FINM6041 · Applied Derivatives
The Binomial Option-Pricing Model
The Binomial Option-Pricing Model (Lecture 5) is the course's first real pricing engine and the intuition-builder for everything after it. The key idea is a delta-hedged riskless portfolio: combine the option with Δ = (f_u − f_d)/(S_u − S_d) shares so the position is riskless, and it must therefore earn the risk-free rate — which lets you back out the option's value with no reference to anyone's expected return. That logic collapses into risk-neutral valuation: price the option as f = e^{−rΔt}[p·f_u + (1 − p)·f_d] using the risk-neutral probability p = (e^{rΔt} − d)/(u − d). You extend it to multi-step trees by building the price tree forward and rolling back node by node, choose u and d from volatility via u = e^{σ√Δt} and d = 1/u, and handle American options by taking max(rollback value, immediate exercise) at every node. In the exam, binomial trees are capped at 4 steps.
What this chapter covers
- 01The delta-hedged riskless portfolio: Δ = (f_u − f_d)/(S_u − S_d)
- 02Risk-neutral valuation: f = e^{−rΔt}[p·f_u + (1 − p)·f_d]
- 03The risk-neutral probability p = (e^{rΔt} − d)/(u − d) — and why it is not a real-world probability
- 04Building the price tree forward, then rolling back node by node
- 05American options: max(rollback value, immediate-exercise payoff) at every node, including the root
- 06Choosing u and d: u = e^{σ√Δt}, d = 1/u when σ is given; use stated up/down moves otherwise
- 07Adjustments for other underlyings: dividend yield p = (e^{(r−q)Δt} − d)/(u − d), futures p = (1 − d)/(u − d)
- 08The 4-step exam cap and convergence to Black-Scholes as steps increase
Two-step American put — checking early exercise at every node
- 2 marksRisk-neutral probability: p = (e^{rΔt} − d)/(u − d) = (e^{0.04} − 0.80)/(1.25 − 0.80) = (1.040811 − 0.80)/0.45 = 0.5351. Build the price tree: after step 1, 56.25 (up) or 36 (down); terminal nodes 70.3125 (uu), 45 (ud) and 28.80 (dd).
- 2 marksTerminal put payoffs, max(X − S_T, 0): f_uu = max(46 − 70.3125, 0) = 0; f_ud = max(46 − 45, 0) = 1; f_dd = max(46 − 28.80, 0) = 17.20.
- 3 marksUp node: rollback = e^{−0.04}[0.5351·0 + 0.4649·1] = 0.4466; early exercise = max(46 − 56.25, 0) = 0. Take the larger → 0.4466.
- 3 marksDown node: rollback = e^{−0.04}[0.5351·1 + 0.4649·17.20] = 8.1963; early exercise = max(46 − 36, 0) = 10. Since 10 > 8.1963, exercise early → value 10.
- 2 marksRoot: e^{−0.04}[0.5351·0.4466 + 0.4649·10] = $4.70; early exercise at the root = max(46 − 45, 0) = 1 < 4.70, so hold. The European put (rollback only) is worth $3.89, so the early-exercise right is worth about $0.81 — nearly all of it created at the deep in-the-money down node.
Key terms
- Risk-neutral probability (p)
- The 'pseudo-probability' p = (e^{rΔt} − d)/(u − d) under which the stock is expected to grow at the risk-free rate. It is a pricing device, not a forecast — the real-world probability of an up-move is irrelevant to the option's value.
- Delta-hedged portfolio
- A holding of Δ = (f_u − f_d)/(S_u − S_d) shares against one short option that has the same value in both future states, hence is riskless and must earn r. This is the no-arbitrage argument the whole model rests on.
- Up and down factors (u, d)
- The multiplicative moves of the underlying over one step. When volatility σ is given, u = e^{σ√Δt} and d = 1/u; when a question states the stock 'moves up/down by x%', use those factors directly.
- Roll-back (backward induction)
- Working from the terminal payoffs back to today, applying f = e^{−rΔt}[p·f_u + (1 − p)·f_d] at each node until you reach the root — the standard way to price any multi-step tree.
- Early-exercise test
- At every node of an American option, compare the rollback (continuation) value with the immediate-exercise payoff and keep the larger. A European option skips this test and uses rollback only.
- CRR parametrisation
- The Cox-Ross-Rubinstein choice u = e^{σ√Δt}, d = 1/u, which links the tree to the stock's volatility and makes the binomial price converge to Black-Scholes as the number of steps grows.
The Binomial Option-Pricing Model FAQ
How many steps can a binomial tree have in the FINM6041 exam?
A maximum of 4 steps — the lecturer states this explicitly in the exam brief, so you will never have to grind out a huge tree by hand. The skill being tested is the method (compute p, build the tree, roll back, check early exercise), not arithmetic stamina.
Is the risk-neutral probability p a real probability?
No. It is a pricing construct — the probability under which the stock earns the risk-free rate — sometimes called a pseudo, risk-adjusted or martingale probability. The actual real-world chance of an up-move does not enter the option price at all, because the option is valued relative to the stock through a riskless hedge.
When do I need to check for early exercise?
For American options, at every non-terminal node — including today's root. Compare the rollback value with the immediate-exercise payoff and keep the larger. For European options you never check; you just roll back. A common trap is a deep in-the-money American put whose early-exercise value beats continuation at a down node.
How do I choose u and d?
If the question gives volatility σ per year, use the Cox-Ross-Rubinstein formulas u = e^{σ√Δt} and d = 1/u. If instead it says the stock goes up or down by a stated percentage, use those factors directly. For a dividend-yield underlying use p = (e^{(r−q)Δt} − d)/(u − d), and for a futures underlying p = (1 − d)/(u − d).
How does the binomial model relate to Black-Scholes?
They are the same idea at different resolutions. As you add steps, the binomial distribution of the terminal price approaches the lognormal, and the binomial price converges to the Black-Scholes-Merton value. Binomial is the tool of choice when no closed form exists — American and path-dependent options.
Studying with AI? Sia — free AI financial modeling tutor works through FINM6041 step by step.
Exam move
Anchor everything to the riskless-hedge story rather than memorising p in isolation: the option plus Δ shares is riskless, so it earns r, and rearranging that gives p = (e^{rΔt} − d)/(u − d). Then drill one clean routine — compute p, build the price tree forward, roll back with f = e^{−rΔt}[p·f_u + (1 − p)·f_d] — until it is automatic on a 2- to 4-step tree. For American options, write 'max(rollback, exercise)' at every node as a physical habit, and always re-check the root. Get the u,d decision right up front (σ given → CRR; percentage moves stated → use them), and know the two variations the exam loves: the dividend-yield and futures versions of p. Keep all decimals through the tree and show each node — a correct price with no working scores zero.