 5.8, meaning extremely aggressive pressing, and many analysts used that figure to criticise South Korea's approach. I split the data into fifteen-minute windows and found Germany's pressing system broke down after Kim Young-gwon came on. Three weeks later FIFA published a report confirming exactly what I had written, but I had already been attacked hard for it. A beautiful metric is never a conclusion. It is an invitation to check again.

So when the 6/6 graphic spread, I did what I always do: opened the raw data and counted.

Core: dissecting six rows

First, confirm what is true. Since 2026, MSI champions have been JDG, Gen.G and BLG. Worlds champions in the same window came from T1 and the LPL. Neither event left the hands of the LPL and LCK. The four names dominating discussion are JDG, BLG, Gen.G and T1, and all four deserve it: stable rosters, long-tenured coaching staffs, and top-tier budgets.

But counting at team level, the number is no longer six out of six. JDG won MSI 2026. T1 won Worlds 2026. If you count by team, that is a complete failure of translation: the mid-season champion was not the end-of-season champion.

So where do the six rows come from? From the definition of "translation." If you count by region rather than by team, the comparison becomes almost true by definition. Worlds has not escaped the LCK and LPL since 2026. Pick a set of MSI editions whose champions are LCK or LPL, compare with a set of Worlds editions whose champions are also LCK or LPL, and you will always get a very high rate. You are not measuring MSI's predictive power. You are measuring something everyone already knows: these two regions have dominated this game for over a decade.

This is what I call "re-counting what is already known and calling it a discovery." It is not arithmetically wrong. It is simply meaningless for forecasting.

To test whether any real signal exists, I split the data into three layers.

The first is the regional layer. Here MSI and Worlds correlate strongly, but that correlation existed before MSI expanded. In other words, the new format did not create the regional correlation; it only made it easier to see. People are crediting the new format for something the discipline already had.

The second is the team layer. Here MSI identifies contenders better than it predicts champions. A team that goes deep at MSI usually has fundamentals good enough to go deep at Worlds, but that is a relationship between "going deep" and "going deep," not between "winning" and "winning." JDG in 2026 is the clearest case: they won MSI with what was considered the strongest roster in the field, then lost in the Worlds semifinal to T1 itself.

The third is bracket position. This is the least discussed and most analytically valuable layer. At MSI, finishing second or third means very different things depending on where and how you lost. But when data is compressed into a single "champion / not champion" column, all of that information disappears.

I did this work at club level once, with midfielder Lee Kang-in at Mallorca. He ranked in the top ten in La Liga for chances created per 90 minutes at 2.8. I proposed signing him for eight million euros. The board rejected it, arguing he could not demonstrate defensive ability. Six months later he shone and helped Mallorca survive, while my club finished eighth. I wrote a fifteen-page internal report that blamed no individual and showed our process had read the wrong kind of data. A transfer fee is the number one person is willing to pay. Real value is the number data does not have to negotiate.

In MSI's case, the wrongly read variable is "champion." It is a crude binary, and it erases everything worth analysing.

The biggest issue: the patch window between MSI and Worlds

There is one variable the 6/6 graphic never mentions, and it is the most important variable in the whole story: the gap between MSI and Worlds.

MSI sits mid-year, Worlds at year end. In between, the game changes. Patches reshuffle entire champion classes, jungle systems, towers, objectives. A team that wins by playing around a specific champion pool can become ordinary after two patches.

This is why football metrics cannot be imported wholesale into esports. In football, the rules are fixed across decades. In esports, the rules are rewritten every few weeks. An xG-style metric in League of Legends would have to be normalised per patch, and once you normalise per patch, your sample immediately fragments.

So which metrics replace it? As a data person, I use four groups, localised for this discipline.

One: pick and ban rates by champion class. This measures a team's ability to adapt to the patch, not its strength. A team with high pick rates on meta champions but a low win rate with them is misreading the patch.

Two: gold difference at fifteen minutes. This is the closest esports analogue to "early-game quality," and it is far more stable than kill counts.

Three: objective control rate, especially elemental drakes and the Herald. This reflects coordinated team play, which individual scoreboards never show.

Four: win rate by side selection, blue versus red. This is the most underrated metric in every discussion I have ever read. An MSI champion with a pronounced side-selection advantage may face an entirely different problem at Worlds.

None of these four groups are pretty. None of them fit into an infographic. That is exactly why they are rarely used.

What 214 empty-stadium matches taught me

In 2026, when the pandemic forced leagues behind closed doors, I was a graduate student and used the rare opportunity to track 214 matches in the Bundesliga and K League 1 from May to August. Home win rate in the Bundesliga fell from 43.2 percent to 37.8 percent, and average goals rose from 2.79 to 3.12.

People called it a natural experiment. I call it an opportunity to measure luck. 214 empty-stadium matches taught me: home advantage is data, not just atmosphere.

I raise this because it connects directly to MSI. When an external condition changes, you get a chance to separate essence from environment. MSI's 2026 format expansion is exactly such a change. And notably, in the three years since, no team outside the LPL and LCK has won MSI.

That can be read two ways. First, the new format made MSI harsher, reinforcing the two strongest regions. Second, the new format has not yet produced any structural change, and three years is far too short to conclude.

