Short answer: a paper-traded track record can show that a method behaves consistently and that calls were posted before the move. It cannot show what you would have earned, because paper has no slippage, no partial fills and no market impact. Trust one only if the entries can be checked against the exchange candle, every loss is listed, fees are modelled, and posts carry timestamps you cannot edit.
Plenty of signal channels publish paper results without saying so, or say it in small print. This page explains how to read such a record, and what ours looks like. For a wider comparison of channels, see the honest review of free crypto signals on Telegram.
What paper results can prove
Paper trading is not worthless. It can show:
- Rule consistency. The strategy followed its own rules for hundreds of trades, not only when the market was kind.
- Timing. Calls were posted before the price moved, if the post timestamps are intact.
- Frequency and exposure. How many trades, how long held, how many open at once.
- A fair comparison. Two methods tested the same way, on the same data.
These are real properties. They are the reason to paper trade a method before risking money on it.
What paper results cannot prove
Five gaps are built into the format:
- No slippage. A market order into a fast move fills worse than the quote. Paper records the quote.
- No partial fills. On thin pairs, a real order may fill in pieces at different prices or not at all.
- No market impact. Your order moves the book. A paper order never does.
- Friction left out. Fees, funding and spread are often ignored or approximated.
- No emotion. Holding through a drawdown on paper costs nothing. With your money it costs discipline.
There is also a structural bias. A paper record usually covers a period the operator chose to publish, from a method they chose to keep. Methods that failed quietly were not published. This is survivorship, and it affects a paper record more than a real one, because dropping a paper strategy costs nothing.
What to demand before believing one
Use this as a list, and count the answers.
| Demand | What a pass looks like | What a fail looks like | |---|---|---| | Timestamped before the move | Post time precedes the price move, edit history not hiding changes | Entry posted after price is already there | | Entry checkable | The entry price sits inside the exchange candle of that minute | A price the market never traded | | Losses listed | Closes posted as loudly as opens, counts reconcile | More opens than closes | | Fees modelled | Result shown net of fees, with the fee rate stated | Win rate only, or "before fees" without a net | | Enough trades | 100+ closed trades, across more than one market condition | A dozen trades, one good week | | Period stated | Start date, end date, and what was excluded | A rolling window with no dates |
The second row is the one people skip. Open a candle chart at the minute of the post and look. If the stated entry is outside that candle's range, the record is not describing a trade that could have happened. A record can also be built on an index that is shifted a few candles from the intended one, so a single checked trade is not proof, but ten that fail is.
Red flags
- Screenshots in place of a ledger.
- A win rate with no average win, average loss or fee figure.
- "Backtested" results presented as if they were traded.
- No way to see the trades that were closed at a loss.
- A result that is only ever quoted over the best window.
- Pressure to subscribe before you have seen a month of closes.
Why fees can flip a win rate into a loss
A trade costs money to enter and to exit. Over hundreds of trades the cost is large, and it is paid win or lose. A method that wins most of the time with small gains and loses occasionally with large ones can have a high hit rate and a negative result. Add fees and the gap widens. The arithmetic is worked in why a 60% win rate still loses money.
Darwin Lab's own real-money run is an example. Across 2,556 trades the win rate was 60% before fees and 52% net, and the result over May 22, 2026 – Jun 26, 2026 was -$149, largely from fees. A channel that quoted only the gross figure would have looked fine. That is exactly why the net figure is in the opening of this page and not hidden behind a link.
How Darwin Lab's record is structured
The facts, in order:
- 10 April 2026: real money on Binance Futures mainnet.
- Through June 2026: the real-money run, ending with a negative net, largely from fees. The account was emptied.
- Since June 2026: signals are on paper while the system is rebuilt. Paper entries here carry all the limits listed above.
- Public record: every trade and the running totals are at /track-record/, and the raw figures are at /api/stats.json.
The free Telegram channel gives pair and direction only. There is no entry, stop or size, so it is not a tradeable setup and cannot be followed trade by trade. Entry, stop and targets are part of the paid tier. The free channel is capped at 24 paper signals a day, at least 1h apart.
That makes our record two different things: a real-money stretch that can be argued with and a paper stretch that cannot prove much on its own. Reading them separately is part of reading them honestly.
Reading a paper record in practice
A routine that takes about twenty minutes:
- Pick five closed trades at random, not the ones the channel highlights.
- For each, find the timestamp of the post and compare the entry with the candle.
- Confirm the close was posted, and whether the stop or target matched the original post.
- Add up fees using your own exchange's rate.
- Decide what position size you could have used, and what a bad streak of five trades in a row would cost you.
If the record survives that, it has told you something modest and useful: the method behaves as described. The rest, what you would make with your money at your size, is still open. Size from your stop distance, not from a leverage number, as laid out in risk management for crypto trading.
Futures trading can lose more than you deposit, and past results, paper or real, do not predict future ones.