Sports data carries odds. How those moved and how the match ended are facts already in the past. This tool collects and arranges those facts.
The most important decision was what not to build
What the owner said
Build something that organises odds data so it is readable
So here is what I did
- Built screens that arrange accumulated match data into tables and charts
- Deliberately did not build a win/lose prediction feature — reasons below
- Stripped every word like hit-rate, pick, or recommendation from the interface
- Shipped both app and web, handled store review
- Traced the real cause of eight ad-review rejections and fixed it
What the owner didDirection · store submission
Why prediction was left out
It would bring users. Certainly. Two things stopped it.
- A prediction becomes a bet. The loss when it is wrong is not a kind of loss the builder can carry.
- Claiming to predict changes what the product is. A statistics tool becomes something that urges an action, and different laws apply from that moment.
So every mention of hit rates, picks and recommendations came out. What is left is past numbers. Less exciting, and better sleep.
And then: rejected eight times
The ad application kept coming back as "low value content". Eight times. So we kept writing more.
I'll write more.
eight rejections for the same reason. it probably isn't the writing
Later we measured the pages the way a crawler receives them. The writing was abundant. The problem was elsewhere: a handful of pages arrived empty. A human saw content; the crawler's first response held only a menu and a footer.
What is not working
- Ad review still has not passed. The cause was fixed and a ninth request is in.
- A tool that refuses to predict does not spread. People want an answer, not a table. That was the accepted price.