Interview Revision for Quant Positions

Amit Kumar Jha

Stochastic Calculus
Basic Concepts

1. Brownian Motion (Wiener Process):

A stochastic process Bt where t ≥ 0, with the following properties: ​

1. B0 = 0. ​

2. Bt has independent increments.

​

3. Bt − Bs ∼ N (0, t − s) for 0 ≤ s < t.

​ ​

4. Bt has continuous paths.

​

2. Itô's Lemma:
If f (t, Bt ) is a twice-differentiable function where Bt is a Brownian motion, then
​ ​

∂f ∂f 1 ∂2f 2 ∂f
df (t, Bt ) = ( + μ+ 2
σ ) dt + σdBt
∂t ∂x 2 ∂x ∂x
​ ​ ​ ​ ​ ​ ​

For f (t, Bt ): ​

∂f 1 ∂2f ∂f
df (t, Bt ) = ( + 2
) dt + dBt
∂t 2 ∂Bt ∂Bt
​ ​ ​ ​ ​ ​

​ ​

Stochastic Integrals

1. Itô Integral:
t
Given a process Xt , the Itô integral ∫0 ​ Xs dBs is defined as the limit of sums of the form
​ ​

∑ Xti (Bti+1 − Bti ).

​
​

​
​

​
​

Properties:
t t t
1. Linearity: ∫0 (aXs ​ ​ + bYs ) dBs = a ∫0 Xs dBs + b ∫0 Ys dBs .
​ ​ ​ ​ ​ ​ ​ ​

2
2. E [(∫0 Xs dBs ) ] = E [∫0 Xs2 ds].
t t
​ ​ ​ ​ ​

2. Stratonovich Integral:
t
Defined similarly to the Itô integral but using the midpoint: ∫0 ​
Xs ∘ dBs =
​ ​

limn→∞ ∑ X(ti +ti+1 )/2 (Bti+1 − Bti ).

​

​ ​
​

​
​

​
​

t t
It is related to the Itô integral by the correction term: ∫0 ​ Xs ∘ dBs = ∫0 Xs dBs +
​ ​ ​ ​ ​

1 t
2
​
∫0 ∂X
​
s
∂Bs ​
​

​ ds.

Advanced Concepts

1. Stochastic Differential Equations (SDEs):

General form: dXt = μ(Xt , t)dt + σ(Xt , t)dBt . ​ ​ ​

Solution Methods:
1. Analytical solutions for simple cases.
2. Numerical solutions using discretization methods like Euler-Maruyama and Milstein.

Tricks for Memorization

Mnemonic for Itô's Lemma: "Partial time, half double space, full single space."
Link Brownian Motion Properties: "Zero start, independent steps, normal increments,
continuous path."

Printed using Save ChatGPT as PDF, powered by PDFCrowd HTML to PDF API. 5/10
 Discretization Techniques

1. Euler-Maruyama Method:
For dXt = μ(Xt , t)dt + σ(Xt , t)dBt :
​ ​ ​ ​

Xt+Δt = Xt + μ(Xt , t)Δt + σ(Xt , t)ΔBt ,

​ ​ ​ ​ ​

where ΔBt ​ ∼ N (0, Δt).

2. Milstein Method:
Improves Euler-Maruyama by including terms involving the derivative of σ :

1 ∂σ(Xt , t)
Xt+Δt = Xt + μ(Xt , t)Δt + σ(Xt , t)ΔBt + σ(Xt , t) ((ΔBt )2 − Δt) .
​

2 ∂Xt
​ ​ ​ ​ ​ ​ ​ ​ ​

Financial Math: Option Pricing Models

1. Binomial Model:
Assumptions: Discrete time steps, two possible outcomes per step.
Formula: Price of option is the discounted expectation of future payoffs.
Steps:
1. Construct binomial tree.
2. Calculate option value at each node using backward induction.

Ci,j = e−rΔt (pCi+1,j+1 + (1 − p)Ci+1,j )

​ ​ ​

e(r−q)Δt −d Δt ,
where p = u−d
,​ u = eσ ​

d = e−σ Δt . ​

2. Multinomial Model:
Extension of binomial with multiple possible outcomes per step.

Ci,j = e−rΔt ∑ pk Ci+1,j+k ​ ​ ​

where pk are the transition probabilities.

