Paper trading, explained

Paper trading is placing orders that behave like real orders, against prices that came from a real market, using money that does not exist. Nothing is bought. Nothing is sold. No broker is involved, and no order ever leaves the simulation.

That sounds like it should be useless. It is not — but what it is good for is narrower than most people assume, and knowing the boundary is most of the value.

What the simulation actually gives you

Three things, reliably.

A place to be wrong cheaply. The first strategy you write will do something you did not intend. It will buy at the top of every rally, or hold a position for eleven months because the exit condition can never be true, or place four hundred orders in a day because you compared the wrong two numbers. Finding that out costs nothing here.

Mechanical fluency. Order types, position sizing, what a fill looks like, what happens to your cash when a limit order sits unfilled — these are learnable by doing, and doing them once for real is an expensive lesson.

A falsifiable record. A strategy that you ran and can point at is a different object from a strategy you thought about. It produced trades. You can read them.

Where the simulation ends

This is the part worth being precise about, because a simulator that flatters you is worse than no simulator at all.

Your orders do not move the price. In a real market, a large order consumes the book and the price moves against you before you are filled. In a simulation your order is matched against a book that does not know you exist. For small sizes on liquid instruments the difference is small. For anything else it is the whole game.

Prices are not live. CodeBull's quotes are cached snapshots of a third-party feed, refreshed on an interval — not a streaming tick-by-tick feed from an exchange. A strategy whose edge depends on reacting within seconds is not something this environment can honestly evaluate.

There is no emotional cost. The single largest difference between a simulated drawdown and a real one is that you can walk away from the first. Discipline under a real 20% loss is not a thing paper trading can teach, and anyone who tells you otherwise is selling something.

Costs are approximations. Spreads, fees and slippage in a simulation are modelled, and a model is a simplification by definition. A strategy whose returns are thinner than its modelled costs is not a strategy — it is a rounding error.

So what is it for?

It is for the question "does this idea do what I think it does?" — and that question is worth answering before any other.

A moving-average crossover either generates the trades you expected or it does not. A mean-reversion rule either triggers on the days you thought or it sits idle for a month. That is a question about your logic, not about the market, and a simulation answers it exactly.

It is not for the question "will this make money?" Nothing answers that in advance, and a simulated result is one of the least reliable ways to guess.

Starting on CodeBull

Everything on CodeBull is simulated. There is no real-money mode to graduate into, no funding step, and no way to accidentally send a real order — the platform has no connection to a broker at all. That is deliberate: it is an educational environment, not a trading account with training wheels.

Two ways in:

1. Place orders yourself. Market, limit and stop orders against the simulated book. Watch what a fill does to your cash and your position.
2. Write a strategy. Python, or visual blocks if you would rather not. The engine runs it and produces a report you can read.

The second is where it gets interesting, and it is the subject of your first strategy bot. Once you have a report in front of you, what a backtest can — and can't — tell you is about not over-reading it.

None of this is financial advice, and nothing here predicts what any strategy will do with real money. It is a description of a simulator and what a simulator can honestly show you.