Day Two of the World Cup, Picked
After a perfect opening day, here is what the model likes for Canada vs Bosnia and USA vs Paraguay.
Day one could not have gone better. Two games, and the model went a clean six-for-six. Every winner, every total, every both-teams-to-score call landed, and it even nailed the Mexico scoreline on the nose at 2-0. I’m not going to pretend that happens every day. It won’t. But it’s a nice way to start, and the whole point of this is to keep score in the open, so there it is on the board.
On to day two. Two more games, both with the host nation involved again. Here’s exactly what we’re playing before we get into the why.
Today’s picks
Canada vs Bosnia & Herzegovina
Winner: Canada, draw no bet. Take the draw out of the equation and the model makes Canada about an 80% shot. The market has them nearer 70%. We think Bosnia is weaker than the price suggests.
Total: Under 2.5 goals. This is the big one. The model says under lands about 89% of the time. The market has it closer to 57%. That’s a gap you don’t see often.
Both teams to score: No. Same story. We’re at about 86%, the market about 53%.
USA vs Paraguay
Winner: USA. We make them about 54%, the market about 50%, and that’s before you remember they’re at home, which our model doesn’t take into account for the World Cup.
Total: Over 2.5 goals. We lean over at about 51% against a market near 41%.
Both teams to score: Yes. About 52% for us versus 45% for the market.
Notice the two games pull in opposite directions on goals. One screams under, the other leans over. More on why below.
Model update: the ratings moved, the trophy didn’t
Before the picks, a quick word on what yesterday’s results did under the hood, because this is the part I find genuinely fun.
Every result reshuffles the team ratings. Win and your number goes up. Lose and it drops. After day one:
Mexico beat South Africa and climbed +31 points.
South Korea beat Czechia and jumped +40.
South Africa fell −59.
Czechia dropped −63.
(The losers move more than the winners here for a simple reason. The system is less sure about them, so each result counts for more.)
Now here’s the interesting part. You’d think shaking up four teams’ ratings would shuffle the “who wins the World Cup” board. It doesn’t. Argentina is still out in front, with Austria, France, and Belgium behind, exactly the same picture as yesterday. And the reason is blunt: none of these four were ever going to win the thing. They’re all sitting at a fraction of a percent to lift the trophy, so nudging their ratings up or down can’t move a race they were never in.
Where it does matter is the only race these teams are actually in, which is getting out of the group. And there it moved a lot:
- South Korea: 86% to 90% to advance.
- Czechia: 88% to 83%.
- Mexico: 93% to 94%
So that’s the lesson the model is quietly teaching. A single group-stage result is almost meaningless to the trophy, but it can swing a team’s knockout hopes by several points overnight.
Game one: Canada vs Bosnia, and the most lopsided number on the board
History first, because it’s a good one. This is the first men’s World Cup match ever played on Canadian soil, at BMO Field in Toronto, with Canada as co-host. For a country that had never won or even drawn a World Cup game across its two previous appearances, that’s a real moment.
And the model likes them, sort of. It makes Canada the most likely single winner at around 47%, but here’s the catch: it also thinks a draw is wildly live at around 41%. This projects as a tight, tense, low-event game, the kind of nervy opener a host plays when the whole country expects a win. The most likely scoreline the model spits out is 0-0, with a Canada 1-0 close behind.
That’s why the winner pick is Canada with the draw taken out. We don’t have an edge backing Canada to win outright. The market actually prices them a touch higher than we do. But strip out the draw, and the model is far more confident in Canada than the market is, because it rates Bosnia notably weaker (we have them at about 12% to win, the market says about 22%). So: back Canada, get your money back if it’s level, and let the draw, which we think is underpriced anyway, be the escape hatch.
The standout, though, is the goals. The model wants Under 2.5 at nearly 89%, against a market sitting around 57%. That is one of the biggest gaps I’ve seen it produce. Both teams to score: No is the same bet wearing a different hat, at about 86% versus 53%. Two cautious sides, a host with nerves, a low-scoring grind: that’s the picture, and it’s a strong one. One honest caveat, same as always: the model doesn’t account for Canada’s home crowd, so if anything it’s *underrating* the hosts, not over.
Game two: USA vs Paraguay, the open one
Down in Inglewood at SoFi Stadium, the USA open their tournament at home against Paraguay, and this game is the mirror image of the Toronto one. The model sees it opening up.
We make the USA about 54% to win, a touch above the market’s 50%, and again that’s before you credit a partisan home crowd the model ignores. Paraguay are stubborn and well-organized, the type to make you earn it, but the model still leans toward the hosts and toward goals: Over 2.5 at about 51% versus a market near 41%, and both teams to score: Yes at about 52% versus 45%. Expected scoreline lands around 2-1 USA, most likely single result 1-1.
So where Canada vs Bosnia is a bet on tension and a clean sheet, USA vs Paraguay is a bet on an end-to-end night. I like that the model isn’t just defaulting to “home team, under, no goals” across the board. It’s reading the two matchups differently, and the goal markets are where it’s planting its flag most firmly.
The ledger rolls on
That’s day two: a draw-no-bet on a nervy host, two big goal calls in opposite directions, and a coin-flip-plus on the USA at home. Every pick and every price is already logged, same as yesterday, so by tomorrow morning we’ll know exactly how it went, win or lose, with the receipts out.
Six-for-six was a dream start. Now we find out if it was the model or the variance. Enjoy the football.