Advanced Financial Mathematics Chapter 13 Market marking and delta hedging These notes are greatly inspired from the book Derivatives Markets Chapter 13-Advanced Financial Mathematics-UEH-F2023 2 Market makers • A market maker stands ready to buy from sellers and sell to buyers, ensuring a liquid market with quick/immediate transactions. • Proprietary trading operations are conceptually distinct from market making and are conducted by a financial institution for its own account, following an investment strategy. • Clients and proprietary traders of financial institutions profit from market movements, while market makers profit from transaction fees. Chapter 13-Advanced Financial Mathematics-UEH-F2023 3 Delta hedging • Without hedging, a market maker ends up with an arbitrary position based on customer orders. • Market makers can control their risk by engaging in delta hedging. • They calculate the delta of the option they have sold and take a perfectly opposite position in the underlying asset. • A delta-hedged position is not typically perfectly costless and requires capital. Chapter 13-Advanced Financial Mathematics-UEH-F2023 4 Option risk in the absence of hedging • We have already seen the concept that for every options bought/sold (traded), there are always two parties involved, often a long position versus a short position, but not always. • The risk is associated with one of the positions for the option or the strategy in question. • Let's suppose a client wants to buy a call option from a market maker, and the market maker sells the client one with: • S0 = K = 40$, π = 30 %, π = 8 %, π = 3 months ( 91 365 years) Chapter 13-Advanced Financial Mathematics-UEH-F2023 5 Chapter 13-Advanced Financial Mathematics-UEH-F2023 6 Detailed calculation • Call price: 91 πΆ π = 40, πΎ = 40, π = 0.3, π = 0.08, π − π‘ = ,πΏ = 0 365 = 40 π 91 −0∗365 π π1 − 40π 91 −0.08∗365 1∗0.58240416=βπΆ with π1 = ππ π π2 πΎ + π−πΏ+ 2 π (π−π‘) (π−π‘) = π π2 = 2.7804$ 0.5232266 ππ =0 40 40 + 0.32 91 0.08−0+ 2 365 0.3 91 365 and π2 = π1 − π (π − π‘) = 0.20804775 − 0.3 91 365 = 0.20804775 = 0.05825337 Chapter 13-Advanced Financial Mathematics-UEH-F2023 7 • Greeks: βπΆ = π −πΏ(π−π‘) π π1 = π Γπ = π −πΏ(π−π‘) π π1 ππ (π−π‘) = π 91 −0∗365 91 −0∗365 π 0.20804775 = 0.5824>0 π 0.20804775 40∗0.3 91 365 = 0.06515618>0 ππ = β― ≅ −0.017<0 • For the market maker: • β < 0: risk is high when the stock price increases. • π > 0: his position is profitable in time. Chapter 13-Advanced Financial Mathematics-UEH-F2023 8 Discussion • We assume that the stock price increases by $0.75 to reach $40.75. • Option price is revaluated: 91 πΆ π = 40,75, πΎ = 40, π = 0.3, π = 0.08, (π − π‘) = ,πΏ = 0 365 = 40.75 91 −0∗365 π π π1 − 40π 91 −0.08∗365 1∗0.63007814=βπΆ π2 with π1 = ππ π πΎ + π−πΏ+ 2 (π−π‘) π (π−π‘) = π π2 = 3.2352$ 0.57231299 =0.01857639 ππ 40.75 40 + 0.32 91 365 0.08−0+ 2 0.3 π2 = π1 − π (π − π‘) = 0.33206032 − 0.3 91 365 91 365 = 0.33206032 = 0.18226594 • The market maker has thus lost $2.7804 - $3.2352 = -$0.4548. • If one day had passed before reaching this new value, then we would have had a lower value due to the negative theta. However, in this scenario, we are only varying the underlying asset's value without the effect of time passing. Chapter 13-Advanced Financial Mathematics-UEH-F2023 9 Chapter 13-Advanced Financial Mathematics-UEH-F2023 10 Delta and Gamma as measures of exposure • If we had used the delta to approximate the new price of the call option, we would have obtained: New Call Price = $2.7804 (Call Price with S = $40) + $0.75 (Change in S) * 0.5824 (Delta) = $3.2172. • However, if we rely solely on the delta, the price change in the option would be estimated as $0.4368, which underestimates the actual change of $0.4548. • The reason for this discrepancy is that the delta was measured for an underlying asset with an exact value of $40. Once it reaches $40.75, the underlying asset has a delta of 0.6301, which is greater than 0.5824. • When the value of the underlying asset increases, the delta also increases for a call option. Therefore, you would need to continually recalculate the new delta (and integrate the delta for each increment in the underlying asset's value between $40 and $40.75). • In conclusion, the delta can be useful for small changes in the underlying asset's price, but for larger changes, it underestimates the change in the option price. Chapter 13-Advanced Financial Mathematics-UEH-F2023 11 Question How to address this issue without getting involved in complex calculations? • Solution : For larger variations, one will need to consider convexity with gamma, and thus: 0.0652 2 2.7804 + 0.75 ∗ 0.5824 + 0.75 = 3.2355$ 2 β option pric Variation πΆ square of variation 2 Γ of S π of S with S = 40$ 2 • Conclusion : The combination of delta and gamma provides a good approximation for moderate variations. Chapter 13-Advanced Financial Mathematics-UEH-F2023 12 Taylor approximation π′ π π π₯ =π π + 1! π ′′ π π₯−π + 2! (3) π π π₯−π 2+ 3! π₯−π 3 +β― • The delta overestimates the decrease in the option price when the underlying asset's value decreases. • In fact, delta will generally provide a slightly lower approximation for the option's value for any change in the underlying asset's value because it neglects all the 2nd, 3rd, 4th derivatives, and so on. Chapter 13-Advanced Financial Mathematics-UEH-F2023 13 Delta hedging • We revisit the example from page 385 of the reference book, section 13.3. • The market maker sells a call option and hedges the position with stocks. Based on the previous example, this means buying 0.5824 shares. • We will now examine the evolution of the market maker's position. Chapter 13-Advanced Financial Mathematics-UEH-F2023 14 Day 0 π = π/πππ • Selling a call option on 100 shares: • Each option (per unit of the underlying) is worth $2.7804. So, 100 * $2.7804 = $278.04 is received by the market maker. • The market maker buys 100 * 0.5824 = 58.24 shares for $58.24 * $40 = $2329.60. • The net initial cost is: • $2329.60 - $278.04 = $2051.56. Chapter 13-Advanced Financial Mathematics-UEH-F2023 15 Day 1 π = π/πππ • π = 40.50$ 1 365 • πΆ π 1 365 = 40.50$, πΎ = 40, π = 0.3, π = 0.08, π − π‘ = • For the market maker: 58.24 ∗ 40.50 − π π‘πππ π£πππ’π 90 ,πΏ 365 100 ∗ 3.0621 − 2051.56 ∗ π new option price 0.08 365 = 0 = 3.0621$ = 0.5003 • Hence, a profit about 0.50$ in one day Rebalancing the portfolio is necessary because the delta has changed! • The new delta is 2 π −πΏ(π−π‘) π π ππ π πΎ + π − πΏ + (π − π‘) 2 π π 2 =π 0.3 ππ 40.50 40 + 0.08 + 2 0.3 Chapter 13-Advanced Financial Mathematics-UEH-F2023 90 365 90 365 = π. ππππ 16 Day 1 π = π/πππ • The market maker needs 100 * (0.6142 - 0.5824) = 3.18 new shares, which requires an additional investment of 3.18 * $40.50 = $128.79. • When summing up the accumulated initial cost, the day's profit, and the purchase of new shares, we get: $2051.56 * π^(0.08/365) + 0.5003 + $128.79 = $2181.30. • In other words, if you wanted to enter into an identical position at π‘=1/365, you would have to buy 61.42 shares (with a delta of 0.6142 times 100 shares) at a price of $40.50 each and sell 100 call options at a price of $3.0621 each. • This would result in a new initial cost on day 1 of: 61.42 * $40.50 - 100 * $3.0621 = $2181.30. Chapter 13-Advanced Financial Mathematics-UEH-F2023 17 Day 2 π = π/πππ • At π‘ = 2/365: • The underlying asset's price is $39.25. • - The option price for the market maker is C S 2 365 = 39.25$, K = 40, σ = 0.3, r = 0.08, T − t = • For the market maker: ππ. ππ ∗ 39.25 − 100 ∗ 2.3282 − ππππ. ππ ∗ e • • • 0.08 365 89 , δ = 0 = 2.3282$ 365 = −3.8631 Value of their shares at day 2: $61.42 * $39.25. New value of the option at day 2: 100 * $2.3282. Interest on the cost of the position at the previous day for the market maker: $2181.30 * π^(0.08/365). • This results in a loss of approximately $3.8631 in one day. • Rebalancing the portfolio is required because the delta has changed! The new delta is: σ2 0.32 89 ln S K + r − δ + (T − t) ln 39.25 40 + 0.08 + 2 2 365 e−δ(T−t) N =N σ T 89 0.3 365 = 0.5311 • The market maker needs to sell 8.31 new shares to balance their position since the delta is negative. Chapter 13-Advanced Financial Mathematics-UEH-F2023 18 Day 2 π = π/πππ • From the sale of 8.31 shares, the market maker will receive 8.31 * $39.25 = $326.66. • When summing up the accumulated initial cost, the day's profit, and the repurchase of new shares, we get: ππππ. ππ ∗ 0.08 π 365 + −3.8631 − πππ. ππ = ππππ. ππ$ • This is the value of the position at this moment. • In other words, if you wanted to enter into an identical position at π‘=2/365, you would need to buy 53.11 shares (with a delta of 0.5311 times 100 shares) at a price of $39.25 each and sell 100 call options at a price of $2.3282 each. • This would result in a new initial cost on day 2 of: 53.11 * $39.25 - 100 * $2.3282 = $1851.66. Chapter 13-Advanced Financial Mathematics-UEH-F2023 19 Chapter 13-Advanced Financial Mathematics-UEH-F2023 20 Day 0 • Selling 100 call options: • In exchange, buying 58.24 shares with a loan of $2051.56 in addition to the $278.04 obtained from selling the call options. • Note: Our borrowing capacity is assumed to be equal to the value of the shares in our portfolio, which is $2051.56 on day 0. Chapter 13-Advanced Financial Mathematics-UEH-F2023 21 Jour 1 The portfolio value rises to $2181.30, and our borrowing capacity increases by $129.74 in one day. To remain delta-neutral (delta-hedged), you need to buy 3.18 shares at $40.50 each for a total of $128.79. You pay $0.45 in daily interest on the loan. The difference between $129.74 (borrowing capacity increase) - $128.79 (purchase of shares) - $0.45 (interest) = $0.50 is the daily profit. The market valuation profit is equal to the net cash flow of a 100% leveraged position. Market Valuation Profit = Change in the value of the underlying + Interest expenses for borrowed funds - Cash outflows for options. Remarks: - If the net cash flow is positive, you can pocket the amount. - If the net cash flow is negative, you need to inject funds. - A hedged portfolio that never requires fund injections is called self-financing. Chapter 13-Advanced Financial Mathematics-UEH-F2023 22 Delta hedging for multiple days The daily gains and losses result from the combined effect of three elements, always using the example of the call option. 1. Gamma: For large movements in the value of the underlying asset St, the market maker loses money. • For small movements in St, the market maker makes money. The larger the movement, the less adequate delta hedging becomes. • When St increases, β_πΆ increases, and the written call option loses value (for the market maker who has a short position in the option) faster than the proportion β_πΆ of shares that the market maker holds gains. • Conversely, when St decreases, β_πΆ decreases, and the written call option by the market maker gains value more slowly than the proportion β_πΆ of shares that the market maker holds loses. 