금융수학 101 송성주, 고려대학교 통계학과 What is Math Finance/ Financial Engineering? History of Math Finance Issues Basics of Derivative Pricing and Hedging Black-Scholes Model Risk Measure 3/13/2009 1 Mathematical Finance/Financial Engineering From Wikipedia: Mathematical finance is the branch of applied mathematics concerned with the financial markets. Mathematical finance will derive and extend the mathematical or numerical models suggested by financial economics. While a financial economist might study the structural reasons why a company may have a certain share price, a financial mathematician may take the share price as a given, and attempt to use stochastic calculus to obtain the fair value of derivatives of the stock. 3/13/2009 2 Mathematical Finance/Financial Engineering Mathematical finance also overlaps heavily with the fields of financial engineering and computational finance. Arguably, all three are largely synonymous, although the latter two focus on application, while the former focuses on modeling and derivation. Financial engineering applies mathematical finance results, along with techniques such as statistical data analysis, mathematical optimization, stochastic process modeling, and MC simulation. Many universities around the world now offer degree and research programs in mathematical finance. 3/13/2009 3 Origin of Mathematical Finance Louis Bachelier's doctoral dissertation: “Theorie de la speculation”(1900) Application of probability to stock markets Main objective: valuation of options Historically the first attempt to use advanced mathematics in finance Witnessed the introduction of Brownian motion: Wiener-Bachelier process (Feller, 1966) Rediscovered in 50s (Savage, 1954; Samuelson, 1965) 3/13/2009 4 Subsequent research in Math. Finance Wiener, Kolmogorov, Ito, Doob, Meyer, Kunita and Watanabe... Construction of BM, Stochastic integration, Ito formula, Doob-Meyer decomposition, martingale representation theorem, ... Markowitz, Sharpe (1990 Nobel Prize) Mean-variance analysis, Capital Asset Pricing Model Savage, Samuelson , Fama, McKean… Martingale, Efficient market hypothesis, Geometric BM, Option valuation 3/13/2009 5 Subsequent research in Math Finance Merton, Black, Scholes (1973; 1997 Nobel prize) Option pricing formula, Perfect replication, Arbitrage-free pricing Harrison, Kreps, Pliska (1979, 1981) Close link with martingale theory has been established/ Fair price of the an option is the expected discounted payoff of the option under an equivalent martingale measure 3/13/2009 6 Issues Derivative Pricing and Hedging Portfolio Optimization Risk Management Others: Modeling, Volatility, Inference, Computation... Mathematical Tools: Stochastic Calculus, Martingale theory, Stochastic control, PDE, Numerical methods … Statistical contribution: Model fitting, Time series (ARCH, GARCH), Inference for stochastic processes, Measuring/Modeling the risk, Statistical data analysis … 3/13/2009 7 Portfolio Optimization An economic agent decides how much to consume, how much to save, and how to allocate his capital between different assets in order to maximize his expected utility. Optimal solution depends on the mean return rates of the risky assets, which are difficult to estimate, and also depends heavily on the kind of utility function we choose. 3/13/2009 8 Portfolio Optimization Optimal Mean-variance portfolio (Markowitz) Minimize the variance with given mean value or maximize the mean value with given variance, usually considered in a single-period framework Optimal consumption/asset allocation for utility maximization Lagrange Multipliers (Single-period) Dynamic Programming (Multiperiod, Backward Algorithm) Hamilton-Jacobi-Bellman PDE (Continuous time) 3/13/2009 9 Basics of Derivative Pricing Derivative Securities: Forward, Future, Swap, Option (Call, Put)… Forward: an agreement to buy or sell an asset at a certain future time for a certain price. Future: an agreement to buy or sell an asset at a certain time in the future for a certain price, traded on an exchange. Swap: an agreement to exchange two cash flows with different features. eg: fixed interest rate vs. floating interest rate 3/13/2009 10 Basics of Derivative Pricing Option: gives the owner the right to buy or sell a certain asset at a pre-specified price, on or before a pre-specified