Proof of Take
Hot takes disappear.
Receipts don't.
Anyone can claim they called it. Proof of Take makes the claim checkable: the content of your prediction is hashed and timestamped onchain before the result is known.
Write
A Take is one falsifiable claim with a confidence number, supporting evidence, a countercase and a date it can be graded. Vague takes can't be stamped meaningfully, so the form pushes you toward specificity.
Hash
Zaply builds a canonical string from the Take — a fixed field order with normalised whitespace and punctuation — then takes a SHA-256 of it. The same Take always produces the same hash, on any machine, forever.
Sign
Your wallet signs a Solana transaction carrying that hash in a memo, formatted as ZAPLY:v1:<hash>. No custom smart contract, no token, no custody. You pay only the network fee.
Prove
The verification page recomputes the hash from the stored content and compares it to the recorded value and to the memo actually written onchain. If someone edited the Take afterwards, the hashes stop matching and the page says so.
What a proof shows
- This exact text existed at this time.
- It was signed by this wallet.
- It hasn't been edited since — any change breaks the hash.
- The confidence number was set before the outcome was known.
What it doesn't show
- That the prediction is correct. A stamped Take can still be badly wrong.
- That the author is an expert. It's a record, not a credential.
- Ownership of anything. There is no token and no financial interest involved.
- That every Take was stamped — creators choose which ones to put on the record.
- Accuracy (70%)
- 56 pts
- Calibration (20%)
- 14.5 pts
- Sample size (10%)
- 8.9 pts
Sample creator from Zaply's demo dataset.
Signal Score
How the score works
Accuracy (70%) is your hit rate across resolved Takes. PUSH and VOID results are excluded entirely — nothing was really predicted, so they neither help nor hurt.
Calibration (20%) asks whether your confidence was honest. It's a Brier-style measure: saying 90% and being wrong costs far more than saying 60% and being wrong. Inflating every number is actively punished.
Sample size (10%) ramps with the square root of your resolved count, maxing out at 25. A 2-for-2 record shouldn't outrank a long history, and this is what stops it.
Below 5 resolved Takes the score is shown as provisional, because it doesn't mean much yet.
Signal Score describes a creator's resolved Takes on Zaply. It is a Zaply reputation metric, not an objective measure of expertise, and it says nothing about work published anywhere else.
Put your next call on the record.
Three onchain stamps a month on the free plan. No wallet needed until you use one.
Start free