The FA Cup semi-final draw happened, and within approximately forty-five seconds, every football account on the internet had declared Manchester City the winners. City got Southampton. Chelsea got Leeds. The consensus was immediate, unanimous, and — if history is any guide — almost certainly wrong about at least one of these ties.

I ran the numbers. You won't like them.

Let's start with the big one. Since the Premier League era began in 1992, there have been 23 FA Cup semi-finals where one side was from a lower division. The higher-ranked team won 15 of those. That's a 65.2% success rate, which sounds comfortable until you realise it means that roughly one in three times, the "easy draw" team gets sent home with their tracksuit bottoms metaphorically round their ankles. One in three. That's not a rounding error. That's a pattern.

Southampton, specifically, have a semi-final record that should make City's analytics department quietly close their laptops and stare at the wall. Saints have reached the FA Cup semi-final eight times since 1992. They've won four of them. A 50% conversion rate in semi-finals is, for context, higher than Arsenal's (44.4% from nine) and identical to Manchester United's from their last eight. Southampton in a cup semi-final are not Southampton in the league. They are a different, stranger, more chaotic organism entirely.

But here's the stat I keep coming back to. In the last decade, when a team from outside the top flight has reached the FA Cup semi-final, they have scored first in 57% of those matches. More than half the time, the underdog draws first blood. They don't always win, obviously. But they score first. They make it uncomfortable. They create the fifteen-minute window where the favourites' fans go very quiet and start typing things like "it's the history of the" before deleting the tweet.

Now, Chelsea versus Leeds. This one is fascinating for entirely different reasons. Since 2000, FA Cup semi-finals between two clubs with an active historical rivalry have produced an average of 3.8 goals per match, compared to 2.1 for semi-finals without significant needle. Hatred, it turns out, is a goal catalyst. The data is unambiguous on this. When two sets of fans genuinely cannot stand each other, defensive discipline collapses by the 60th minute. Players start doing things that look brave on the highlight reel and idiotic on the expected goals model.

Leeds, for their part, have won their last three matches against sides currently in the Premier League across all competitions. Chelsea have lost two of their last four against lower-division opposition in cup ties. I'm not saying Leeds will win. I'm saying that anyone who looked at this draw and immediately pencilled in a City-Chelsea final is operating on vibes rather than evidence, and the evidence has a 34.8% chance of making them look foolish.

The broader point is this: the FA Cup semi-final is not the quarter-final. It is a different competition played in a different emotional register. Wembley semi-finals, specifically, have produced more upsets per fixture than any other round since the neutral venue format was reintroduced. The big pitch. The big occasion. The weird acoustics that make 30,000 Southampton fans sound like 60,000. It does things to supposedly superior teams.

So by all means, assume City will stroll through. Assume Chelsea's quality will tell. Assume the draw was kind and the final is already decided. The numbers suggest that approximately one third of you are about to have a very bad afternoon in late April, and I for one will be here with the spreadsheet open, waiting.

Actually, I've already formatted the "I told you so" column. It's colour-coded and everything.