What paper trading is
Paper trading means trading with pretend money at real prices. You open positions, set stops and targets, close trades and keep a running profit and loss, but nothing is on the line. People use the same term for stocks, forex, options and crypto, and it is also called demo trading, simulated trading or, when someone writes it as one word, papertrading.
The idea is old. Before simulators existed, traders wrote their intended trades in a notebook and checked later how they would have done. Today an exchange or broker does the bookkeeping for you, or you run your own script that records hypothetical orders.
Paper trading is used for three different jobs, and it helps to know which one you are doing:
- Learning the platform. Where the buttons are, how a stop order behaves, how margin is shown.
- Rehearsing a plan. Following your own written rules for a few weeks to see whether you can actually follow them.
- Testing a strategy. Collecting a forward record that you hope says something about future results.
The first two are things paper trading does well. The third is where most of the confusion lives.
Paper trading versus backtesting
A backtest takes historical candles and asks what your rules would have done. A paper trade takes the live market as it unfolds and records what your rules do right now.
The difference sounds small and it is not. Past data is already complete, so it is easy to let information from after the decision leak into the decision. A common version is an indicator calculation that is shifted by a few candles, so a trade appears to enter at a price that was only known later. Another is choosing the strategy settings after looking at which ones worked on that same history. Both produce a beautiful equity curve that mostly measures how well you fitted the past.
Forward paper trading removes the first problem completely, because the future has not happened when you decide. It does not fix the second one if you keep tweaking the rules after each bad week, and it does not fix the cost problems below. But a forward record is typically harder to flatter than a backtest, which is why it is a useful second check after one.
What a good paper account has to simulate
A simulator is only as honest as its assumptions. Four matter most.
Fees. Every real trade pays a fee on the way in and again on the way out. On futures, a market order pays a taker fee each side. If the simulator charges nothing, every strategy looks better than it is, and the strategies hit hardest are the ones that trade often with small targets. A full round trip should be charged on every trade, winners included. More on this in risk management in crypto trading.
Slippage. The price you see is not always the price you get. A market order eats into the order book, and in a fast candle the fill can be noticeably worse than the last price. Thin pairs are worse than liquid ones. A simulator that fills you at the exact price on your screen assumes a market that does not exist.
Fill assumptions. Did your limit order really fill, or did price merely touch your level without trading through it? In real markets you are often behind a queue of other orders. Paper systems that count every touch as a fill overstate how many trades you would have gotten, and usually the good ones.
Funding and leverage costs. On perpetual futures you pay or receive a funding rate while a position is open, which can add up on long holds. See funding rate explained and crypto leverage explained.
A worked example
Say you paper trade a simple rule on a futures pair for one month. The rule opens a trade when a signal fires, with a stop 1% below entry and a target 1.5% above. Your tracker shows 40 trades, 22 winners and 18 losers, and a total gain of 6% on the notional traded.
That looks good. Now apply costs honestly. Say a full round trip of taker fees plus a bit of slippage costs 0.15% of the position per trade. Over 40 trades that is 40 x 0.15% = 6%. The entire gain is gone, and it was gone before you counted funding.
Nothing about the strategy changed. Only the assumptions did. The first number answered "did price move my way?" and the second answered "did I make money?" Those are different questions, and the gap between them is widest for short-term, small-target strategies.
Now look at the sample. Forty trades is a small number. If the true chance of a winner were exactly 50%, you would still see 22 or more winners out of 40 fairly regularly by luck alone. A good-looking month is weak evidence by itself. A rough comparison helps: run the same exits on random entries and see whether your signal beats them. If entries chosen by a coin flip do just as well with the same stops and targets, the signal is not adding much.
What paper trading cannot show you
Even a clean, fee-inclusive, forward paper record leaves gaps.
Emotions. Following a rule is easy when the loss is imaginary. Real money changes behavior: people close winners early, widen stops, skip the trade after two losses, or double down. A paper account tells you what the rules do, not what you will do.
Small samples. Trading outcomes are noisy. A strategy can have no edge and still look excellent for weeks. Or it can have a real edge and look terrible for weeks. Without enough trades across different conditions, you cannot separate skill from luck.
Market regimes. A strategy that thrives in a trending month may bleed in a range. If your paper period happened to be one kind of market, the result only describes that kind. Market regimes explained covers why this matters so much.
Market impact and liquidity. At small size you do not move the price. At larger size you might. Paper trading is blind to that.
Execution failures. API errors, rejected orders, exchange outages and delayed stops happen in live trading. A simulator rarely reproduces them.
The biggest beginner mistake
The most common mistake is treating a paper profit as a green light. Someone paper trades for three weeks in a rising market, sees a positive balance, and moves to real money with high leverage and a bigger size than the practice account had.
Three things then go wrong at once. The costs they ignored start to count. The market changes character. And the emotional weight of real losses shows up for the first time, on a size they never practiced. The paper result was not a lie, but it answered a smaller question than the one they thought it answered.
A better habit is to treat paper trading as a filter for bad ideas, not a certificate for good ones. If a strategy cannot make money on paper after realistic costs, it will not make money live. If it can, that is only permission to test it further, at small size, with a hard cap on what you can lose. The stop and exit side of this is covered in stop-loss and take-profit explained.
How to paper trade well
- Write the rules down before you start, including entry, stop, target and size. Changing them mid-run resets the sample.
- Record every trade, including the ones you skipped and why.
- Charge a full round-trip fee on every trade and add a slippage allowance that is on the pessimistic side.
- Pick one timeframe and run long enough to cover more than one kind of market.
- Judge the result by expectancy per trade after costs, not by win rate or a single total.
- Compare against a dumb baseline, such as random entries with the same stops and targets.
- Decide in advance what result would make you drop the idea.
How a systematic engine handles it
A systematic engine has an advantage here: it has no emotions to protect, so paper trading tests the rules rather than the person. It also has a hazard, which is that software can be generous to itself without anyone noticing. A fill assumption, a missing fee or an off-by-a-few-candles index can quietly flatter every trade.
Darwin Lab is a self-evolving genetic-algorithm trading bot built for Binance Futures. It traded real money there from April to June 2026 and has run its signals on paper since June 2026 while it rebuilds. Its paper engine charges a full round-trip taker fee on every trade. Measured net of fees, its published paper signals have not yet shown an edge over random entries with the same stops and targets. That is the honest reading of the record so far, and the full fee-inclusive record is public at /track-record/, so you can check it instead of taking our word for it.
A negative or flat result is still useful. It is exactly what a forward, fee-inclusive paper record is supposed to be able to tell you. For background on what a signal does and does not contain, see what is a trading signal.
How Darwin Lab works | Join the free channel
This article is educational and not financial advice. Trading futures carries a high risk of loss, and a paper result does not predict a real one.