Methods for Alternative Investments

Question 1

Q-measure and P-measure

Answer 1

Q-measure: probability like value from a model that assumes risk-neutrality while the actual world is probably not risk neutral.

P-measure: probability that represents the likelihood of a real world outcome.

Question 2

Four key concepts of risk-neutral modeling

Answer 2

1. Often an infinite set of values
2. Expected returns in a risk-averse world are unobservable.
3. Value obtained from Q-measure is identical P-measure in a risk averse world.
4. Q-measures are tractable

Question 3

Three fallacies generated by averaging compounded rates of return

Answer 3

1. ETFs incur wealth destroying performance when volatility occurs even in an efficient market.
2. Inverse ETFs also incur wealth destroying performance.
3. Rebalancing of portfolios cannot generate positive NPVs when the underlying assets are efficiently priced and offer NPV=0

Question 4

Two paradoxes of informational market efficiency

Answer 4

1. If financial markets are efficient, no one will have an incentive to collect information so there would be no mechanism to keep the market informationally efficient.
2. If financial markets are perfectly efficient then fees paid to active managers imply that the market for asset managers is highly inefficient

Question 5

When do efficiently inefficient markets occur?

Answer 5

inefficient enough to compensate managers for the cost of pursuing skill-based strategies, but too inefficient to present a large number of money managers with easy arbitrage opportunities.

Question 6

Two methods of implementing a momentum strategy

Answer 6

Cross-sectional (relative to index)

Absolute (own performance)

Question 7

Market Divergence Index

Answer 7

The average signal-to-noise ratio for a group of submarkets

Question 8

Generic algorithms vs neural network

Answer 8

Generic algos are modeled after the natural selection process while neural networks are modeled after the learning process of the human brain with various nodes and layers. Both seek to identify patterns in data.

Question 9

Crisis alpha

Answer 9

The potential for some strategies to generate superior returns during periods of financial crisis.

Question 10

Two popular valuation approaches among bottom-up investors

Answer 10

DCF Models and enterprise values models

Question 11

Four mechanics of the traditional fundamental investment process

Answer 11

1. Idea generation (most critical step)
2. Optimal idea expression
3. Sizing the position (sized to conviction)
4. Executing the trade

Question 12

3 top-down schools of thought

Answer 12

Feedback-based global macro: markets are rational most of the time, these managers try to read the markets psychology.

Information-based global macro: collect micro-level information as there is a delay between micro information release and the effect on macro.

Model-based global macro: focus on financial/economic theories to analyze market movements (e.g. policy mistakes)

Question 13

Two risks of directional fundamental strategies

Answer 13

1. Fundamental risk (unexpected change in fundamental value of a security).
2. Noise risk (investors trading for reasons not related to fundamental value)

Question 14

Two building blocks of behavorial finance

Answer 14

1. Limits to arbitrage (rational investors unable to undo dislocations by unrational traders)
2. Cognitive psychology

Question 15

3 Sentiment indicators

Answer 15

(1) Discount on closed-end funds
(2) Turnover NYSE
(3) Number of IPOs

Question 16

Anchoring

Answer 16

Give too much weight to previous observations

Question 17

Loss aversion/disposition effect

Answer 17

investors prefer to avoid losses more than to acquire gains

Question 18

Prospect Theory (Kahneman)

Answer 18

Investors tend to underweight probable outcomes vs. certain outcomes

Question 19

Two primary outputs of a PCA analysis

Answer 19

eigenvalue: proportion of return variance of a factor

factor loadings: responsiveness of each asset to each principal component

Question 20

Three key differences between PCA and FA

Answer 20

1. FA makes specific statistical model assumptions
2. FA generates different factor scores when a different number of factors are used (PCA = constant)
3. PCA can identify a factor driven by one security, FA seeks factors that drive at least two securities

Question 21

What is multicollinearity and what are two adverse effects?

Answer 21

When independent variables in a regression analysis have a high correlation.

1. Slope coefficients may be inaccurate
2. Standard errors might be inflated

Question 22

Stepwise regression

Answer 22

Iterative process whereby variables are added and deleted based on their statistical significance.

Question 23

Overfitting models

Answer 23

Explain the past well

Question 24

Three dynamic risk exposure models that can help estimate the effectiveness of market-timing strategies

Answer 24

1. Dummy variable approach
2. Separate regression approach
3. Quadratic approach

Question 25

Dummy variable approach

Answer 25

Add a dummy variable of 1 when excess return is positive and 0 when excess return is negative

Question 26

Separate regression approach

Answer 26

Similar to dummy variable approach but using separate regressions based on subsamples

Question 27

Quadratic approach

Answer 27

Assesses market-timing skills using a quadratic curve

Question 28

Two approaches to model changing correlations

Answer 28

Conditional modeling approach (specified conditions)

Rolling window modeling approach (subperiods)

Question 29

Limits to arbitrage

Answer 29

Degree of leverage that an investor can cost-effectively arrange and the level of risk they can tolerate

Question 30

Four steps for implementing a pairs trading strategy

Answer 30

1. Identify the candidates
2. Identify pairs with a divergent spread
3. Construct and size the portfolio
4. Exit strategy