Paper trading bots: forward-test before risking real money
Paper trading runs your bot live on real market data with fake money. It is the essential bridge between a backtest — which can lie — and risking real capital. A paper-trading phase catches the bugs, slippage surprises, and overfit strategies that look fine in history but fall apart in real time. Here is how to use it well.
What paper trading is
Paper trading (also called forward testing or demo trading) runs your bot against the live market in real time, placing simulated orders with no real money at stake. Unlike a backtest, which replays history, paper trading happens in the present — so it tests your strategy and your code against conditions that haven't happened yet. Many exchanges offer a testnet or demo account for exactly this. See our how to paper trade guide for setup.
What paper trading catches that backtests miss
- Code bugs — an off-by-one in your data handling, a misfiring order, a timezone error. Backtests often hide these; live execution exposes them.
- Real slippage and latency — your live fills won't match the perfect prices a backtest assumed, revealing whether the edge survives real slippage.
- Overfit strategies — a system that was curve-fit to history often starts losing immediately on fresh, unseen data, which paper trading reveals before it costs you money.
- Operational reality — API outages, rate limits, reconnections — the unglamorous things that break live bots.
Its limits
A simulated order usually assumes you got filled at the quoted price, but in reality a real order moves the book and may fill worse, especially in size or on thin pairs. Paper trading also can't replicate the psychology of real money — but for a bot that runs automatically, that matters far less than it does for a discretionary trader. Treat paper results as a best case, not a guarantee.
Running a paper-trading phase
After a strategy passes the backtester, run it on paper for long enough to see varied market conditions — ideally weeks, not days, covering at least one volatile stretch. Compare the live results to the backtest: large divergence means slippage, bugs, or overfitting. Only when paper results roughly match the backtest, and the bot has run without operational failures, should you go live — and then with the smallest position size you can, scaling up slowly. Avoid the common bot mistakes of skipping this step.
Frequently asked questions
What is paper trading?
Paper trading runs a strategy live on real market data with simulated (fake) money. It forward-tests your bot against current conditions, bridging the gap between a backtest and risking real capital.
Why is paper trading important for bots?
It catches code bugs, real slippage and latency, operational failures like API outages, and overfit strategies — problems that backtests often hide. It reveals whether the edge survives in real time before real money is at risk.
How long should I paper trade?
Long enough to experience varied market conditions — typically weeks rather than days, including at least one volatile stretch. The goal is to see the strategy handle different regimes and confirm the bot runs without operational failures.
Are paper trading results accurate?
They are optimistic. Simulated fills usually assume the quoted price, while real orders can fill worse, especially in size or on thin pairs. Treat paper results as a best case, then go live small and scale up slowly.