Estonia–Georgia

0–1

HT 0–1

Kick-off: Saturday, 5 September 2020 at 16:00
Status: Match Finished
Venue: A. Le Coq Arena, Tallinn

What the model said, and what happened

Prediction correct

This verdict goes into the public record unchanged; we do not filter out the ones we get wrong. The model's track record

Referee

Donatas Rumsas, Lithuania→

There's no card data for this referee from before this match yet. We don't invent an average; it will appear here once the data exists.

Model prediction

Why this prediction?

The model puts the most likely outcome at 44.6%: “Georgia to win”. Second comes “a draw” at 28.8%.
The highest double chance value is 73.3% (X2 — draw or Georgia win).
These values were computed from 1,480 past matches.

This is a prediction, not a guarantee; a probability is not a certainty.

These sentences are built by a fixed template; every number in them comes straight from the model's computation, none invented while the text was written.

Estonia to win26.7%

Draw28.8%

Georgia to win44.6%

When the model has said 40–45% in the past, 2,399 of 5,395 predictions were correct (44.5%). Accuracy record →

Double chance

1X — Estonia win or draw

What these probabilities rest on

Expected goals comes out of a team's attack weighed against its opponent's defence. A strong attack is damped when it meets a strong defence — which is why comparing the ratings on their own is misleading.

Estonia

Attack: Estonia score below the league average (39% fewer).

Opponent's defence: Georgia's defence is close to the league average (0.92×), so it doesn't move the expectation much.

Georgia

Attack: Georgia score below the league average (20% fewer).

Opponent's defence: Estonia concede 40% more than the league average — the defence is the weakest link in this match.

Week 1 of the season. This prediction was computed from previous seasons. Summer transfers, budget changes and squad rebuilds aren't in the ratings yet — the model can only see that a team has been rebuilt once new results come in.

We measured this and we're not hiding it: in the opening week of past seasons the model came in 3.3 points below the accuracy it claimed (n=220) — for the rest of the season it beats its own claim. The tendency points this way, though it is not proof on its own: in the opening week it slightly over-calls home wins (said 42.7%, got 39.5%) and slightly under-calls draws (said 26.1%, got 28.2%).

This sample is small and not statistically established, which is why we do not correct the model for it. The correction we tried (pulling probabilities toward the league average) didn't win on holdout, so it wasn't added. Read opening-week predictions with a little more caution.