Test the process before the money
Paper trading submits simulated orders without risking live capital. A backtest applies a strategy to historical data. An allocation preview only shows a proposal. Each answers a different question.
Paper trading is useful for finding missing permissions, duplicate orders, bad quantities and rules that produce unexpected decisions. A profitable simulation does not prove the same strategy will make money live.
Set up a trial you can learn from
- Use a separate simulation environment
Confirm that the account and order destination are explicitly paper trading. Alpaca, for example, documents separate paper credentials and an endpoint. Do not keep live credentials in a trial that does not need them.
- Match your intended budget
Use the capital, assets and position sizes you expect to trade. Keep a written starting plan and model version. Changing the rules after every loss makes the result impossible to interpret.
- Record decisions and execution
For each review, save its time, input data, intended action, order status and reason. Include no-trade decisions. Compare what the agent said with what the simulator recorded.
- Exercise the stopping controls
Check pausing, cancelling an open order and recovering from a disconnected session. Verify that an uncertain response does not cause the same order to be sent twice.
What a simulator leaves out
Alpaca’s paper trading documentation explains differences in fills, liquidity and data. Live trading adds execution uncertainty and the pressure of actual losses. A short trial in one market environment provides limited evidence.
Set a review date in advance. Count rule violations and execution errors alongside returns. Do not switch to live money simply because the first week was positive.
Is Agency’s first look paper trading?
No. Agency’s first look proposes a percentage allocation and explains the choices. It does not simulate a continuing account or establish a performance history. Agency currently has no paper trading or backtesting mode.
Use the first look to check whether a plan was understood. Use a dedicated simulation environment if you need repeated simulated orders. For the eventual results, follow the same measurement checklist.