Between a promising backtest and risking real money sits paper trading: running your strategy live against real market data with simulated orders. It feels like the responsible step, and it is — but most people misunderstand what it actually proves. Done naively, a few weeks of green paper-trading results give false confidence. Done right, paper trading catches the specific failures that backtests are blind to. The difference is knowing what question it answers. None of this is investment advice.
What paper trading actually validates
The real value of paper trading isn't confirming your strategy is profitable. It's confirming your system works in the messy conditions a backtest never simulates.
A backtest runs against clean, complete historical data, all at once, with perfect hindsight. Paper trading runs forward in real time, which surfaces an entirely different class of bugs: does your code handle a data feed that's delayed or has gaps? Does it react correctly to a market open, a halt, or a missing bar? Does your signal actually fire when you think it does, on data that arrives bar by bar rather than as a complete history? These integration failures are invisible in a backtest and common in production.
The most valuable thing paper trading does is run your strategy on data it has never seen and was never tuned on. If your backtest was secretly using future information — a classic look-ahead bug — paper trading exposes it, because the future genuinely isn't available yet. A strategy that paper-trades far worse than it backtested almost always has this problem.
What it can't tell you
Paper trading has two hard limits, and ignoring them is how people get burned.
First, simulated fills are optimistic. Most paper-trading environments fill your orders at or near the quoted price, without modeling the slippage, partial fills, and queue position you'd hit with real money — especially in less liquid instruments. So your paper P&L is systematically better than your live P&L will be. Paper trading validates that your orders would have been placed, not that they'd have filled where you assumed.
Second, the sample is too short to prove an edge. A few weeks or months of paper results is a tiny sample, dominated by whatever market regime happened to occur. A trend-following strategy will look brilliant in a trending month and broken in a choppy one, and neither tells you about its long-run expectancy. You cannot distinguish skill from luck in a handful of trades.
Resist the urge to treat a profitable paper-trading month as proof. With a small number of trades, randomness dominates. Paper trading confirms your system runs correctly; it does not confirm your edge is real. Those are different claims, and conflating them is expensive.
How to run it so it means something
Treat paper trading as a system test, not a profit test. A sound approach:
Run it forward on a genuinely out-of-sample period — data that came after everything you used to design and tune the strategy. Watch for divergence between paper and backtest performance: a large gap is a red flag pointing at look-ahead bias or unrealistic backtest assumptions, not bad luck. Log everything — every signal, every order, every fill — and reconcile it against what your backtest would have done on the same bars, so you catch logic discrepancies. And when you eventually go live, start with a small fraction of your intended size, because the optimistic-fill gap means real results will be worse than paper.
The mindset that serves you: paper trading is where you discover your code is wrong, not where you discover your strategy is right. If you come out of it having fixed three integration bugs and confirmed your strategy behaves identically forward as it did in the backtest, it did its job — regardless of the P&L.
Paper trading is a powerful tool aimed at the wrong target by most people. Point it at your execution system and your assumptions instead of your P&L, and it becomes the step that catches the bugs and biases that would otherwise cost you real money.
Originally published at pickuma.com. Subscribe to the RSS or follow @pickuma.bsky.social for new reviews.
Top comments (0)