Financial Modelling
Week 1
Hedge Funds:
- A private investment pool, open to
institutional or wealthy investors
• Exempt from much of SEC regulation,
they can pursue more speculative
strategies
- Richer people need less protection, as
they have more knowledge and more
money
- Focused on absolute returns
Hedge Fund Fees and Performance:
- Fees + performance fee (2/20 scheme, 2% of AUM, 20% of gains)
• The incentive fee can be modelled as a call option
- This may encourage excessive risk taking by manager
- High water mark
• If funds experience losses, incentive fee will only be paid once the losses are made up
- Encourages managers to shut down funds after poor performance, and start a new one
- Funds of funds
• Invests in other hedge funds
• Only works if underlying hedge funds have di erent strategies
• There is a dual layer of fees, meaning net returns are very low
- Hedge funds are correlated to stocks the most, followed by commodities. This e ect worsens
during recessions
- Hedge fund diversi cation strategies and alpha’s are disappearing
Hedge Fund Strategies
- Directional
• Bets on the direction of markets
(increase, decrease)
• Based on fundamental investment
approach
• Not market neutral (positive or negative
exposure)
- Non-directionals
• Exploits temporary misalignment in
prices
• Quantitative investment approach
• Strives to be market neutral
fi ff ff
,- Hedge fund alphas and betas
• Alpha is the abnormal return after adjusting for exposure
• No point to pay a fee for beta exposure that we can earn ourselves via an etf
- Alpha transfer (portable alpha)
• Earning a beta in one asset and alpha on another, by using future contracts
• The goal is to separate asset allocation (beta) from security allocation (alpha)
• Three steps:
- Invest where you have a skill, and can nd alpha
- Hedge systematic risk away to isolate alpha
- Establish exposure to desired asset class
by using index or ETF futures
Why do/did hedge funds do so
well?
- Mutual funds do not outperform the market
after fees, so this raises the question of why
hedge funds did so well. Reasons to the right
Lecture 1
- Model inputs are key, as a bad input will generate a bad model.
- One needs to focus on the needs that will be met by conducting a model before collecting data
and building the model
- Model Classi cations
• Empirical vs Theoretical
• Deterministic vs probability (stochastic)
- Deterministic does not have a random term (CAPM)
- Probabilistic does (regression)
• Discrete vs Continuous
• Cross sectional vs time-series vs panel model
- Cross sectional compares units at a certain point in time
- Time tracks one unit over time
- Panel uses multiple units over multiple periods
- Model Optimization
• 2 main approaches to optimization
- Analytical
• Use calculus, it is fast and allows to
calculate sensitivity of inputs
- Numerical
• Uses brute force, it is more time
consuming and does not guarantee
a unique solution
fi fi
, - Model Simulation (Monte Carlo simulation)
• Computer-based technique used to account for uncertainty in decision making
• It repeatedly samples model input variables to obtain the distribution of possible outputs
- Model risk
• The potential loss an institution can incur due to the decisions taken based on a model.
• Sources of model risk
- Speci cation: conceptual mistakes
- Implementation: programming aws
- Application: misinterpretation of outputs
- Environment: time-varying parameters that breakdown relationships
Week 2: Performance Evaluation
Intro to performance evaluation
- Sharpe ratio: measures return compensation per unit of total risk, with risk being sd
• It is the appropriate measure when the portfolio represents the entire investment of an
individual
• Slope is the CAL
• Pros
- Simple to compute
- Allows comparison between assets (since the benchmark is the rf)
- Intuitive interpretation
- Not a ected by leverage
• Cons
- Assumes returns follow normal distribution (not great to evaluate hedge funds)
- Does not distinguish between good and bad volatility (downside vs upside)
- Does not distinguish between systematic and idiosyncratic risk
- Doesn’t say much in itself, must be compared
- Sortino ratio: measures return compensation per unit of bad (downside risk)
• Same as Sharpe but only takes into account downside risk (returns below rf)
• Pros
- Distinguishes between good and bad volatility
- Does not assume that returns follow normal distribution
- Allows comparison across assets
- Intuitive interpretation
• Cons
- Noisier than Sharpe when distribution is normal (less observations used to compute sd)
- Does not distinguish between systematic and idiosyncratic risk
- Doesn’t say much in itself, must be compared
- Treynor ratio: measures return compensation per unit of systematic risk
• Appropriate when portfolio evaluated is part of a fully diversi ed portfolio
• Slope is the SML
• Pros
- Distinguished systematic and idiosyncratic
