How useful is a Monte Carlo model for predicting TI2026?

November 18, 2026 4 answers
Asked by 8bitButtonMinder

A model simulates The International 2026 about a million times using Glicko-2 team ratings. What can those results tell viewers about the likely bracket and tournament winner, and what should we be cautious about?

Answers

GBCaudioMux

Think of it as a map of possibilities, not a prediction of one exact bracket. If a team wins 20% of the simulations, the model is saying it wins roughly one in five times under the assumptions it uses. It can also estimate how often teams reach each round.

The million runs make the simulated percentages less noisy, but they don't fix bad assumptions. The team ratings, the way those ratings become match win probabilities, and the tournament format all matter.

8bitButtonMinder:

So a low chance of winning doesn't necessarily mean a team is unlikely to make a deep run?

mechLoadoutJ7:

Right. Reaching a late round and winning the whole event are different probabilities. A team can be reasonably likely to do the former while still having a small title chance.

Reply to this answer
4K!SpawnVoucher

I'd pay close attention to updates rather than treating the first set of numbers as fixed. New results can change the Glicko-2 ratings, and newly released tournament details could change the paths teams have to play through.

When a percentage moves, check what changed in the inputs. A shift in the model isn't automatically evidence that a team suddenly got better.

Reply to this answer
QTEbuffer~Thimble

The bracket is the fun part for me. Two teams with similar ratings can have different routes to the title, so their chances might differ even before considering individual matchups. Look at round-by-round odds as well as the outright winner table.

Reply to this answer
LANparty_Quicksand

Check the labels too. One team in the initial model had an uncertain regional classification and was placed in Western Europe. That doesn't tell us whether its simulated match odds are affected, but it is worth checking when comparing results by region.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions