I built a World Cup predictor in 5 weeks as a solo dev - here's what nearly broke me

The 2026 World Cup starts in 6 days. I launched CupRoute last week after about 5 weeks of solo building. Happy to share what the experience was actually like in case it's useful to anyone here.

The hardest part wasn't the code. It was the data model.

The 2026 tournament has 48 teams across 12 groups, with the 8 best third-place finishers advancing to a Round of 32. That means there are 495 possible combinations of which third-place teams go to which knockout bracket slot, and I needed the route tracker to handle all of them correctly.

Getting that logic right while also making sure every prediction scores properly (including teams whose knockout round opponents aren't known yet) took longer than building the entire frontend.

A few other things I didn't expect to be hard:

• Timezone display. The tournament spans US, Canada and Mexico, so games are spread across 6 time zones. Getting every fixture to display in the user's local time, reliably, across devices, without storing anything server-side, was fiddlier than expected.

• The scoring system. I went through 3 iterations before landing on something that felt fair. Final version: 5pts exact score, 3pts correct margin, 1pt correct result, plus bonus points for every team you correctly predict to advance through each knockout round (up to 200pts for calling the winner). Max possible score is 1,640.

• Private leagues. Simpler to build than I expected, but explaining how knockout round predictions work to users who haven't read the docs is an ongoing support challenge.

The product is live, free, and taking predictions now: cuproute.com

Would genuinely love feedback from this community, on the product, the scoring, or anything else. And if anyone here is a football fan, I'd love to know who you're tipping to win it. 👇