I lean toward the second reading, with a caveat. I was once attacked for doubting PPDA, and FIFA later confirmed what I said. That taught me that the sceptical reading is sometimes right — but it also taught me that sometimes the sceptical reading is right in part and wrong in the whole. Here, I do not yet have enough data to say which.

The transfer market: where MSI actually has value

If you ask what MSI is worth, my answer is not about predicting Worlds. It is about the transfer market.

MSI is the first international stage of the year and the only pre-Worlds setting where Korean and Chinese teams meet directly in knockout play. For a market administrator like me, it is the most important evaluation window of the first half.

But for that same reason, MSI produces the most mispricing. Three reasons.

Sample size. A player plays only a handful of MSI games. At that size, an individual can stand out simply because opponents were weak, or because his team peaked at the right moment. This is survivorship bias: we remember who shone at MSI and forget who shone at MSI and then vanished.

Patch distance. A player who excels on champions favoured in the MSI patch can lose much of his value when the patch shifts. Clubs that buy the patch pay for it.

Team effect. A player can look outstanding because four teammates create space for him. Move him, and the space disappears.

Because of these three, I never price a player off an international event alone. I normalise across leagues and I always state where the number's limits are. Don't trust the standings, ask the expected-value data. Standings tell the past, data tells the future. But in esports I must add a clause: data only tells the future within the same patch.

Contrarian angle: 6/6 is a language trap

The six-out-of-six rate is not a discovery. It is a narrative device, built by choosing a starting point and a definition that produce the prettiest possible rate.

Want to prove MSI champions predict Worlds champions? Start from 2026, when the MSI format changed, and compare by region. You get a very high rate. Now keep 2026 and compare by team. JDG won MSI 2026, T1 won Worlds 2026. The rate collapses instantly.

Or move the start to 2026. A European team won MSI 2026 and a Chinese team won Worlds 2026. Another failed translation, this time even at regional level.

The data is not wrong. Three years is not enough to call anything a law, and a rate constructed by choosing a definition is not a rate — it is an argument.

I said something similar once before. In 2026, as a first-year student in Busan, I collected data on Asan Mugunghwa in K League 2. They topped the table but averaged only 1.02 expected goals per match, below Busan IPark's 1.48. They relied on penalties: six in six matches. I wrote on my blog that they would slide. They finished fourth and lost in the playoffs. That post reached two thousand views, enormous for a student blog, and it taught me something I still use: a team scoring penalties six times in six matches is not playing football, it is playing a lottery.

MSI's 6/6 is the same in nature. It is pretty, it spreads easily, and it makes people forget that behind it sit three seasons — roughly fifteen to twenty knockout matches — and in a discipline where one patch can change everything, fifteen knockout matches is a very small sample.

What is actually worth tracking

If I had to pick three signals for the coming MSI, none of them would be who wins.

First, the list of knockout-stage teams not from the LPL or LCK. If one reaches the semifinals, the 6/6 rate everyone shares will have to be rewritten.

Second, the second- and third-place teams and where they lost. Information lives in bracket position, not final ranking. A team that finishes third after losing a five-game series to the eventual champion is more interesting than a team that finishes second after a sweep.

Third, the window between MSI and Worlds — specifically the patches inside it. If the signature champion pools of the MSI champions are heavily adjusted before Worlds, every MSI-based forecast loses value within weeks.

None of these produce a green checkmark on social media. They only produce a spreadsheet. That is the kind of spreadsheet twelve years of watching this discipline have taught me to trust.

The view from Vietnam: expectation, viewing hours and a mis-measured gap

Vietnamese fans watch MSI under very specific conditions. Time-zone differences put most North American and European matches in the early morning or late night in Vietnam. Having followed Vietnamese forums and commentary channels across many seasons, I find it interesting that those hours create a distinct collective memory. Viewers in Vietnam remember matches not by date but by "that night." Chronology compresses, and when chronology compresses, random patterns start to look like real ones.

I am not saying Vietnamese fans read the game wrong. I am saying the conditions of receiving data shape how conclusions form, and that is an effect anyone working with data must account for.

On the competitive gap, I think one point is frequently missed in domestic discussion. The LPL and LCK are strong not only in rosters. They are strong in development systems, in the number of high-level matches a player can log in a year, and in their ability to turn a patch change into an advantage within weeks. The gap with the VCS is not located in a few individuals. It is located in the number of exposures to peak-level competitive conditions.

That also means that when a VCS team achieves a milestone internationally, its value lies elsewhere — not in whether they won a title, but in the extra layer of experience they now hold, experience that becomes data for the following season.

What I take away from this graphic

I started writing from a student blog with two thousand views, and I learned that data does not care who you are, only whether you read it correctly.

The 6/6 graphic will be shared many more times, before every MSI and every Worlds, for years. It will keep being true until someone changes the definition. The job of a data person is not to oppose it. The job is to state the sample size, state the definition, and let the reader decide how much to trust it.

What I will do at the coming MSI is simpler. I will reopen my spreadsheet before the final and log gold difference at fifteen minutes, pick and ban rates by champion class, objective control rate, and win rate by side selection. Then I will wait. Three months later, when the patch has changed, I will compare.

If what I logged still holds after the patch shifts, I will start to believe it. If not, I will add a line to the limitations section of the report and move on to next season.

Data does not reward people who speak loudly. It only rewards those who are patient enough to count a second time.