引導式測驗 · 中等 等級
AI 投資組合優化測驗
測試估計風險、投資組合約束和樣本外評估,包括為什麼機器學習不能保證更好地實現分配。
問題 1 的 8
What does a portfolio optimizer do once its inputs and constraints are specified?
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。
- What does a portfolio optimizer do once its inputs and constraints are specified?
- Guarantees the portfolio’s realized return
- Chooses asset weights that meet an objective such as balancing expected return and risk
- Predicts with certainty which assets will rise next month
- Removes the need to estimate risk and expected returns
- Why can mean-variance weights change sharply after small input revisions?
- The objective ignores all risk information
- The covariance matrix is fixed by the optimizer and cannot be estimated
- The weights are chosen independently of the inputs
- The solution can be sensitive to noisy estimates of expected returns and covariances
- In a portfolio workflow, where may ML contribute?
- Estimating return or risk inputs, or regularizing how those estimates feed into weights
- Guaranteeing a globally optimal realized return
- Replacing all portfolio constraints with historical averages
- Eliminating the need for a benchmark
- Why compare an optimized portfolio with a simple 1/N baseline?
- To remove the need to account for turnover
- To prove that all assets have identical risk
- To check whether estimation complexity improves held-out results over a transparent benchmark
- To ensure the optimizer uses equal weights as its final answer
- Which validation design is appropriate when tuning on financial time series?
- Preserve time order so later observations do not inform earlier predictions
- Randomly shuffle every month before splitting train and test data
- Tune on the full sample and report the same sample as out of sample
- Choose the test period after seeing which period gives the best result
- What should a cost-aware portfolio test include?
- A gross return series without the rebalance schedule
- The number of parameters, without portfolio weights
- Turnover, transaction costs, and relevant liquidity or market-impact assumptions
- Only the model’s training loss
- What do position and leverage limits contribute to an optimization?
- They guarantee that all risk is diversified away
- They replace estimates of expected return
- They ensure every asset receives a positive weight
- They constrain the set of permitted allocations and can limit concentration or borrowing
- What does a good historical optimization result establish?
- That future inputs will match the estimation period
- Performance under the tested data, objective, constraints, and cost assumptions—not guaranteed future results
- That the same weights are optimal for every investor
- That losses cannot exceed the historical sample maximum