Chapter 5.

Correlation & Performance Measurement

March 4, 2015 jhud

5.1 Demonstrate knowledge of various measures of correlation between assets.

For example:
· Recognize the importance of correlation in alternative investment portfolio
management
· Define and apply the concept of covariance
· Define and apply the concept of the correlation coefficient
· Define and apply the Spearman rank correlation coefficient
· Discuss the role of correlation in portfolio diversification
· Define and apply the concept of beta in the context of the CAPM
· Define autocorrelation
· Apply the concepts of the first order autocorrelation coefficient
· Recognize and apply the Durbin-Watson statistic

Risk of the portfolio = how correlated the components are => how diversified is it
Covariance is the degree to which two assets move in the same direction. Similar to
correlation -1 to 0 to +1
similar to the formula to variance except instead of squaring you cross multiply
Correlation Coefficient aka Pearson Correlation: degree of association of two variables
scales covariance to -1 to 1
Spearman Rank Correlation Coefficient; Pearson can be more susceptible to outliers than
Spearman
Correlation Coeficient often used to demonstrate risk reduction through combination of
assets that are not perfectly correlated
The portion of a portfolios risk that can be diversified away is called: diversified, non-
systematic, unique, or idiosyncratic risk.
Beta

beta of an asset = covariance between asset & market, over market volatility, similar to
correlation coefficient except that it’s not bound by -1 to 1
beta of 0.95 means a 1% move 0.95%. beta of 2 would move 2%
beta is also slope coefficient in linear regression
because beta is linear it represents weighted average of portfolio constituent asset betas
beta is estimated rather than observed (as is correlation/covariance) and is likely
difference in future

Auto-Correlation aka serial correlation

relationship between returns of an asset across different time periods
also range between -1 to 1 (like correlation)
use similar formula to covariance except returns are from same asset (separated by time
periods (‘k’)
first order autocorrelation is correlation between “t” and “t-1”
Durbin Watson is test for AutoCorrelation in a time series
2 indicates no autocorrelation
If DW Stat statistically significant
>2 indicates negative autocorrelation (mean reversion)
<2 indicates positive autocorrelation (trend)
0 autocorrelation is rejected at 1 or 3
5.2 Demonstrate knowledge of the internal rate of return (IRR) approach to
alternative investment analysis. For example:
· Define and apply the IRR
· Define and apply the three types of IRR based on time periods for which cash flows are
available (i.e., lifetime, interim, and point-to-point) and their relationship to valuation of
alternative investments

IRR is used when valuation data is not available e.g. equity, real estate
IRR is the discount rate that makes NPV (Net Present Value) = Zero
“Cash Flow Mode” on calculator:
e.g. 250 milion dollar investment (bank deposit)
year 1 returns: 150M (withdraw)
year 2: 100M (withdraw)
year 3: 80M (withdraw)
RoR = 17.33%

IRR can be done on realized or expected cash flows

3 types of IRR
lifetime: include all cash flows 0 to T
interim aka since inception IRR: usually uses an appraised value prior to termination
point-to-point: some time in between beginning and end

5.3 Demonstrate knowledge of problems with the use of IRR in alternative

investment analysis.
For example:
· Recognize complex cash flow patterns and discuss their effect on the
computation and interpretation of IRRs
· Discuss the challenges (e.g., scale differences) of comparing investments based on IRRs
· Discuss the difficulties of aggregating IRRs
· Recognize factors that contribute to the sensitivity of IRRs to cash flows
· Discuss the reinvestment assumption inherent in the IRR and how it is
addressed by the modified IRR
· Compare and apply time-weighted and dollar-weighted returns

Problems with IRR, difficult to apply when initial cash outflow is in fact an inflow in the form
of a loan, followed by cash outflows
IRR is very sensitve to scale, aka time period of cash flows
Aggregation of IRR doesnt work, instead have to consider sum of cash flows at each point
in time.
High IRR’s are insensitve to distant cash flows, low IRRs are highly sensitive to distant
cash flows; Size and longevity of cash flows can have varying impact on IRR
A modified IRR can be used to address the assumption of reinvestment (which is that the
IRR can obtained for any cash subsequent cash flow)
Time weighted vs Dollar weighted returns, time weighted returns do not have cash flows,
dollar weighted do. IRR is a method for calculating Dollar weighted returns.
Dollar weighted returns can be used to estimate the return to an average investor in a
hedge fund (treat new allocations as cash flows).

5.4 Demonstrate knowledge of returns based on notional principal.

For example:
· Define and apply the concepts of notional principal and full collateralization
for forward contracts
· Apply the concept of the log return to a fully collateralized derivatives
position
· Apply the concept of the log return to a partially collateralized derivatives
position

Some contracts/investments have zero or no clear starting value so the rate of return must
be calculated differently e.g. with forwards and options

In the case of forwards notional principal can be used, i.e. the dollar value of the contract.