2. Theta: The option loses value over time in absolute terms, which is advantageous for the market maker (who has a short position in the call option). 3. Interest: Interest expenses on borrowed funds must be paid. Chapter 13-Advanced Financial Mathematics-UEH-F2023 23 Chapter 13-Advanced Financial Mathematics-UEH-F2023 24 Underlying asset price movement and autofinancing portfolio • For a one-standard deviation (σ) move in the underlying asset's value (up or down), you will obtain Table 13.3 (p. 389 DM). • These movements resemble those in the forward tree model (term tree) from Chapter 10. • For movements of exactly one standard deviation, the portfolio will be self-financed. In Figure 13.2, this represents the two points where the curve crosses the axis, resulting in a zero daily profit (moderate underlying asset movement, neither too high nor too low). Lemma: If ππ = ππ−1 π πβ±π β, the delta-neutral portfolio is auto-financing. Chapter 13-Advanced Financial Mathematics-UEH-F2023 25 Mathematics of delta hedging • The delta (β), gamma (Γ), and theta (π) influence the profit on a delta-hedged (delta-neutral) position. • Note: The other components (π,π,πΏ) do not change, so there's no impact on vega, rho, and psi. Chapter 13-Advanced Financial Mathematics-UEH-F2023 26 Delta-gamma approximation πΆ(ππ‘ ) + πΆπππ πππππ (ππ‘+β − ππ‘ ) stock price variation ΓπΆ(ππ‘) ∗ βπΆ(ππ‘) + (ππ‘+β − ππ‘ ) ∗ ≅ πΆ(ππ‘+β ) 2 2 • En 13.2.3. on avait pris l’exemple suivant : 0.752 0.0652 2 2.7804 + 0.75 ∗ 0.5824 + = 3.2355$ βπΆ Variation Prix du call Variation Γπ prix de S avec S = 40$ prix de S 2 au carré • Et le véritable prix de l’option avec que le prix du sous-jacent soit passé de 40 à 40.75 est de 3.2352$. • Concrètement, l’approximation delta-gamma indique que (résultat des séries de Taylor) : • Remarques : • Ceci ignore l’effet du temps (l’impact de « h ») et fonctionnera donc pour un h petit! • L’approximation tient également pour une option de vente. • Le delta et le gamma dépendent du prix de l’action. Chapter 13-Advanced Financial Mathematics-UEH-F2023 27 Chapter 13-Advanced Financial Mathematics-UEH-F2023 28 Delta-Gamma-theta approximation Lemma: For small π = ππ‘+β - ππ‘ and h, we have πΆ ππ‘+β = πΆ ππ‘ + π ≅ πΆ ππ‘ , π − π‘ + π Call price stock price variation ∗ βπΆ ππ‘ ,π−π‘ Chapter 13-Advanced Financial Mathematics-UEH-F2023 + π 2 π€πΆ ππ‘,π−π‘ 2 + βπ(ππ‘,π−π‘) 29 Chapter 13-Advanced Financial Mathematics-UEH-F2023 30 Market maker profit • The value of the market maker's position at time t is βπΆ ππ‘ ,π−π‘ ∗ ππ‘ − πΆ ππ‘ , π − π‘ = βπ‘ ∗ ππ‘ − πΆ ππ‘ . • The change in value from t to t+h is: ≅πβ • βπ‘ ∗ ππ‘+β − ππ‘ − πΆ ππ‘+β − πΆ ππ‘ − βπ‘ ππ‘ − πΆ ππ‘ π πβ − 1 = βπ‘ ∗ ππ‘+β − ππ‘ − βπ‘ ∗ ππ‘+β − ππ‘ + ππ‘+β − ππ‘ − βπ‘ ππ‘ − πΆ ππ‘ π πβ − 1 ΓπΆ ππ‘ =− ∗ ππ‘+β − ππ‘ 2 2 +βπ ππ‘ ,π−π‘ 2 ΓπΆ ππ‘ ∗ + βπ 2 + βπ‘ ππ‘ − πΆ ππ‘ Chapter 13-Advanced Financial Mathematics-UEH-F2023 ππ‘ ,π−π‘ π πβ − 1 31 Remarques • This is the profit of the market maker when the price goes from ππ‘ to ππ‘+β within a time interval of length h. We can break down the last equation into three components: • Gamma : ΓπΆ ππ‘ − 2 • Theta : −βπ ∗ ππ‘+β − ππ‘ ππ‘ ,π−π‘ 2 ∗ 365 • Intérêts : − βπ‘ ππ‘ − πΆ ππ‘ π πβ − 1 Chapter 13-Advanced Financial Mathematics-UEH-F2023 32 Tableau 13.5 DM Delta-gamma-theta approximation for different values of ππ‘+β − ππ‘ Chapter 13-Advanced Financial Mathematics-UEH-F2023 