← Citibank Interview Insights

Citibank·Data Scientist·Technical Phone Screen·Intermediate

IntermediatePrefer not to say
Jul 2026

Summary

Quant-heavy technical screen for a Data Scientist role at Citibank. The whole thing was binomial tree pricing, both equity options and fixed income, which I wasn't expecting to be this deep for a DS position. Left feeling like I'd done okay on the mechanics but probably fumbled the conceptual explanation on the American call question.

Questions Asked (5)

Q1

Given a two-period binomial stock model with S0=100, up factor 1.25, down factor 0.8, and a 5% per-period risk-free rate, compute the risk-neutral probability and price a European call with strike 100 using backward induction.

Algorithms & Data StructuresTechnical Trade-offs
Author's notes

The risk-neutral probability calculation itself is fine, just plug into the formula.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, compute the risk-neutral probability using the formula p = (e^{rΔt} - d) / (u - d), ensuring consistency with the given per-period risk-free rate. Then, construct the binomial tree for the stock price, calculate the option payoff at maturity, and work backward to discount expected payoffs at each node to find the initial call price.

Pro tip: Always verify that the risk-neutral probability lies between 0 and 1; if not, check for arbitrage or input errors. Also, mention that the same risk-neutral probability applies to all nodes, simplifying calculations.

1. Compute risk-neutral probability

Use the formula p = (e^{rΔt} - d) / (u - d) with r = 0.05, Δt = 1, u = 1.25, d = 0.8. Calculate p and verify 0 < p < 1.

2. Build stock price tree

Calculate stock prices at each node: S0=100, up to 125 or down to 80 after one period, and then to 156.25, 100, or 64 after two periods.

3. Compute option payoffs at maturity

At each final node, compute call payoff as max(S_T - K, 0) with K=100. Payoffs: max(156.25-100,0)=56.25, max(100-100,0)=0, max(64-100,0)=0.

4. Backward induction for option value

At each earlier node, compute option value as e^{-rΔt} [p * V_up + (1-p) * V_down]. Start from period 1 nodes and then period 0.

5. State final call price

After backward induction, the initial call price is the value at the root node. Present the numerical result clearly.

Key Points to Mention

  • Risk-neutral probability formula and its derivation from no-arbitrage condition
  • Assumption of continuous compounding for risk-free rate (or clarify if simple compounding is used)
  • Construction of the binomial tree for stock prices
  • Option payoff calculation at maturity for a European call
  • Backward induction method: discounting expected payoffs at each node
  • Verification that the risk-neutral probability is between 0 and 1

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q2

Using the same binomial tree, price an American put with strike 100 and identify any nodes where early exercise is optimal.

Algorithms & Data StructuresTechnical Trade-offs
Author's notes

You have to compare the hold value against the intrinsic value at every node, not just the terminal ones.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, clarify the binomial tree parameters (u, d, p, r, Δt) and then compute the American put value via backward induction, comparing intrinsic value and continuation value at each node. Identify nodes where intrinsic value exceeds continuation value, as those are optimal early exercise points.

Pro tip: Emphasize that early exercise for an American put is only optimal when the stock price is sufficiently low, and that the early exercise boundary can be characterized by a critical stock price at each time step. This shows deeper understanding beyond just computing values.

1. Clarify tree parameters

Confirm the binomial tree parameters: up factor u, down factor d, risk-neutral probability p, risk-free rate r, and time step Δt. If not given, state typical assumptions (e.g., u=1/d, p=(e^{rΔt}-d)/(u-d)).

2. Compute terminal payoffs

At maturity, compute the put payoff max(100 - S_T, 0) for each terminal node.

3. Backward induction with early exercise

At each node, compute the continuation value as the discounted expected value of the two child nodes. Compare with intrinsic value max(100 - S, 0). The American put value is the maximum of the two.

4. Identify early exercise nodes

Mark nodes where intrinsic value > continuation value. These are the nodes where early exercise is optimal. Optionally, describe the boundary of these nodes.

5. Summarize and interpret

State the final American put price and list the early exercise nodes. Explain that early exercise occurs when the stock price is low enough that the benefit of immediate exercise outweighs waiting.

Key Points to Mention

  • American options can be exercised at any time before maturity, unlike European options.
  • Backward induction is used to price American options, incorporating the possibility of early exercise at each node.
  • Early exercise is optimal when the intrinsic value (immediate payoff) exceeds the continuation value (expected discounted future payoff).
  • For a put option, early exercise is more likely when the stock price is low, as the time value of waiting is small and the benefit of receiving the strike price sooner is high.
  • The early exercise boundary separates nodes where exercise is optimal from those where holding is better.
  • The binomial tree must be consistent with risk-neutral valuation to avoid arbitrage.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q3

Prove whether early exercise of an American call on a non-dividend-paying stock can ever be optimal in this binomial framework, and explain the intuition.

Technical Trade-offsPricing & Monetization
Author's notes

I knew the answer is no, and I knew it had something to do with time value, but my explanation came out muddled.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Set up a one-period binomial model for a non-dividend-paying stock and compare the value of immediate exercise with the value of holding the option. Show that the no-arbitrage price of the American call equals the European call price, so early exercise is never optimal. Explain the intuition using the time value of money and the insurance value of waiting.