Fully collateralized means no margin/leverage. You also earn the risk free rate.
= ln(1+R) +rfr

Partially collateralized means leverage amplifies returns, you earn less from the risk free
rate

5.5 Demonstrate knowledge of the distribution of cash waterfall.

For example:
· Explain the distribution of cash waterfall provision of a limited partnership
agreement
· Recognize terminology associated with the cash waterfall provision (e.g.,
carried interest, hurdle rate, catch-up provision, vesting, clawback clause)
· Discuss factors (e.g., management fees, incentive-based fees) to consider in a
fund’s compensation structure and the potential effects of decisions regarding
compensation structure
· Contrast and apply fund-as-a-whole carried interest and deal-by-deal carried
interest
· Define and apply the concept of clawback provisions, including their purposes
and limitations
· Compare and apply hard and soft hurdle rates and their sequences of
distribution
· Discuss the potential effects of incentive fees on decision making and their
option-like nature

Vesting: can be pro-rata over the investment period or other period, vested rights cant be
traded or sold and may be subject to forfeiture. Vesting is the opposite of Clawback.

Clawback provision (aka give back or look back): return fees to LP when losses are
followed by early profits (return of incentive fees). Escrow could be used for collected
incentive fees.
Compensation structure: aligns incentives with investors, fees, GP investment in the fund,
vesting, etc
aggregating profit and losses, PE: can be deal by deal carried interest (incentive) or on
fund as a whole. Deal by deal doesnt align that well with LPs interests as less risk for the
GP
Hard vs Soft hurdle rates: hard means you only earn incentive fees on the amount of profit
above “x” whereas soft means you earn incentive fees on all profit once the rate of “x” is
exceeded.
Upon liquidation of the fund:
repay investment
HARD HURDLE: incentive fee only charged on profit above hurdle rate, e.g. if 10M profit,
hurdle rate works out to half of 10M then manager gets 20% incentive from the 5M portion
of the profit
the catch up provision on a soft hurdle rate requires that -> the example in the book is
written extremely poorly, terrible!
incentive fee = long call option for the manager, higher assets = higher payoff. lower
assets = no payout. time to expiration = time to next payout
strike price = fund NAV, hurdle rates increase the strike price
incentive to increase vol, with limited losses, e.g. fees
managers may be more risk diverse when deep in the money (i.e. owed a lot of incentive
fees) – and vice versa (as opposed to making decision based on trading opportunities).
5.6 Demonstrate knowledge of performance measures used in alternative
investment analysis.
For example:
· Define the ratio-based performance measure type
· Recognize and apply various ratio-based performance measures (i.e., the
Sharpe ratio, the Treynor ratio, the Sortino ratio, the Information ratio, and
return on VaR)
· Define the risk-adjusted performance measure type
· Recognize and apply various risk-adjusted performance measures (i.e.,
Jensen’s alpha, M2 [M-squared], and average tracking error)

Ratio based Performance Measures: return over risk with some bench or rfr incorporated.
Sharpe ratio: return-rfr / stdev = annual risk premium earned per percentage point of
standard deviation (risk).
e.g. sharpe of 0.35 means excess of 35 basis points above the risk free rate for each point
of standard deviation.
Sharpe ratio for portfolio returns rather than components of a portolio, ok for comparison if
on the same time intervals and period
Sharpe is based on stdev for risk therefore may not be good for non-normally distributed
returns i.e. skew and kurtosis
Treynor ratio uses Beta instead of Standard Deviation as the input for Risk. i.e. the risk
premium the investment earns per unit of beta e.g. Treynor of 0.0467 would be an return
of 4.67% higher than the risk free rate
Treynor is best for comparing components to be added to a well-diversified portfolio and
should be used on a standalone basis. It’s not commonly used in Alternative Investments
because beta isnt always appropriate as a risk measure.
Sortino like sharpe but uses a benchmark or target instead of risk free rate and downside
deviation or semistandard deviation ie. deviation below bench/target instead of stdDev
Information Ratio is like sharpe except that it uses a benchmark return instead of the risk
free rate and a tracking error instead of standard deviation. It uses deviation from the
benchmark instead of standard deviation
Return on VaR: (VAR is a measure of risk for a specified time for a level of confidence).
RoVaR divided by a specified VaR, fine for normal distibution
Risk-Adjusted Returns:
Jensens Alpha: measure of the % an asset is estimated to outperform, Expected Return
minus Risk Free Rate minus Beta multiplied by (market return minus risk free rate) =
Jensens Alpha
M Squared: expresses market return normalized by the risk of the market portfolio, i.e.
deleveraged or leveraged according to the market risk, m-squared is equal to the amount
the return is greater (or less than) than the market when normalized to the same leverage
Average Tracking Error: average differences in returns from benchmark, the numerator of
the information ratio.