date. Arbitrage-free market Forward Price ( F S 0 e ) : delivery price r (i) F S 0 e : long forward, shortsell stock, and invest S 0 to moneymarket r (ii) F S 0 e moneymarket 3/13/2009 r : short forward, buy stock, and borrow S 0 from 11 European call option price: single-period binomial model (Cox, Ross, and Rubinstein 1979) Conditions: S1d K S1u , S1d S0 e r S1u , S1u K a 0 a 1 S1u S1d v a( S0 S1d e r ) : price 3/13/2009 12 Arbitrage-free pricing If V0 v, buy a call, shortsell a shares of stock, and invest aS0 V0 in the moneymarket. If V0 v, S1 S1u ( S1u K ) aS1u (aS0 V0 )e r ( S1u K ) aS1u (aS0 v)e r ( S1u K ) aS1u aS1d ( S1u K ) a( S1u S1d ) 0 If V0 v, S1 S1d aS1d (aS0 V0 )er aS1d (aS0 v)e r aS1d aS1d 0 3/13/2009 13 Arbitrage-free pricing (cont’d) If V0 v, sell a call, buy a shares of stock, and borrow aS0 V0 from the moneymarket. If V0 v, S1 S1u ( K S1u ) aS1u (aS0 V0 )e r ( K S1u ) aS1u (aS0 v)e r ( K S1u ) aS1u aS1d ( K S1u ) a( S1u S1d ) 0 If V0 v, S1 S1d aS1d (aS0 V0 )er aS1d (aS0 v)er aS1d aS1d 0 3/13/2009 14 Risk-Neutral Measure Equivalent martingale measure Pricing with risk-neutral measure r S e 0 S1d r V0 e (S1u K ) S1u S1d er ( p(S1u K ) (1 p)(S1d K ) ) E (er (S1 K ) ), S0 e r S1d p S S 1u 1d S0 e r ( pS1u (1 p ) S1d ) E (e r S1 ) 3/13/2009 15 Martingales Fair game, one of main tools in the study of random processes Expected fortune of a gambler at time n given all the information in the first n-1 trials is the same as the actual value of the gambler’s fortune before the nth round. S0 E (e rt St ), e rt St E (e rT ST | all information up to time t) V0 E (e rtVt ) e rtVt E (e rT ( ST K ) | all information up to time t) 3/13/2009 16 Hedging Reducing the financial risk by trading available securities Hedging strategy: how many shares of which securities we should hold at each time How Black and Scholes originally obtained the pricing formula Completeness/Incompleteness of markets Replication(Perfect Hedge), Quadratic Hedge (Mean-variance, Local risk minimization), Super(conservative) Hedge, Quantile Hedge, Efficient Hedge, ... 3/13/2009 17 Black-Scholes Model Options can be completely hedged and their prices can be uniquely determined under some assumptions. Assumptions: Underlying asset ~ geometric Brownian Motion. Constant volatility, constant expected rate of return No transaction costs or taxes, all securities are perfectly divisible No dividends No arbitrage opportunities Continuous trading Short rate is constant 3/13/2009 18 Black-Scholes formula European Call option C ( x, t ) x( z ) e r (T t ) K ( z T t ) x log( x / K ) (r 2 / 2)(T t ) where z T t Needs volatility input Perfect hedging strategy: Hold Cx ( St , t ) shares of stock at any time t T : Delta hedging Not a good fit to the real data: kurtosis, skewness, volatility smile 3/13/2009 19 Financial risk Financial risk: Market risk, Credit risk, Operational risk, Liquidity risk Market risk: Risk that the value of an investment will decrease due to market movements Credit risk: Risk of loss due to counter-party defaulting on a contract Operational risk(Basel II Accord): Risk of losses resulting from inadequate or failed internal processes, people and systems, or external events Liquidity risk: Additional risk in the market due to the timing and size of a trade 3/13/2009 20 Risk Measure: Value at Risk Value at Risk: the most prominent risk measure (J.P. Morgan's RiskMetrics,1993) Given a risk X with cumulative distribution function FX and a probability level (0,1) , VaR ( X ) FX1 ( ) inf{x R : FX ( x) }. 3/13/2009 Does not measure the potential size of a loss, lacks the property of subadditivity, CDF unknown in practice 21 Coherent Risk Measure Artzner, Delbaen, Eber and Heath (1999) ( X c) c ( X ) Translation invariance: Positive homogeneity: Monotonicity: X Y a.s. ( X ) (Y ) Subadditivity: ( X Y ) ( X ) (Y ) 3/13/2009 (cX ) c ( X ), c 0 22 Expected tail loss Expected shortfall, Tail conditional expectation, TailVaR ES ( X ) E ( X | X VaR ( X )) Well-known in insurance: ETL=Mean excess loss+ VaR ( X ) For continuous risks, ETL is a coherent risk measure 3/13/2009 23 Estimation of VaR or ETL Assume normality (Variance-Covariance method) Historical simulation Monte Carlo simulation Asymptotic distribution of order statistics Quantile regression Econometric approach Extreme value approach 3/13/2009 24