Global Asset Allocation and Stock Selection
Assignment 5: Final Overview
Due 24 hours from the time you look at the assignment, no later than Tuesday February 28, 2006 6:00pm
This assignment is to be completed on an individual basis. No discussion
of the questions is allowed. Collaboration is a violation of the Fuqua Honor Code. Turning in the assignment more than 24 hours after viewing the assignment is
also a violation of the Fuqua Honor Code.
You are required to turn in no more than 2 pages of a Word file with no smaller than 10-point font. Email me a file called A5_Lastname_Firstname.DOC (i.e. A5_Smith_Jean.DOC for Jean Smith). Do not include any embedded graphics. No empirical work is necessary for any of the questions.
If any of the questions or parts of the question are unclear, make a reasonable assumption so that I understand your interpretation.
Please scan your file for virus infection before emailing. Do not post to the student drive!
Your company is thinking of making an allocation the Alpha Vista Fund, LLC. You are in charge of due diligence. Your firm is looking for long U.S. equity exposure
and over the past 5 years (Alpha Vista started up 5 years ago and has grown to $1.2B under management) has out-performed the S&P 500 every year. On a year by year basis, the fund's beta has fluctuated between
0.97 and 1.03 and over the whole period the beta is 0.99. However, the average annual alpha is 160 basis points. It has ranged from 190 basis points to 90 basis points.
The Alpha Vista's strategy involves holding the underlying stocks of the S&P 500 and writing out of the money calls on both the index and some large market cap constituents
of the index. The amount of writing has been consistent over the past 5 years (with notional equal to the assets under management). You request some additional tests. First,
you ask for a three factor alpha (adding a size and value factor) -- but the alpha from this is basically the same as the simple alpha. Next you add two fixed income factors and
experiment with some macroeconomic factors. Again, the alpha is roughly unchanged. You even experiment with a dynamic beta where the market return is interacted with
a yield curve variable. Again, the alpha is the same. Your job is assess whether this alpha is genuine. What is your recommendation? [Quarter to half page]
Again, you are doing due diligence. This time you are visiting Robert Arnott's Research Affiliates, LLC (RA) in Pasadena. RA has launched a new product called "Fundamental Indexing" (RAFI).
The idea is simple. While the market might be relatively efficient on average, from day to day, it makes sense that stocks are not exactly priced at the unobservable "true" price.
Hence, some stocks are overvalued and some stocks are undervalued. Even if you don't know the true price, the market capitalization weighting scheme embedded in most indices will
cause immediate problems. Market cap weighting will overweight overvalued stocks and underweight undervalued stocks. This is a powerful intuition which leads your firm to consider the
the RAFI. In the RAFI, indices are constructed using alternative measures of size (like total sales, total book value, number of employees, and other fundamental variables). The index weights
are averaged to get a new weighting scheme. On backtests, the RAFI indices outperform the market cap weighted index by about 120 basis points per year. You are thinking of doing the same
exercise as you did with Alpha Vista Fund, however, you quickly discover that similar analysis has already been done by RA - and published by RA. While the simple alpha is 120 basis points per year,
the three factor alpha is zero. The RAFI loads up on the value factor. You ask RA to explain. RA says that the problem lies with the use of the value "factor." They argue that it is not
is not really a risk factor. It proxies for something else -- and it is exactly what RAFI is designed to capture. Discuss what you think about this? [Quarter to half page]
The RAFI exercise gets you thinking. The RAFI under and overweights based on fundamental variables. A stock might get a higher than market cap equivalent weight because the
fundamentals look relatively better than the market capitalization. However, you are worried about the quality of some of the fundamental data. If the firm is "managing" some
of the fundamental data, then the RAFI might overweight a firm that is already overvalued. You know that there has been research on earnings
quality. Quality is often represented by the amount of accruals a firm is doing. How might you potentially improve the RAFI method by taking quality into account? Do you have another other ideas that might improve their performance?
[Quarter to half page]
Think of an ex post mean variance frontier, five assets and five years of monthly data. Think of plotting the individual assets and the frontier. The individual assets are all inside
the frontier and they represent 'buy and hold' portfolios (of just one asset). The frontier represents the best fixed weights over the five year period, i.e. weights that are
rebalanced every month. However, there are no buy and hold portfolios represented in the frontier (a buy and hold portfolio is one where you choose the initial weights in period 0
do not rebalance for the 60 months). Discuss how you might of add buy and hold portfolios to the optimization problem? [No calculations please.] How relevant is this for our monthly
tactical asset allocation exercise?
Your final scoring screen performance looks great -- however, you get hammered in 1998 and 1999. The problem is that the value factor works well on average but not all the time.
Outside the selection model, you come up with a time-series model that has the ability to predict when value outperforms growth. The predicted values of the model range between
0 and 1, with 0 meaning the model predicting a zero chance value will outperform growth, 0.7 meaning there is a 70% chance that value will outperform growth, and 1.0 meaning
there is a 100% predicted chance that value will out perform growth.
Currently in our scoring screen, you assign a "3" (i.e. 3 out of 5) to the value factor. Given the results of this auxiliary model, how might you change
the way the scoring screen works? Be specific.
In some of the presentations today, we saw that the long-short portfolio had greater volatility than the S&P 500. Perhaps the extra performance we get over and above the S&P is just the result
of taking on extra variance risk. Comment. Indeed, how do we think of the 2x2 that I mentioned in class (with $100 million under management, you go long $200 and short $200). It was far larger volatility than
the S&P 500? Comment. [Quarter page]
Provide a group evaluation. Names and percentage contributions (including your own).
If a group member participated in a subset of assignments, please assess the contribution based on the sum of assignments they participated in -- and note which assignment they did not participate in.
Thanks. If you leave this question blank, I am assuming equal weights.