Active Equity - Strategies

Question 1

Statistical Arbitrage - Pairs Trading

Answer 1

extensive use of technical stock price and volume data to exploit pricing inefficiencies
aim to profit from mean reversion in related share prices

identifies two securities in the same industry that are historically highly correlated with each other and aims to profit from taking advantage of a temporary breakdown in this relationship

The risk is that the breakdown of the observed previous relationship is long term in nature, there is no mean reversion,

Question 2

Pitfalls in Quant strategies

Answer 2

Survivorship bias is one of the most common issues affecting quantitative decision making. While investors are generally aware of the problem, they often underestimate its significance. When back-tests use only those companies that are currently
in business today, they ignore the stocks that have left the investment universe due to bankruptcy, delisting, or acquisition.

Look-ahead bias: using information that was unknown or unavailable at the time an investment decision was made

Data-mining
Lack of turnover
Limits on the availability of long and short positions
Transaction costs

Question 3

Holdings based vs, Returns based Style analyses

Answer 3

Some active equity managers may maintain one investment style over time in the belief that that particular style will outperform the general market.
Others may rotate or switch between styles to accommodate the then-prevailing investment thesis.

Returns-based style analysis regresses the portfolio’s historical returns against the returns of the corresponding style indexes (over 60 months in this example). Its output indicates the average effect of investment styles employed during the period. While the holdings-based analysis suggests that the current investment style of the equity fund is value oriented, the returns-based analysis indicates that the style actually employed was likely in the growth category for a period of time within the past five years.

Question 4

Back-testing

Answer 4

The purpose of back-testing is to identify correlations between the current period’s factor scores, FS(t), and the next period’s holding period strategy returns, SR(t + 1).

Question 5

Impact of adding a new Factor to Investment-style classifications using either the returns-based or holdings-based approaches.

Answer 5

changing its factor model by adding a new factor, the correlations of the fund’s returns with the factors would likely change and the returns-based style would change. Even though the investment universe is unchanged, the portfolio holdings would likely change and the holdings-based style classification would also will be affected.

Question 6

Returns-Based Style Analysis (an approach to Style Classification)

Answer 6

• useful when full details of Portfolio holdings are not known
• statistical tools are used to identify the style indices that provide the most significant contribution to portfolio return
• key inputs to the analysis are the historical returns for the portfolio and the returns for the style indexes
• Selection of style indexes is crucial as stock returns can be highly correlated with the same sector, across sectors, and across global markets
• commercial investment information providers such as Lipper and Morningstar collect and analyze fund data, classifying funds into style groups

Drawbacks
- Using statistical analysis may generate inaccurate results if input data is limited or there are flaws in the application design
- May limit results within certain boundaries making it difficult to detect aggressive positions

Question 7

Strengths of Holdings-based analysis compared to Return-based

Answer 7

• more accurate as actual portfolio holdings are used
• managers can see how each individual holding contributes to style, verify that each style is in line with stated investment philosophy, and take any action to prevent style deviation
• the styles a portfolio is exposed to is visible

Holdings-based approaches are done bottom-up but executed differently by various commercial information.

Size (Small-cap, mid-cap etc), Value vs. Growth

A simple value scoring model uses one factor, such as price-to-book ratios to rank the stock

A value score of 0 to 1 is assigned OR Comprehensive models use multiple factors, assign a fixed weight to each factor