Demo vs Live Trading: What Paper Trading Proves
Paper trading validates strategy logic — not fills, slippage, or your own psychology. How to graduate safely.
What paper trading is good for #
Paper (demo) trading runs a strategy against live market data with simulated fills. It answers the logic questions cheaply: does the signal fire when it should, do position sizes compute correctly, do risk limits trigger, does the agent behave through volatility — all without risking capital.
Every major venue supports this workflow: Binance maintains a dedicated spot testnet, and both dYdX and Hyperliquid document testnet environments for practicing perps mechanics before mainnet.
What paper trading cannot replicate #
Simulated fills are optimistic. Paper engines typically assume mid-price or top-of-book fills with no market impact; live trading pays the spread, suffers slippage on size, and gets partial fills in fast markets. Fees and funding are often approximated or ignored.
The bigger gap is psychological: clicking through a 20% paper drawdown is trivial; watching real capital do it triggers intervention — widening stops, doubling down, disabling the agent — which invalidates the backtest you just ran.
A graduation checklist #
Promote a strategy to live capital only when: it is profitable net of realistic fees and slippage assumptions over at least dozens of paper trades; its worst paper drawdown is a fraction of what you could tolerate live; risk limits (per-trade loss, daily loss, kill switch) are enforced by the platform, not by willpower; and you start at minimum size — paper size means nothing until live fills confirm the edge.
JW3.ai's paper mode exists for exactly this pipeline: validate agent behavior in simulation, then deploy the same agent live with identical risk limits and human-approval gates.
Frequently asked questions #
Is paper trading worth it?
Yes for validating logic, sizing math, and risk-limit behavior. No as proof of profitability — simulated fills ignore spread, slippage, and market impact.
Why do strategies that work in demo fail live?
Optimistic fills, ignored fees and funding, and trader psychology: live drawdowns trigger manual interference that the paper test never modeled.
How should I size my first live trades?
Minimum size with hard platform-enforced loss limits. Scale only after live fills confirm the edge paper trading suggested.