fffi fl fi
Week 1
Hedge Funds:
- A private investment pool, open to
institutional or wealthy investors
• Exempt from much of SEC regulation,
they can pursue more speculative
strategies
- Richer people need less protection, as
they have more knowledge and more
money
- Focused on absolute returns
Hedge Fund Fees and Performance:
- Fees + performance fee (2/20 scheme, 2% of AUM, 20% of gains)
• The incentive fee can be modelled as a call option
- This may encourage excessive risk taking by manager
- High water mark
• If funds experience losses, incentive fee will only be paid once the losses are made up
- Encourages managers to shut down funds after poor performance, and start a new one
- Funds of funds
• Invests in other hedge funds
• Only works if underlying hedge funds have di erent strategies
• There is a dual layer of fees, meaning net returns are very low
- Hedge funds are correlated to stocks the most, followed by commodities. This e ect worsens
during recessions
- Hedge fund diversi cation strategies and alpha’s are disappearing
Hedge Fund Strategies
- Directional
• Bets on the direction of markets
(increase, decrease)
• Based on fundamental investment
approach
• Not market neutral (positive or negative
exposure)
- Non-directionals
• Exploits temporary misalignment in
prices
• Quantitative investment approach
• Strives to be market neutral
fi ff ff
,- Hedge fund alphas and betas
• Alpha is the abnormal return after adjusting for exposure
• No point to pay a fee for beta exposure that we can earn ourselves via an etf
- Alpha transfer (portable alpha)
• Earning a beta in one asset and alpha on another, by using future contracts
• The goal is to separate asset allocation (beta) from security allocation (alpha)
• Three steps:
- Invest where you have a skill, and can nd alpha
- Hedge systematic risk away to isolate alpha
- Establish exposure to desired asset class
by using index or ETF futures
Why do/did hedge funds do so
well?
- Mutual funds do not outperform the market
after fees, so this raises the question of why
hedge funds did so well. Reasons to the right
Lecture 1
- Model inputs are key, as a bad input will generate a bad model.
- One needs to focus on the needs that will be met by conducting a model before collecting data
and building the model
- Model Classi cations
• Empirical vs Theoretical
• Deterministic vs probability (stochastic)
- Deterministic does not have a random term (CAPM)
- Probabilistic does (regression)
• Discrete vs Continuous
• Cross sectional vs time-series vs panel model
- Cross sectional compares units at a certain point in time
- Time tracks one unit over time
- Panel uses multiple units over multiple periods
- Model Optimization
• 2 main approaches to optimization
- Analytical
• Use calculus, it is fast and allows to
calculate sensitivity of inputs
- Numerical
• Uses brute force, it is more time
consuming and does not guarantee
a unique solution
fi fi
, - Model Simulation (Monte Carlo simulation)
• Computer-based technique used to account for uncertainty in decision making
• It repeatedly samples model input variables to obtain the distribution of possible outputs
- Model risk
• The potential loss an institution can incur due to the decisions taken based on a model.
• Sources of model risk
- Speci cation: conceptual mistakes
- Implementation: programming aws
- Application: misinterpretation of outputs
- Environment: time-varying parameters that breakdown relationships
Week 2: Performance Evaluation
Intro to performance evaluation
- Sharpe ratio: measures return compensation per unit of total risk, with risk being sd
• It is the appropriate measure when the portfolio represents the entire investment of an
individual
• Slope is the CAL
• Pros
- Simple to compute
- Allows comparison between assets (since the benchmark is the rf)
- Intuitive interpretation
- Not a ected by leverage
• Cons
- Assumes returns follow normal distribution (not great to evaluate hedge funds)
- Does not distinguish between good and bad volatility (downside vs upside)
- Does not distinguish between systematic and idiosyncratic risk
- Doesn’t say much in itself, must be compared
- Sortino ratio: measures return compensation per unit of bad (downside risk)
• Same as Sharpe but only takes into account downside risk (returns below rf)
• Pros
- Distinguishes between good and bad volatility
- Does not assume that returns follow normal distribution
- Allows comparison across assets
- Intuitive interpretation
• Cons
- Noisier than Sharpe when distribution is normal (less observations used to compute sd)
- Does not distinguish between systematic and idiosyncratic risk
- Doesn’t say much in itself, must be compared
- Treynor ratio: measures return compensation per unit of systematic risk
• Appropriate when portfolio evaluated is part of a fully diversi ed portfolio
• Slope is the SML
• Pros
- Distinguished systematic and idiosyncratic
fffi fl fi
