What makes a track record real
A record that only contains wins is not a record. It is an advertisement with dates on it.
Every analytics product in this category publishes something it calls proof. Almost none of it survives contact with five simple questions, and the questions are the same whether you are reading a competitor, a Discord, or us.
The five tests
- 1. Is it complete?Every signal, or a selection? If the losses are on a different page, in a smaller font, or absent, the record is a highlight reel and its headline number means nothing.
- 2. Was it timestamped before the outcome?A call published after the fact is not a call. The emission time has to be fixed at emission and never edited afterwards.
- 3. Is the rule written down?If the criteria that fire a signal are not published in advance, the selection can move to fit the results and nobody outside can tell.
- 4. Can somebody else re-derive it?Are the underlying addresses, markets and prices public, so an outsider can recompute the same number and get the same answer? On a public chain there is no excuse for anything less.
- 5. Does it say what following it would have paid?Including the losers, stated per unit staked. “Hit rate” without realised return is the oldest trick in the category.
Why so few pass
Because a complete record is a liability and a curated one is an asset, right up until somebody checks. The dominant products in adjacent categories publish verified winners, scores with no audit, or a label with no performance history attached to it at all. None of that is fraud; it is selection, and selection is enough.
What we do about it
Our calls are timestamped at emission and never backfilled. Every one of them lands on the public ledger, and when a market resolves the same line is marked hit or miss with the realised return of having followed it, stated per $100. Misses render exactly like hits, in the same type, in the same list.
The rule that fires a call is published in full, in advance, on the methodology page, along with the wallet scoring and the public endpoints to recompute both. The ledger is free and needs no account, permanently, because a record you have to pay to audit is not an audit.
And the honest part: the record starts the day it starts. It is thin, and it says so on its face. A long record would be more persuasive and we do not have one — what we have instead is a record that has never been edited, which is the only version of this that is worth anything later.
Questions
How can I tell if a trading track record is real?
Ask five things: is it complete including losses, was each entry timestamped before the outcome, is the selection rule published in advance, can an outsider re-derive it from public data, and does it state the realised return of following it.
What is survivorship bias in a track record?
Showing only the entries that worked. With enough signals, a selected subset will look excellent even if the underlying process has no edge at all.
Why does a short record matter less than an unedited one?
Because length can be manufactured by backfilling and selection, and integrity cannot. A thin record that has never been edited becomes evidence over time; a long curated one never does.