33 Movement of one standard deviation. • We revisit the example of a one standard deviation change in the value of the underlying asset with ππ‘+β − ππ‘ 2 = π 2 = π 2 ππ‘2 β. • The profit is equal to: ΓπΆ ππ‘ 2 2 =− π ππ‘ + π ππ‘,π−π‘ β − βπ‘ ππ‘ − πΆ ππ‘ π πβ − 1 2 For exactly one standard deviation of variation, the profit should be approximately zero − ΓπΆ ππ‘ 2 π 2 ππ‘ 2 + π ππ‘ ,π−π‘ β − βπ‘ ππ‘ − πΆ ππ‘ πβ = 0. Chapter 13-Advanced Financial Mathematics-UEH-F2023 34 Black-Scholes PDE Theorem: Suppose that the stock price S follows the following geometric Brownian motion with drift r and volatility π, π. π. S solves the following SDE πππ‘ = πππ‘ ππ‘ + πππ‘ πππ‘ Then, the call option price satisfies the following Black-Scholes partial differential equation (PDE): π€πΆ ππ‘ 2 2 π ππ‘ + πππ‘ βπ‘ + π ππ‘,π−π‘ = ππΆ ππ‘ . 2 Chapter 13-Advanced Financial Mathematics-UEH-F2023 35 Frequency of portfolio rebalancing • In practice, transaction costs introduce frictions, meaning that portfolio rebalancing can be expensive every time the delta changes. • One can measure the impact of less frequent rebalancing: 1 π β,π = ππ‘ 2 π 2 ΓπΆ ππ‘ π₯π2 − 1 β 2 • With π₯π representing the number of standard deviations of the movement of S and π β,π being the return in period i, assuming an initially delta-neutral position, and h is the time between adjustments in the portfolio. • Here, we assume that the market maker has sold an option. Hence, 1 πππ[π β,π ] = π 2 π 2 Γh 2 2 • For daily rebalancing : 2 1 π2π2Γ πππ π 1 = ,1 2 365 365 • For hourly rebalancing : 24 πππ 24 π π=1 1 , 365∗24 1 = π=1 1 π2π2Γ 2 24 ∗ 365 πππ π 2 = 1 , 365 1 24 Chapter 13-Advanced Financial Mathematics-UEH-F2023 36 Delta-neutral hedigng in pratice • Delta-neutral hedging is not sufficient to completely eliminate risk. • It is possible to make a position gamma-neutral, but it requires derivative products. • Static replication of the option is possible, a strategy in which options are used to hedge options. • Protection against significant underlying asset movements is possible by buying insurance, i.e., purchasing deeply out-of-the-money call and put options. • The problem can be addressed by creating a new derivative product, such as a variance swap. Chapter 13-Advanced Financial Mathematics-UEH-F2023 37 Gamma-neutral • Retake Table 13.6 with π = 40, π = 0.3, π = 0.08, πΏ = 0. 3 πΆπππ πΎ = 40, π‘ = 12 4 πΆπππ πΎ = 45, π‘ = 12 Achat 1,2408 Call(K = 45), Vente d’un Call(K = 40) Prix (C) 2.7804 1.3584 -1.0993 Delta Δ 0.5825 0.3285 -0.1749 Gamma Γ 0.0651 0.0524 0 Vega π 0.0781 0.0831 0.0250 Theta π -0.0173 -0.0129 0.0013 Rho π 0.0513 0.0389 -0.0013 Chapter 13-Advanced Financial Mathematics-UEH-F2023 38 Discussion • If you have a sold option with a strike price (K) of 40 and you want to neutralize the gamma: −0,0651 ∗ 1 + 0,0524 ∗ ππππππ = 0 πππππ = 1,2408 calls to buy • Then, you seek to neutralize the delta: −0,5825 ∗ 1 + 0,3285 ∗ 1,2408 + ππππ‘ππππ ∗ 1 = 0 ππππ‘ππππ = 0,1749 shares to buy • Neutralizing gamma improves profitability, so why not do it? • Possible responses: • It requires taking positions in other options, which can become complex and often involves costs. • Market maker clients buy options that typically have positive gammas (recall, gamma is positive for both call and put options). • Therefore, market makers have negative gammas in aggregate. While a market maker could theoretically seek gamma neutrality, they certainly cannot all be gamma-neutral at the same time. Chapter 13-Advanced Financial Mathematics-UEH-F2023 39 Chapter 13-Advanced Financial Mathematics-UEH-F2023 40