​

3. Black-Scholes Model:
Assumptions: Log-normal price distribution, constant volatility, no dividends.
Formula:

C = S0 N (d1 ) − Ke−rT N (d2 ),

​ ​ ​

ln(S0 /K)+(r+σ 2 /2)T

where d1 = and d2 = d1 − σ T .
​

​ ​ ​ ​ ​

σ T ​

Concepts:
Normal Distribution: Used for modeling continuous price changes.
Cumulative Distribution Function (CDF) N (d): Probability that a standard normal
variable is less than d.
Greeks: Sensitivities of option price to various parameters.
Delta (Δ): Rate of change of option price with respect to the price of the
underlying asset.

Printed using Save ChatGPT as PDF, powered by PDFCrowd HTML to PDF API. 6/10
 Gamma (Γ): Rate of change of Delta with respect to the price of the underlying
asset.
Vega (ν ): Sensitivity to volatility.
Theta (Θ): Sensitivity to time decay.
Rho (ρ): Sensitivity to interest rate changes.
4. Local Volatility Model:
Volatility is a function of both time and the underlying asset price.
Dupire's Formula:
∂C
2 ∂t
+ rS ∂C
∂S
+ qSC − qKC
(t, S) =
​ ​

σlocal​

1 2 ∂2C
​

2 S ∂S 2
​ ​

where C is the market price of a European call option.

5. Stochastic Volatility Model:
Example: Heston model, where volatility follows its own stochastic process.

dSt = μSt dt +
​ ​
vt St dWt1
​ ​ ​ ​

dvt = κ(θ − vt )dt + σ vt dWt2

​ ​ ​ ​ ​

where dWt1 and dWt2 are correlated Brownian motions with correlation coefficient ρ.
​ ​

6. Levy Jump Model:

Incorporates jumps in price process.
Assumptions: Asset prices follow a jump-diffusion process.

dSt = μSt dt + σSt dBt + St dJt

​ ​ ​ ​ ​ ​

where Jt represents the jump process.

​

Probability Concepts

1. Basic Concepts:
Probability Distribution: Describes likelihood of outcomes.
Expected Value: Mean of a random variable.
∞
E[X] = ∑ xP (X = x) (discrete),
​
E[X] = ∫ ​ xfX (x) dx (continuous)
​

x −∞

Variance and Standard Deviation: Measure spread of a distribution.

Var(X) = E[(X − E[X])2 ] = E[X 2 ] − (E[X])2 , σX = ​ Var(X) ​

Normal Distribution: Bell-shaped curve, characterized by mean (μ) and variance (σ 2 ).

1 (x−μ)2
fX (x) = e− 2σ2 ​

2πσ 2
​ ​

Printed using Save ChatGPT as PDF, powered by PDFCrowd HTML to PDF API. 7/10
 2. Advanced Techniques:
Law of Total Probability:

P (A) = ∑ P (A∣Bi )P (Bi ) ​ ​

Bayes' Theorem:

P (B∣A)P (A)
P (A∣B) =
P (B)
​

Central Limit Theorem (CLT):

Xn − μ d
→ N (0, 1)
​

σ/ n ​

Moment Generating Functions (MGFs):

MX (t) = E[etX ]
​

Risk Management Concepts

1. Value at Risk (VaR):

Definition: Maximum loss over a specified period within a given confidence level.
Methods:
Historical Simulation: Use historical data to simulate portfolio returns.
Variance-Covariance: Assumes returns are normally distributed.

VaR = Φ−1 (α)σp T − μp T ​ ​ ​

Monte Carlo Simulation: Simulate returns based on assumed distributions.

2. Expected Shortfall (ES):
Average loss beyond the VaR threshold.

ESα = −E[R∣R ≤ VaRα ]

​ ​

3. Exposure Measures:
Expected Exposure (EE): Average exposure over time.

EE(t) = E[Exposure(t)]

Potential Future Exposure (PFE): Maximum exposure at a future point with a certain
confidence.
Maximum Likely Exposure (MLE): Highest exposure at any time.
Effective Expected Positive Exposure (EEPE): Time-weighted average of EE.
Expected Negative Exposure (ENE): Similar to EEPE but for negative exposure.
Risk-Weighted Assets (RWA): Regulatory measure for risk exposure.

