第 49 / 209 页
非代码面试题
显示 20 / 4169 道匹配题目
答题状态：未尝试未正确已正确
ID题目领域难度题型进度权限
1902Vendor Lag Alignment 2A vendor distributes a dataset with a fixed 3-business-day lag. If your model key is the Monday file and you trade at the open, what is the first valid trading session that may use that observation?统计中等derivation未尝试免费1903Vendor Lag Alignment 3A vendor distributes a dataset with a fixed 1-business-day lag. If your model key is the Wednesday signal and you trade at the open, what is the first valid trading session that may use that observation?统计中等数值题未尝试面试订阅1904Vendor Lag Alignment 4A vendor distributes a dataset with a fixed 5-business-day lag. If your model key is the Friday accounting snapshot and you trade at the close, what is the first valid trading session that may use that observation?统计中等数值题未尝试面试订阅1905Vendor Lag Alignment 5A vendor distributes a dataset with a fixed 2-business-day lag. If your model key is the Thursday borrow-fee file and you trade at the open, what is the first valid trading session that may use that observation?统计困难essay未尝试面试订阅1906Minimum Embargo Length 1A label is defined as the forward return over the next 5 trading days. If you split train and test windows by time, what is the minimum embargo gap needed to avoid leaking label information across the boundary?统计简单essay未尝试免费1911False Launch Probability 1You test 40 independent noise strategies. A strategy is launched only if it passes an in-sample screen at level 1/20 and then passes a second independent validation at level 1/10. Under the global null, what is the probability that at least one noise strategy still gets launched?统计简单essay未尝试免费1916Look-Ahead Audit Scenario 1A daily factor is z-scored using the full-sample mean and standard deviation before the backtest is run. What is the main problem?统计简单essay未尝试免费1917Look-Ahead Audit Scenario 2A backtest ranks stocks each month using the latest fully restated accounting data available today, even for years long ago. What is wrong with that setup?统计简单essay未尝试免费1918Look-Ahead Audit Scenario 3A research notebook chooses the top decile each day using today's close and then records the same day's close-to-close return as PnL. What is the leak?统计中等essay未尝试面试订阅1919Look-Ahead Audit Scenario 4A training set includes observations whose labels use returns that overlap with the future test window. What should be added to the split procedure?统计中等数值题未尝试面试订阅1920Look-Ahead Audit Scenario 5A PM wants to backfill today's index membership list into the whole historical universe before evaluating a factor. Why is that unsafe?统计困难essay未尝试面试订阅1921Required Observation Count 1A research lead uses the approximation n >= K * sigma 2 / delta 2 with K = 7.84, signal standard deviation 24 bp, and target detectable mean shift 3 bp. What is the minimum integer sample size?统计简单数值题未尝试免费1923Required Observation Count 3A research lead uses the approximation n >= K * sigma 2 / delta 2 with K = 9, signal standard deviation 30 bp, and target detectable mean shift 4 bp. What is the minimum integer sample size?统计中等数值题未尝试免费1926Minimum Detectable Edge 1Using the same approximation n >= K * sigma 2 / delta 2, suppose K = 7.84, sigma = 25 bp, and you can collect only n = 400 observations. What is the smallest detectable mean edge delta, in basis points?统计简单essay未尝试免费1928Minimum Detectable Edge 3Using the same approximation n >= K * sigma 2 / delta 2, suppose K = 9, sigma = 30 bp, and you can collect only n = 900 observations. What is the smallest detectable mean edge delta, in basis points?统计中等数值题未尝试免费1931Effective Sample Under Serial Dependence 1A return series has 240 observations and lag-1 autocorrelation 1/5. Using the heuristic n eff = n * (1-rho)/(1+rho), what is the effective sample size?统计简单essay未尝试免费1936Paired Design Observation Ratio 1Two strategies are tested on the same names and dates, giving correlation 1/5 between their noise terms. Relative to an unpaired comparison with the same marginal variance, by what factor does the paired design reduce required sample size?统计简单数值题未尝试免费1941Power Score Design Choice 1Compare two research designs using the score delta * sqrt(n) / sigma. Design A has delta=4 bp, n=225, sigma=2 bp. Design B has delta=3 bp, n=400, sigma=1 bp. Which design has the larger score?统计简单数值题未尝试免费1946Log Utility Versus Impact Budget 1A signal is strong enough that the PM scores larger size through a log benefit but pays quadratic impact. The desk chooses a scalar exposure x > -1 to maximize F(x) = 3 ln(1+x) - 2 x 2. What exposure x maximizes F?数学简单数值题未尝试免费1947Log Utility Versus Impact Budget 2A desk chooses a participation tilt with diminishing marginal benefit and quadratic implementation drag. The desk chooses a scalar exposure x > -1 to maximize F(x) = 4 ln(1+x) - 1 x 2. What exposure x maximizes F?数学简单数值题未尝试免费