Project 3 Exercises

See solutions.md for complete solutions and the guide for the default run instructions. Exercises P3-01, P3-03, P3-04, and P3-06 can be answered using the inputs and historical tables provided in this book. P3-02 requires the historical primary_ledger.csv, and the historical calculation in P3-05 requires French inputs; neither is generated by the default constructed branch. If these materials are unavailable, read the complete historical reference solutions; there is no need to start the market-data branch.

P3-01 (Manual ledger calculation) At the start of month 1, a two-asset portfolio has target weights (0.6, 0.4). The assets return +5% and −10% in month 1. At the start of month 2, the target weights are again (0.6, 0.4), with a cost of 25 bp per unit of turnover. The assets return +2% and +1% in month 2. Calculate the drifted weights at the end of month 1, turnover at the start of month 2, cost, and gross and net returns in month 2.

P3-02 (Ledger identities) Read outputs/primary_ledger.csv. For every row, verify $\text{cost}=0.001\times\text{turnover}$ and $\text{net}=(1-\text{cost})(1+\text{gross})-1$. Also verify that, for each strategy, train_end in each of two consecutive months precedes holding_month.

P3-03 (Why lag does not affect equal weight) In sensitivity.csv, the equal-weight metrics are identical under lag 0 and lag 1, while Minimum Variance differs slightly. Explain why, and what this implies for conclusions about the importance of the timing assumption.

P3-04 (Interpreting intervals) ERC's annualized net Sharpe is 0.016 higher than equal weight, with a 95% block-bootstrap interval of [−0.014, 0.050], while the mean net-return difference is −0.38% per year. Can you say that ERC outperforms equal weight? Write wording for a conclusion you would accept.

P3-05 (Unregistered exploration) Someone proposes changing the window to 24 months “to see if it gets better.” Use the functions in src/project03_backtest.py to run Maximum Sharpe with W=24 and all other main-specification parameters unchanged. Report annualized net Sharpe and explain how this result should be labeled in the report.

P3-06 (Research design) If this project were to become a tradable study, list the data, timing, costs, and evaluation elements that must be replaced or added, and give one piece of evidence that would overturn the claim that “ERC reduces risk concentration without sacrificing returns.”