Project Revision Notes

1. Valuation of European Put Option Using Finite Differencing Method:

Explicit Method:

Vi,j = αVi+1,j+1 + βVi,j+1 + γVi−1,j+1 ,

​ ​ ​ ​

Printed using Save ChatGPT as PDF, powered by PDFCrowd HTML to PDF API. 8/10
 where coefficients depend on discretization.
Implicit Method:
Requires solving a system of linear equations.
Crank-Nicholson Scheme:
Combines explicit and implicit methods for stability.

Vi,j+1 − Vi,j 1
= [αVi+1,j + βVi,j + γVi−1,j ]
​ ​

Δt 2
​ ​ ​ ​ ​

2. Monte Carlo Simulations:

Heston Model: Simulate paths for both asset price and variance.
vt
St+Δt = St exp [(r − ) Δt + vt ΔBt1 ]
​

2
​ ​ ​ ​ ​ ​

vt+Δt = vt + κ(θ − vt )Δt + σ vt ΔBt2

​ ​ ​ ​ ​ ​

Merton Jump Diffusion Model: Incorporate jumps in price paths.

σ2
St+Δt = St exp [(μ − λk − ) Δt + σΔBt + J ]
2
​ ​ ​ ​

where J represents the jump component.

3. Interest Rate Predictions Using Stochastic Model Simulations:
Vasicek Model: drt = a(b − rt )dt + σdBt .
​ ​ ​

rt+Δt = rt + a(b − rt )Δt + σ Δtϵt

​ ​ ​ ​ ​

Hull-White Model: Extension of Vasicek with time-dependent parameters.

drt = (θ(t) − art )dt + σdBt

​ ​ ​

Cox-Ingersoll-Ross (CIR) Model: drt ​ = a(b − rt )dt + σ rt dBt . ​ ​ ​

rt+Δt = rt + a(b − rt )Δt + σ

​ ​ ​ rt Δ t ϵ t ​ ​ ​

Black-Derman-Toy (BDT) Model: Lognormal interest rate model.

4. Valuation of a Credit Default Swap (CDS) Under Constant Intensity Model:
Steps:
1. Estimate default probability.
2. Calculate expected payments and recoveries.

∑i P (τ > ti )Δti
CDS spread =
​ ​ ​

∑i P (τ > ti )Δti ⋅ (1 − R)
​

​ ​ ​

where τ is the default time and R is the recovery rate.

5. Pricing FX Options:
Black-Scholes Model: Apply to foreign exchange context.

C = S0 e−qT N (d1 ) − Ke−rT N (d2 ) ​ ​ ​

Printed using Save ChatGPT as PDF, powered by PDFCrowd HTML to PDF API. 9/10
 ln(S0 /K)+(r−q+σ 2 /2)T
where d1 ​ = ​

σ T ​
and​ d2 = d1 − σ T .
​ ​ ​

PDE Approach: Solve differential equation governing option price.

∂C 1 ∂2C ∂C
+ σ 2 S 2 2 + (r − q)S − rC = 0
∂t 2 ∂S ∂S
​ ​ ​ ​

Quadrature Method: Numerical integration for option pricing.

∞
C=e −rT
∫ ​ max(ST − K, 0)f (ST )dST​ ​ ​

−∞

Monte Carlo Simulation: Simulate exchange rate paths and average payoffs.

N
1
C=e −rT
∑ max(STi − K, 0)
​ ​ ​

N i=1

Revision Tips
Thanks for reading !!
Organize: Group similar topics together for better retention.
Practice Problems: Solve as many problems as possible to understand concepts deeply.
Mnemonics: Use memory aids to remember complex formulas.
Flashcards: Create flashcards for key concepts and formulas.
Regular Review: Schedule regular reviews to reinforce learning.

Let me know if you need further details on any specific topic!

ChatGPT can ant info.

Printed using Save ChatGPT as PDF, powered by PDFCrowd HTML to PDF API. 10/10