What paper trading can — and cannot — prove

Paper trading is the default for every new CryptoMetric AI bot. It runs the decision pipeline on public market data but intercepts the order and creates a simulated fill instead of sending a real order to an exchange. This is the appropriate place to learn how a template behaves and whether the surrounding automation works as expected.

What paper mode can show

A supervised paper run can reveal whether a template produces the expected BUY, SELL and HOLD pattern; whether risk checks refuse unsuitable entries; whether stops and targets are recorded; and whether the operator notices alerts and can halt the bot. CryptoMetric AI records simulated trades and charges a flat 0.1% fee in its paper-order model, making that explicit cost visible.

It can also expose practical mistakes: the wrong symbol or timeframe, limits that are too loose, too little candle history, excessive turnover, or a strategy that conflicts with the market regime. These are valuable findings even when the simulated profit-and-loss number is poor.

What it cannot reproduce

The paper executor uses synthetic fills. It does not reproduce the full order book, queue position, partial fills, changing liquidity, real network latency, exchange rejection, account restrictions or every fee tier. A simulated limit order is treated as immediately filled at its stated price. Those simplifications can make a paper result materially different from a live result.

Simulation also cannot prove that a historical pattern will continue. Repeatedly adjusting a strategy after seeing its results can fit noise rather than a durable relationship. A favourable paper period may reflect one market regime and say little about the next one.

A gate, not a certificate

CryptoMetric AI can require a bot to build its own paper history before a live transition is even considered. Clearing that gate is only a precaution. It is not a certificate of profitability, an approval of the strategy, or a recommendation to use real funds. Operational and regulatory gates remain separate, and live trading can remain unavailable regardless of paper results.

A useful review routine

1. Keep the configuration fixed long enough to observe it across different conditions.
2. Review rejected signals and operational failures, not only profitable simulated trades.
3. Compare assumed fills and fees with conditions actually available at the exchange.
4. Record why a setting changed and restart the evaluation after a material change.
5. Assume that future live outcomes can be worse than the paper record.