金融辅导猿

MPT、CAPM、CAL这些原理背后的变量都是weight1，2，最终却全部化成了R和Variance 两个没有Weight的表达方式，真是妙的很呀。

Chapter 01

1.1 课程范围

• Modern Port folio Theory and the equilibrium pricing of assets
• Pricing of derivative instruments and related concepts
• A few practical applications of financial modelling

Topics such as time series analysis/forecasting models and hi gh frequency trading
algorithms will only be briefly mentioned as appropriate.

行，名字里加个计算机就能想讲啥讲啥😁

1.2 Different Financial Markets

1.2.1 金融术语

1.2.2 金融市场

1.3 Fundamental Financial Concepts

1.3.1 Risk and return (Variance and Expectation)

1.3.2 Time Value of Money

1.3.3 Calculation of returns

In summary, the first chapter requires understanding and simple calculation.

Chapter 2&3 Modern Portfolio Theory

2.1 学习目标

🀄️3 Steps in the asset allocation process
• Combining risky assets (投资组合）
• Allocation between risk free and risky assets （搭配无风险资产，调整投资决策）
• Combining with investor’s preference （结合个人风险偏好）

搞学术吹牛必备历史：

Frank Fabozzi, PetterKolm, DessislavaPachamanova, Sergio Focardi, Robust Portfolio         Optimization and Management, Wiley, 2007

Step1. Modern Portfolio Theory (MPT) was pioneered by Harry Markowitz in a paper published in 1952

1. 用E(r) 和 Variance表示单个金融产品的return和risk

2. Matrix convention【用线代来算投资组合的risk和return】

3. 相关性为+1：没有分散风险的作用；相关性为-1，well- diversified

非系统性风险【可以通过构建portfolio降低】和系统性风险【用beta来衡量】

Minimum variance frontier； Efficient frontier; Global minimum variance portfoliois

【用拉格朗日来求最优解，听起来就牛逼🧱】

 Step 2: Allocation between risky and risk free assets

找出可行的最优配置

Step 3: combining with the investor’s preference 

不讲武德，想到第三层了。Step1如何配置金融产品间的风险分散最大化比例；Step2如何配置无风险和这个最优portoflio间的比例【你愿意承担多少风险】；Step3：【二次函数型】投资人获得最大满足的点在哪。

精彩花絮

Single factor model/Multi-index model

单因素的模型最后就拆成了市场风险和股票风险，所以才能Step1计算两个股票间的相关性的假设。

知道模型还不够，得了解优缺点才能显出水平。

 

•Examples of factors include economic growth, business cycle, long and short term interest rates, inflation, industrial production, strength of US dollar etc.
•For the model to work well, the factors should be independent from each other •Empirical studies were carried out to test the performance of these models
•Numerous possibilities exist, so we could not draw a definitive conclusion
•Some models give reasonable results, but many multifactor models do not perform as well as a single index model; this illustrates the difficulty in choosing appropriate factors (e.g. factors with high explanatory power in one period may become unsuitable in the future)

所以，挑选的方式就有很多了：

•Select individual stocks/bonds which have the best risk/return characteristics
•Accurate forecasts are essential

或者 Dividend assumption/ESG/Ethical issue

•Complex practical issues such as the inclusion of transaction costs, optimization across multiple client accounts, and the need to take tax consequences into considerations

•A number of different constraints can be added to the optimization problem：

I. Linear and quadratic constraints

II. Combinatorial constraint

III. Availability of a risk free asset

•Assumptions behind the theory
•The mean-variance criterion
        •Samuelson (1970) showed that, although the distribution of returns shows higher         moments, disregarding these would not affect portfolio choices
•Quadratic utility functions
        •In Markowitz original work in 1959, it was shown that quadratic utility functions are         good local approximates of more complicated forms
•Normal distribution of returns
        •Even though individual asset returns may not be normal, portfolio returns will         resemble a normal distribution

•Difficulty in obtaining reliable estimates for variances / covariances
        •Changing variance due to change of circumstances (“heteroscedasticity”)
        •Changing correlation between assets
        •Increasing correlation in turbulent periods
        •Using history to forecast the future
        •The forecast is usually better for variances than for the returns; but is it accurate         enough?
•Use more sophisticated econometric models, e.g. GARCH, for better predicting the future variances

对冲巨佬提过的：正相关不一定真相关

Rationality arguments
•Investor’s could be inconsistent, i.e. they make choices in an irrational (or even random) manner
•It is sometimes impossible to quantify their preferences in terms of a utility function
•Behavioral finance topics [refer to lecture 11]
•Overconfidence, framing, mental accounting, regret avoidance etc.
•Daniel Kahnemanshared the Nobel prize in Economics in 2002 (joint work with Amos Tversky)
•Richard Thalerwon the Nobel prize in Economics in 2017

How could we use the MPT?
•According to this theory, everyone should hold the same risky portfolio (in the same proportion)
•if this is the case, many finance professionals would lose their jobs!!!
•Because of various reasons, many investorsmay be reluctant to invest in some asset classes
•E.g. Hong Kong investors may find it inconvenient to invest in stocks in many foreign countries
•While it is often impractical to include all kinds of assets as inputs to the optimization process, we could still use Modern Portfolio Theory to optimize within each asset class

Problems with MPT
•Mean-variance optimization using historical data often yields unrealistic results
•Unintuitive weights of assets in the portfolio
•Concentrated in a few representative stocks
•Small changes in the input parameters lead to drastic changes in portfolio allocations
•Note that the optimization results are less sensitive to errors in estimating variance, and that the estimation of population covariance is more stable over time than estimating returns from historical data