Backtesting vs Paper Trading vs Live
Published Jul 26, 2026 · Updated Oct 16, 2026
Backtesting, paper trading, and live trading validate different things: a backtest replays historical market data under a simulation model; paper trading runs a strategy on current market data with virtual funds; live trading submits real orders through an exchange API connection or a Hyperliquid trading account. A live run shows the fills, fees, and funding that actually occurred, but cannot guarantee future results. Before carrying a configuration between stages, check that the template supports that stage and that its parameters mean the same thing.
What does each stage actually validate?
Each stage answers a different question. A backtest tells you whether the logic ever worked, paper trading tells you whether it still works today, and live tells you whether it works at your size with your fees. None of them substitutes for another.
| Stage | Market data | Money | Validates | Can’t catch |
|---|---|---|---|---|
| Backtest | Historical replay | None | Strategy logic, parameter ranges, past drawdown | Today’s regime, real fills, order-book depth |
| Paper trading | Current market data | Simulated (engine-managed) | Signal timing, behavior under paper execution rules | Actual fills, your fee tier, funding paid |
| Live | Current market data | Real, through your selected connection | Observed execution and costs at your actual size | Future market conditions, all failure scenarios, larger-size performance |
A strong backtest with no paper run is untested in the present; a good paper run with no backtest has no evidence across market phases. Use both, in that order.
How do you promote a strategy from backtest to live?
Start with a supported template and record its parameters. Backtest across the market periods relevant to your hypothesis, then observe a paper run across different conditions. Check each stage’s available settings before considering a separate live trader with limited exposure; there is no automatic proof that a configuration will behave identically.
Backtest first
Choose a backtest-capable template and run a backtest in Strategy → Strategy backtest, recording the market source, date range, initial equity, and strategy parameters. The Web uses the default execution model; your own Agent can invoke supported CLI commands. Review the drawdown and execution assumptions before continuing. Templates that cannot be backtested yet, such as the Copy templates, start on paper.
Paper trade the survivor
Check the parameters and create a paper strategy. It uses current market data and simulated execution, with a balance recorded by the engine and no exchange account required. Observe different conditions rather than treating short-term profit as sufficient validation.
Go live small
Add a live trading account (an exchange API connection with trade permission only, or a Hyperliquid trading account), copy the paper strategy onto it, and start below your target size. Scale only after live behavior matches paper. Full steps: From Paper to Live.
Why do backtest, paper, and live results still differ?
Data windows, starting positions, simulated fills, fees, funding, and signal timing can all change results. A gap may reflect different assumptions or an actual fault; compare settings, orders, positions, and diagnostics before deciding. See Paper, Backtest, and Live Differences.
Plans include live and paper slots; usage beyond them is charged hourly in credits. The estimated cost at creation and Membership are authoritative. The leaderboard lists both paper and live strategies; read the mode, measurement window, and risk alongside ROI.