Prado2018 pages 260-261: The text discusses the performance of different portfolio allocation strategies in terms of risk management. It compares the performance of the Hierarchical Risk Parity (HRP) strategy against the Minimum Variance (CLA) and Inverse Variance Portfolio (IVP) strategies. The text mentions that HRP finds a compromise between diversifying across all investments and diversifying across clusters, making it more resilient against systemic shocks. It also states that in out-of-sample simulations, HRP outperforms CLA and IVP in terms of variance reduction, which is important for risk parity investors.

Prado2018 pages 178-179: The text discusses the concept of backtesting and its limitations. It states that a backtest is not an experiment and does not prove anything. It mentions that a backtest can be helpful as a sanity check on variables such as betsizing, turnover, resilience to costs, and behavior under a given scenario. However, it also highlights the difficulties of backtesting and mentions common errors made in backtesting, such as survivorship bias, look-ahead bias, storytelling, data mining, transaction costs, outliers, and shorting. It concludes that even if a backtest is flawless, it is likely to be wrong.

Quotes:

• "A backtest guarantees nothing, not even achieving that Sharpe ratio if we could travel back in time in our retrofitted DeLorean DMC-12."
• "A good backtest can be extremely helpful, but backtesting well is extremely hard."
• "These are just a few basic errors that most papers published in journals make routinely."
• "Even if your backtest is flawless, it is probably wrong."

Prado2018 pages 272-276: The text provides information on risk management strategies and methods. It discusses the concept of structural breaks, which refers to the transition from one market regime to another. Structural breaks can create opportunities for profitable strategies as market participants may make costly mistakes during the transition. The text mentions methods for measuring the likelihood of structural breaks, such as cumulative forecasting errors tests (CUSUM tests) and explosiveness tests. These tests help identify deviations from whitenoise and determine if the process exhibits explosive behavior.

Prado2018 pages 347-348: The text discusses the problem of risk management for large asset managers. Excessive turnover and implementation shortfall can erode the profitability of investment strategies. The objective function is defined as a trading trajectory that determines the proportion of capital allocated to each asset over each time horizon. Transaction costs are associated with changes in capital allocations and are re-scaled by asset-specific factors. The Sharpe Ratio is used to compute the performance of the investment strategy. The problem is to compute the optimal trading trajectory that maximizes the Sharpe Ratio, subject to constraints on capital allocation. The problem is not convex due to non-identically distributed returns, non-continuous transaction costs, and a non-convex objective function. An integer optimization approach is used to find the optimal solution.

Prado2018 pages 313-314: The text discusses second-generation microstructural models that focus on understanding and measuring illiquidity in financial ML models. These models explain trading as a strategic interaction between informed and uninformed traders and pay attention to signed volume and order flow imbalance. The text also introduces Kyle's Lambda, a strategic trade model that considers a risky asset, a noise trader, and an informed trader. The market maker adjusts prices based on the order flow imbalance, and there is a positive relationship between price change and order flow imbalance, known as market impact. The informed trader's profits are maximized at a certain demand level, and the market maker must find an equilibrium between profit maximization and market efficiency.

Quotes:

• "Second generation microstructural models focus on understanding and measuring illiquidity."
• "These models have a stronger theoretical foundation than first-generation models, in that they explain trading as the strategic interaction between informed and uninformed traders."
• "The informed trader conjectures that the market maker has a linear price adjustment function, p=𝜆y+𝜇, where 𝜆 is an inverse measure of liquidity."
• "Kyle argues that the market maker must find an equilibrium between profit maximization and market efficiency."