How to verify a track record
You do not need to audit a whole history. Confirm one past call end to end and you learn the thing that matters most: whether the record can be checked at all.
This is the lesson the rest of the primer builds toward. Reading a signal is literacy; verifying one is the skill. The four steps below run from the cheapest, fastest check to the single most decisive one, and you can apply them to any service that publishes a record.
Verify in four steps
1. Start with the denominator
Find the total number of calls and confirm the losing ones are included. A win rate quoted with no count behind it is a headline, not evidence. With the worked example the figure is stated as 67.5 percent across 308 same-session calls; the 308 is the part you are checking for, and the percentage is meaningless without it. More on this in how to read a win rate.
2. Insist on a continuous run
Look for an unbroken period rather than a hand-picked week. A record that shows only its best five sessions is hiding the other forty-five. A genuine record states its period — here, 2026 year to date — and does not skip the bad stretches inside it.
3. Find the named, independent reviewer
Check that a named outside party has examined the underlying record. A platform leaderboard is not an audit and a happy-customer quote is not a review.
4. Confirm one call against its timestamp
This is the decisive step, and the one most services cannot survive. Pick one historical call and line its published fields up against the public, on-chain receipt that was filed for it. Since that receipt was filed ahead of the result, a clean match shows those fields were set in stone before anyone knew the outcome. A single confirmed call is worth more than any number of celebratory screenshots. The mechanism that makes this possible looks like this:
The call below is invented to show the procedure. The steps are exactly what you would run on a genuine published call.
- Write down the five fields as published. Suppose the call reads: long an index ETF,
entry 412.80,target 414.20,stop 412.10,grade B,sent 14:32:05 UTC. - Rebuild the fingerprint. The service runs those exact fields, in a fixed order, through SHA-256 — a one-way function that turns any input into a single fixed-length fingerprint. The same fields always give the same fingerprint; one changed digit gives a completely different one.
- Open the public receipt. The OpenTimestamps receipt filed alongside the call names the Bitcoin block its fingerprint was committed to. Check that the fingerprint you rebuilt is the same one recorded in that receipt.
- Check the clock. Find the timestamp of that Bitcoin block. If it was mined before the trade resolved, then the call — prices and grade together — demonstrably predates its own outcome. That is the entire proof.
Now try to defeat it. Suppose someone shifted the stop from 412.10 to 412.40 once the candle had turned against them. Re-running step 2 on the altered fields yields a different fingerprint, which no longer lines up with the receipt from step 3, and the edit gives itself away. A receipt is therefore stronger evidence than any image precisely because it breaks the moment a single field is altered.
The takeaway: steps 1 to 3 take a couple of minutes and clear away most of the field; step 4 is the one no amount of marketing can fake. When a service survives step 4, what you hold is a history you can pull apart and test for yourself, not one you are asked to accept on trust. The same procedure, in quick screening form, lives on the verification checklist.