Performance Evaluation Flashcards
Performance from Fund Sponor’s Perspective
Performance shows the following:
- How effective IPS is
- Underperformance
- Results of active management
For feedback and control
Components of Performance Evaluation
- Performance measurement
- Performance attribution
- determines source of return
- Performance appraisal
- determine if its skill (should we fire a manager)
Return Calculation with Cash Flows
CF at beginning:
rt = EV - (BV + CF) / BV + CF
CF at end:
rt = (EV - CF) - BV0 / BV
TWR
Unaffected by external cash flows
Calculate returns subtracting cash flows then add together;
(1 + r)(1 + r)(1 +r) - 1
Example: 2.5M start, 2.7M end. 45K CF day 7, 25K CF day 19
Day 7 2.555M, day 19 $2.575M
(2,555,000-45,000) - 2,500,000 / 2,500,000 = 0.4%
(2,575,000-25,000) - 2,555,000 / 2,555,000 = -0.2%
(2,700,000) - 2,575,000 / 2,575,000 = 4.9%
(1 + .004)(1 - .002)(1 + .049) - 1 = 5.1%
MWRR
EV = BV(1 + r)n + CF(1 + r)n
Just plug in figures for r
TWRR vs. MWRR
- *TWR**
- *+** unaffected by timing of cash flows
- *+** required for GIPS
- *-** need to know valuations on each CF date
- *MWR**
- *+** good if manager controls CF
- *-** distorted by size and timing of CF
TWR vs MWRR
Which is better?
Cash Prior to _____ Performance Result
None None Same
+CF Strong MW > TW
+CF Weak TW > MW
- CF Strong TW > MW
- CF Weak MW > TW
Data Quality for Returns
- Stale price:
- Illiquid assets - use estimates
- Fixed income - use matrix
- Should include accrued interest and dividends
- Use trade date
Portfolio Return Components
P = Market + Style + Active
M = market return, Style = B - M, Active = P - B
Client responsible for S
Manager responsible for A
Example: LC Value Fund earns 18.9% for 3 quarters
Russell 1000 Value = 21.7%, Wilshire 5000 = 25.2%
Style = 21.7 - 25.2 = -3.5% Active = 18.9 - 21.7 = -2.8%
Valid Benchmark
SAMURAI
- Specified in advance
- Appropriate
- Measurable
- Unambiguous (Clear/Precise)
- Reflective of manager’s current investment opinons
- Accountable
- Investable
Types of Benchmarks
- Absolute return: i.e. 5%
- Drawback: NOT investable
- Peer group
- Drawback: subject to survivor bias, NOT investable
- Fails every quality test EXCEPT measurable
- Broad market index
- Drawback: manager style may deviate
- Style index
- Drawback: different definitions of style
- Factor-based models
- Return-based
- Custom
Testing Benchmark Quality
- Mutually exclusive with indices/asset classes
- Exhaustive of manager’s investment universe
- Represent distinct sources of risk
Hedge Fund Benchmarks
Hard to assign a benchmark. Use:
- Value-added return for each position
- Separate long/short benchmarks
- sharpe ratio often used but not appropriate
- Market neutral should be Rf
Macro Attribution Analysis Levels
Beginning Value
- Net contribution
- Rf
- Asset categories (pure indexing SAA from IPS)
- Benchmarks (pure indexing TAA from manager)
- Investment managers (active management)
- Allocation effects (error/plug number)
* *Ending Value**
Micro Attribution Components
Pure Sector Allocation
Definition: Sector deviation from benchmark
Formula:
(Wp,s - Wb,s) * (Rb,s - Rb)
Macro Attribution
Formulas
Asset Category: weight(B - Rf)
Style/Misfit: weight * weight * (misfit return)
Investment Manager: weight * weight * (true active return)
Micro Attribution Components
Within-sector Selection
Definition: Security selection deviation from benchmark
Formula:
(Wb,s) * [(Rp,s - Rb,s)]
Micro Attribution Components
Allocation/Selection Interaction
Definition: Sector deviation from benchmark
Formula:
(Wp,s - Wb,s) * (Rp,s - Rb,s)
All Micro Attribution Components
Pure Sector Allocation: (Wp,s - Wb,s) * (Rb,s - Rb)
Within sector selection: (Wb,s) * [(Rp,s - Rb,s)]
Allocation/Selection Interaction: (Wp,s - Wb,s) * (Rp,s - Rb,s)
Fundamental Factor Models for Micro Attribution
Regress historical returns vs factors. Help identify:
- source of portfolio returns
- sector rotation
- ability to time market
+ considers other factors + insight to investment style
- complex
Fixed Income Attribution Details
External Interest Rate Effect
- Just the benchmark
- Not under manager’s control
- Expected: simulated return on default-free benchmark
- Unexpected: return based on actual price changes
*This is what was earned passively*
Fixed Income Attribution Details
Manager Attribution
-
Interest rate management
- Duration, convexity, YC shape change
- Sector/quality management - weighting of sectors/quality
- Security-selection - effect on held securities
- Trading - plug figure (error/unknown)
*Must sum up to actual portfolio return*
5 Types of Risk-Adjusted Measures
- Jensen Alpha: Rp - CAPM [Rf + B(Rm - Rf)]
- Treynor: (Rp - Rf) / Bp
- Sharpe: (RP - Rf) / stdp
- M2: Rf + [(Rp - Rf / stdp) * stdm]
- IR: active return / active risk
Jesen Ex post Alpha
Rp - [Rf + B(Rm - Rf)]
- Cosiders only systematic risk (beta)
- Good for actively managed portfolios
Treynor
(Rp - Rf) / Bp
- Measures return over systematic risk (beta)
- Used to evaluate additions to portfolios
Sharpe Ratio
(RP - Rf) / stdp
- Excess return relative to total risk
- Assumes returns are normally distributed
M2 Measure
RF + [(RP - RF / stdp) * stdm]
^ Sharpe ratio
- Value added/lost if portfolio had same std as market
- Based on total risk
Information Ratio
active return / active risk
RP - RB / [stdp - stdB]
Quality Control Charts
- Presents manager performance over time
- Value added = 0 then goes above/below
- Over time acceptance range narrows
Quality Control Charts Errors
- Type I - retaining poor manager
- Type 2 = firing superior manager
Manager Continuation Policies
- Costly to hire/fire managers
- New managers will make changes adding costs
- Goals are to:
- Retain best managers
- Base decision off of more than just performance
- Apply consistent procedures
