Kindle Notes & Highlights
Started reading
October 9, 2025
we feel strongly that current curriculum of many such programs is often light on portfolio theory and portfolio management, and long on option pricing theory and various microscopic views of market efficiency (or lack thereof). As practitioners and active researchers in the field, we have selected topics essential to quantitative equity portfolio management, from theoretical foundation to recently developed techniques.
the investment team must have (1) a strong philosophy based on commitment to a set of beliefs, (2) a clear approach in translating uncertainty into an appropriate risk/return trade-off, and (3) a comprehensive investment process from beginning to end.
1980s, there came a volume of formal literature that discovered inefficiencies that could lead to abnormal returns if rigorously applied. The list includes size effect, January effect, value irregularities, momentum effect, etc. We called them anomalies1 and reverently acknowledged in the conclusion that these discoveries (1) were likely not repeatable in the future (now that we know them), (2) may be inconclusive because of potential “risk misspecification,” or (3) were lacking the proper allocation of costs in the strategy. In a modern quantitative process we call these anomalies “factors,”
...more
true alpha-generation is available to practitioners who creatively combine modern tools — econometrics, mathematics, investment theory, financial accounting, psychology, operations research, and computer science. Fifth, objective discipline is essential in the implementation of strategies. This is not to say subjective judgment is lacking in the world of quantitative management — but it lies in perfecting the comprehensive portfolio system, rather than in comprehending the perfect stock selection.
Think about the investment process as a whole rather than just forecasting
Risk-adjusted positive skill is the true goal of the game.
neutral. It is often hard, if not impossible, to disentangle what is alpha and what is beta. For a long time, nobody cared because most of the investors in the hedge funds were high-net-worth individuals who had their eyes on the absolute returns, not abstract geeks. Today, the situation has changed dramatically. Equity market neutral managers (mostly quants) manage zero-beta funds with refined risk management systems, and often deliver pure alpha. Institutional investors are increasingly pursuing and paying handsomly for alpha, but are unwilling to pay excessively for beta management. Hence,
...more
The alpha model is often proprietary and highly guarded, reflecting creativity as well as superior systems. It is the most important differentiator within the investment firm.
It is inaccurate to say that fundamental managers dig deep at the solo stock level, but have no models or disciplines. It is also unfair to say that quantitative managers apply skills to so broad a set of stocks that the process is superficial at the fundamental level, and often labeled black-box, datamining nerds. This is a misrepresentation. Many quantitative investment strategies rely on factors that are based on not only solid economic principles, but also on sound fundamental intuition (more on this in Chapters 5 and 6).
To repeat, quantitative management — lies in broadly perfecting the comprehensive portfolio system, whereas, fundamental management lies in deeply comprehending the perfect stock selection. In many instances, the underlying principles of quantitative investment are no different from traditional fundamental research.
right? Many become easily get excited about mean–variance optimization and Monte Carlo simulation but are bored with balance sheet and cash-flow analysis. This is the wrong attitude, perhaps. Some of the most valuable information, quantitative or fundamental, is only garnered through painstaking analysis of financial statements.
Investing without true information is just speculation. How do we know we have true information that can predict security returns? On one level, predicting a market crash is not enough, even if you are correct once. In the same vein, neither is finding the correct target prices for a couple of stocks a proof of skill. The key to investment success is consistency in forecasting (skill) applied repeatedly (breadth). We have Grinold and Kahn (2000) to thank
Great distinction - being right once does not mean you have a repeatable investment system. And a few forecasts does not mean you have a robust framework for security analysis.
However, a quantifiable forecasting edge across a wide array of assets suggests skill.
portfolios. Clarke et al. (2002) generalize FLAM introducing the concept of transfer coefficient to approximate the loss of information due to constraints. These studies highlight the dampening effect of overly stringent constraints on investment performance. This awareness across the investment community has created increased receptivity to long-short portfolios, either “pure” or constrained, in the
ways. First, stocks are idiosyncratic in nature. A onesize-fits-all model assumes that all stocks respond to the factor exposure in the same way all the time. Practitioners know this is not true, and are beginning to analyze factor significance within this context.
Alpha can also be allusive, and today’s alpha could be gone tomorrow or reclassified as beta in the future. However, one thing is constant: investors such as institutional fiduciaries, pension funds, endowments, and the like, will continue to pursue risk-adjusted alpha through active equity management. It might be that the latest surge of formal quantitative investing has, in part, ushered in better metrics for “separating alpha from beta” and therefore led to a higher level of general understanding of the difference.
After years of unsatisfactory efforts to explain market anomalies by efficient market theorists, behavioral economists took an alternative approach to challenge two key tenets of equilibrium pricing models: (1) arbitrage activity eliminates pricing discrepancies completely and (2) investors behave rationally.
On a more sophisticated level, the arbitrageur also faces the noise trader risk. Shleifer (2000) argued that irrationality is to some extent unpredictable, and it is plausible for today’s mispricing to become even more extreme tomorrow. In other words, convergence of price dislocation is not a certainty.
control. Biased self-attribution is that phenomenon in which people attribute success to skill and failure to bad luck. Kahneman and Riepe (1998) noted, “The combination of overconfidence and optimism is a potent brew, which causes people to overestimate their knowledge, underestimate risks, and exaggerate their ability to control events.” Emotions and self-control: Hirshleifer (2001) posited that emotion could overpower reason. For example, people who are in good moods are
First and foremost on the list is an alpha model, which predicts the relative returns of stocks within a specified investment. The second component is a risk model that estimates the risks of individual stocks and the return correlations among different stocks. The third piece is a portfolio construction methodology to combine both return forecasts and risk forecasts to form an optimal portfolio.
Several features of the normal distribution are undesirable or unrealistic when it is used to model stock returns. First, a stock investor has only limited liability — he could not lose more than what he invested in. Therefore, the return of a stock over any time horizon should never be less than −100%. But a normal distribution assigns nonzero probability to losses of any size, even those exceeding −100%. Second, if we assume that a single-period return for a stock is normal, the compound return over multiple periods is no longer normal.
The concept of diversification refers to the fact that the total risk of a portfolio is often less than the sum of all its parts. Diversification arises when the returns among different stocks are not perfectly correlated. The correlation coefficient between two stocks relates to their covariance and standard deviations