Pro tip: Emphasize that the result holds because the stock pays no dividends; if dividends were present, early exercise might be optimal to capture the dividend. Also, mention that this is a model-independent result under no-arbitrage, not just a binomial artifact.

1. Set up the binomial model

Define a one-period binomial tree with stock price S, up factor u, down factor d, risk-free rate r, and strike price K. Assume no dividends.

2. Compare exercise vs. holding

At any node, compare the intrinsic value (S - K) with the continuation value (risk-neutral expected discounted payoff). Show that continuation value is always at least as large as intrinsic value.

3. Prove inequality

Use the risk-neutral pricing formula to show that C >= S - K for all states, hence early exercise is never strictly optimal. This can be done by constructing a dominating portfolio or using Jensen's inequality.

4. Extend to multi-period

Argue by backward induction that if early exercise is not optimal at any final step, it is not optimal at any earlier step, so the American call price equals the European call price.

5. Explain intuition

Discuss that waiting preserves the option's time value and insurance against downside risk, while exercising forfeits these benefits and requires paying K earlier, incurring interest cost.

Key Points to Mention

  • No-arbitrage principle and risk-neutral valuation
  • Time value of money: delaying payment of strike price K
  • Insurance value: keeping the option alive protects against downside
  • Comparison of intrinsic value vs. continuation value
  • Backward induction in binomial trees
  • Dividend-paying stocks as a contrast where early exercise may be optimal

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q4

Using a two-period short-rate tree with given up and down rates and risk-neutral probability of 0.5, price a zero-coupon bond maturing at t=2 with face value 100.

Algorithms & Data Structures
Author's notes

Straightforward backward induction again.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Use backward induction on the short-rate tree: first compute the bond price at t=1 in both up and down states by discounting the face value at the respective short rates, then compute the t=0 price by discounting the risk-neutral expected value of the t=1 prices at the initial short rate. Clearly state the given rates and probability, and show the calculations step by step.

Pro tip: Emphasize that this is a risk-neutral valuation, so the expected value uses the risk-neutral probability, not real-world probabilities. Also, mention that the same tree can be used to price derivatives like options, showing broader applicability.

1. Identify given parameters

List the up and down short rates for each period (r0, r1_up, r1_down) and the risk-neutral probability p=0.5. Note the face value F=100 and maturity T=2.

2. Compute bond prices at t=1

In the up state, the bond price is F/(1+r1_up). In the down state, it is F/(1+r1_down). These are the values at t=1 if the bond matures at t=2.

3. Compute expected value at t=0

Calculate the risk-neutral expected value at t=0: E[P1] = p * P1_up + (1-p) * P1_down.

4. Discount to t=0

Discount the expected value by the initial short rate r0: P0 = E[P1] / (1+r0). This is the final price of the zero-coupon bond.

5. Verify and interpret

Check that the price is less than the face value (since rates are positive) and discuss any assumptions (e.g., no arbitrage, constant risk-neutral probability).

Key Points to Mention

  • Risk-neutral valuation framework
  • Backward induction on the tree
  • Discounting at each node using the short rate
  • Use of risk-neutral probability (0.5) for expected value
  • Zero-coupon bond pays face value only at maturity
  • No-arbitrage assumption ensures unique price

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q5

Price a two-period coupon bond with face 100 and coupon 5 paid at t=1 and t=2 using the same short-rate tree, then report its yield to maturity and explain the no-arbitrage pricing principle behind your approach.

Pricing & MonetizationTechnical Trade-offsData Modeling
Author's notes

Pricing the bond was fine.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, clarify the short-rate tree and any assumptions (e.g., risk-neutral probabilities, recovery). Then compute the bond price by backward induction using the tree, and solve for the yield to maturity that equates the price to the present value of cash flows. Finally, explain that the price is arbitrage-free because it equals the expected discounted payoff under the risk-neutral measure.

Pro tip: Emphasize that the yield to maturity is a single discount rate that equates the bond's price to its cash flows, but it is not the same as the short-rate tree; it's a derived summary measure. Also, mention that if the bond is callable or has credit risk, the approach would need adjustment.

1. Clarify the short-rate tree and assumptions

State the given short-rate tree (e.g., annual rates) and confirm assumptions: risk-neutral probabilities, no arbitrage, and that coupons are paid at t=1 and t=2. If not provided, ask for the tree or assume a simple binomial tree.

2. Price the bond via backward induction

At t=2, the bond pays 105 (coupon + face). Discount back to t=1 using the short rate at each node, add the coupon of 5, then discount back to t=0 using the initial short rate. Average across branches using risk-neutral probabilities.

3. Compute yield to maturity (YTM)

Solve for y such that the bond price equals 5/(1+y) + 105/(1+y)^2. Use a numerical method (e.g., Newton-Raphson) if necessary, and report y as an annualized rate.

4. Explain no-arbitrage pricing principle

Articulate that the bond price is the expected present value of its cash flows under the risk-neutral measure, ensuring no arbitrage opportunities. Mention that any deviation would allow riskless profits via replication or hedging.

Key Points to Mention

  • Risk-neutral valuation: expected discounted payoffs under risk-neutral probabilities
  • Backward induction on the short-rate tree
  • Yield to maturity as a single discount rate solving PV of cash flows = price
  • No-arbitrage condition: price must equal cost of replicating portfolio
  • Assumptions: no credit risk, no embedded options, flat recovery
  • Potential need for numerical methods to solve for YTM

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.