Is "Worlds T1" real? I checked 12 years of Worlds and MSI with Elo ratings

Riccardo Gallotti · 13 October 2026

TL;DR: On Elo ratings Gen.G is the favourite at 40% and T1 is only at 6%. But at Worlds T1 has won 27 more games than its rating predicts (in 157 games), while at MSI it plays exactly as rated. With that "Worlds bonus", T1 goes to 46%.

I am a researcher and data scientist. I would also be an avid gamer and would love to gain rank with my Orianna in Wild Rift (yes, I know, it's not real League). But lately my favourite automaton has been Claude, the AI made by Anthropic. Among the things we analyse together is the real League of Legends, whose pros I started following in 2023. That was the year of Faker's return to winning, and that autumn T1 won Worlds. T1 then won the next two Worlds as well, each time arriving, somehow, as the underdog. Now Worlds is about to start again, and T1 arrives, once more, as the underdog.

So who is likely to win this time? I gave every professional team an Elo rating built on all its games since 2011. Then I had a computer play this year's Worlds a hundred thousand times in its real format.

On ratings alone, the favourite is Gen.G, the best team on paper for most of the last four years. It wins about two Worlds in five. T1? Only fifth, with about one chance in sixteen.

But the same ratings gave T1 long odds before each of its last three titles. Three in a row from those odds had about one chance in 35,000 — so I checked where the ratings get T1 wrong. Out of 157 games at Worlds, T1 has won 27 more than its ratings predicted. No other team comes close. At MSI, against the same opponents, it instead plays exactly as rated. The T1 that turns up at Worlds would beat the T1 of the rest of the year seven times in ten. Call it a Worlds bonus.

It is known that it shows most against Chinese opponents. In 2023, as T1 closed out JD Gaming in the semi-final, the caster Drakos shouted: "It is not LPL versus LCK! It is T1 versus the LPL! And they like those odds!" He had a point. At Worlds T1 has played 13 best-of-three or best-of-five series against Chinese teams, counting the 2025 play-in against Invictus Gaming, and it started as the underdog in most of them. It won all thirteen: about one chance in 220,000. The other Korean teams, against the same opposition, won 7 of their 19 series, against about 10 expected.

Give T1 that bonus (the dashed row in the table) and the picture flips. T1 wins almost one Worlds in two, and Gen.G drops to about one in four.

Two warnings. First, a rating knows results and nothing about players. It ignores transfers, injuries and the draft. Second, the bonus rests on just nine Worlds, so it is far from certain. T1 has played ten, but I left out its first title in 2013, when the ratings were still learning how strong each league was (with it, T1 would beat its rating that year too).

The final in New York on 14 November settles it. Either way, following T1 these past three years taught me that unpredictability is the fun part, and that the underdog stories are the ones we remember.

I did this research with Claude. It wrote and ran the code while I chose the question and checked the numbers, and we wrote this text together. The code and data are open, and the full article, which also compares tournament formats (double elimination, the football World Cup, NBA and league rules), is here: https://bigiftruebits.substack.com/p/worlds-t1