How we prove the record is real
A prediction shown after the race proves nothing. ShaTinMore locks its analysis of every runner in every race on the morning of the meeting, before race 1, and scores it after the results — and both halves can be verified without trusting us. This page describes exactly what happens; the record page holds the sealed files and the running score.
The chain, step by step
- Rate. On the morning of every meeting (06:00 HKT, with a 09:00 backup run), an automated job rates every declared runner in every race. The rating (form+speed v0.2) uses past form and speed figures only — runs dated strictly before that day, and no odds or market data of any kind.
- Seal. The full ranking is written as one JSON file to write-once storage. The store rejects any attempt to overwrite an existing file — there is no edit path, including for us. A card is only sealed once its first race is under 30 hours away, and never within 30 minutes of it.
- Timestamp. The file's SHA-256 fingerprint is submitted to OpenTimestamps, which commits it to the Bitcoin blockchain within a few hours. Backdating the file would require rewriting Bitcoin — the timestamp proves those exact bytes existed before the first race.
- Publish. The sealed picks are posted to our public Telegram channel at the moment of sealing, so there is a second public trail with its own timestamps.
- Score. After the results, the same job scores the sealed file against the market favourite at HKJC's own win dividends and writes a score file next to it — also write-once. Good days and bad days are published the same way; nothing is edited or removed.
Verify it yourself
- Open the record page and download a meeting's sealed file (
.json) and its timestamp proof (.ots). - Hash the file with any SHA-256 tool (e.g.
sha256sum 2026-09-27.json) and compare it with the fingerprint shown next to the meeting. - With the open-source OpenTimestamps client:
ots upgrade 2026-09-27.json.ots, thenots verify 2026-09-27.json.ots. The proof completes a few hours after sealing, once the calendar's commitment is in a Bitcoin block. - Open the
.jsonitself:sealedAtandfirstRaceare both in the file, so the gap between sealing and the first race is part of what is timestamped.
What the seal does — and doesn't — prove
The seal proves one thing only: the predictions existed, in exactly this form, before the race was run. It says nothing about whether they are any good.
Whether they are any good is measured on the record page, against the hardest honest benchmark we know: the market favourite, at the same HKJC dividends. The model is a deliberately simple, fully described baseline; every change to it is a new version and is scored separately, so a new model can never borrow an old model's record. One meeting is noise — judge the record over hundreds of races.
Frequently asked questions
What does “sealed” mean?
On the morning of every meeting, hours before race 1, our model's ranking of every runner in every race is written as one JSON file to write-once storage, and the file's SHA-256 fingerprint is anchored in the Bitcoin blockchain via OpenTimestamps. From that moment the file cannot be changed or backdated — by anyone, including us.
Could ShaTinMore edit a prediction after the race?
Not without being caught. The storage rejects overwrites, and the Bitcoin timestamp proves the exact bytes existed before the first race. If we replaced a file, its SHA-256 would no longer match the timestamped fingerprint, and anyone re-running the verification would see it.
Do you delete bad days?
No. Every sealed meeting is scored after the results and stays on the record page — losing days exactly like winning days. The score files are write-once too.
How do I verify a sealed file myself?
Download the meeting's .json and .ots files from the record page. Hash the .json with any SHA-256 tool and compare it with the fingerprint shown. Then run `ots upgrade <date>.json.ots` and `ots verify <date>.json.ots` with the open-source OpenTimestamps client to confirm the Bitcoin anchoring.
Is this betting advice?
No. ShaTinMore is a free, independent statistics tool built on HKJC's public results. Nothing here is betting advice and no outcome is guaranteed — the record exists precisely so you can see how often the model is wrong. If you bet, do so only where it is legal for you, only at the legal betting age in your jurisdiction (18+), and only with what you can afford.