引導式測驗 · 中等 等級
交易測驗的強化學習
解釋強化學習交易環境、獎勵設計、驗證、歷史模擬對現實世界表現的限制。
問題 1 的 8
In an RL trading environment, what is the reward function?
預覽每個問題8 問題
當您參加測驗時,就會出現答案和解釋。
- In an RL trading environment, what is the reward function?
- The signal that tells the agent how an action scored under the chosen objective
- A calendar of exchange closures supplied to the agent
- A promise that the learned policy will be profitable
- The record identifying who owns the brokerage account
- Why include slippage and transaction costs in simulation?
- They can be omitted because simulated orders always execute at the midpoint
- They are already reflected fully in historical price changes
- They do not affect returns when the policy is trained on past market data
- They can reduce or alter the returns available from trading
- Which condition describes market nonstationarity?
- Prices and trading conditions remain fixed over time
- Market regimes, participants, rules and liquidity can change
- The strategy trades one instrument only
- The agent receives no observations from the environment
- How should training and evaluation periods be arranged?
- Randomize all dates so future observations influence training
- Select hyperparameters using the final test period
- Use chronological or walk-forward tests with an untouched period
- Report only the training-period return
- What does a high simulated return establish?
- That the strategy will be profitable in live markets
- Only that the policy performed under the tested environment and assumptions
- That the policy cannot be overfit
- That every relevant cost was included
- Why compare an RL agent with a simple benchmark?
- To make the RL model appear more sophisticated
- To ensure the benchmark always underperforms
- To see whether added complexity improves performance over a reference
- To avoid specifying the agent’s objective
- What can paper trading reveal?
- All future market regimes and execution conditions
- Absence of regulatory or operational risk
- A guarantee that live returns will match simulation
- Implementation, timing or execution differences in a sandbox
- What does the guide recommend before committing capital to an RL strategy?
- Independent validation, controls and review beyond a simulated score
- A strong score on a single historical test window
- High returns across seeds in the same simulator only
- A brief paper-trading run during calm market conditions