Comprehensive Guide to the Binomial Option Pricing Model

Search Fintrev

Search our guides, reviews and glossary.

The binomial option pricing model values an option by mapping how the underlying asset’s price could evolve over discrete time steps. At each step the price can do only one of two things: rise by a set factor or fall by a set factor. The model builds this branching “tree” of prices out to expiration, calculates the option’s payoff at every final node, then works backward to today. At each step back it discounts the expected value at the risk-free rate, using probabilities chosen so that no arbitrage is possible. The result is the option’s theoretical fair value. Along the way the model also gives the hedge ratio (delta) at every node and, for American-style options, the points where early exercise beats holding.

The approach was published in 1979 by John Cox, Stephen Ross and Mark Rubinstein (“Option Pricing: A Simplified Approach”). In the same year Richard Rendleman and Brit Bartter published related work (“Two-State Option Pricing”). The version with up and down factors tied to volatility is still widely called the CRR binomial tree after its authors.

What Is the Binomial Model in Finance?

In finance, a binomial model is any framework in which a quantity moves to one of two possible values per period. In option pricing, that quantity is the price of the underlying asset: a stock, an index, a currency pair or a futures contract.

The model has a few building blocks:

• Node: a point in the tree representing a possible price at a specific time.
• Step (Δt): the time between nodes. If an option has three months to expiry and the tree has three steps, each step is one month.
• Up factor (u) and down factor (d): the multipliers applied to the price at each step. If u = 1.1, an up move takes $100 to $110.
• Risk-free rate (r): used to discount future values and to set the risk-neutral probabilities.
• Payoff: the option’s value at expiration, such as max(S - K, 0) for a call with strike K.

The name comes from the binomial distribution. After n steps, the final price depends only on how many of those steps were up moves, and the number of paths that lead to each final node follows binomial coefficients. A three-step tree has four terminal prices, reached by one, three, three and one paths respectively.

How a Binomial Tree Works

Building the Price Tree

Start with today’s price S. After one step, the price is either S × u or S × d. After two steps, it is S × u², S × u × d, or S × d². Because an up move followed by a down move ends at the same price as a down move followed by an up move, the branches merge. This is called a recombining tree, and it is what keeps the model manageable. A recombining tree with n steps has n + 1 terminal nodes and (n + 1)(n + 2) / 2 nodes in total. A tree that does not recombine has 2ⁿ paths, which grows out of hand quickly: 20 steps already produce over a million.

Replication and No Arbitrage

The core idea of the binomial model is that an option’s fair price equals the cost of a portfolio that produces exactly the same cash flows. In a one-step tree, there are only two possible outcomes, so you can always solve for a combination of shares and borrowing (or lending) that matches the option’s payoff in both states.

The number of shares in that portfolio is the option’s delta:

Delta = (option value in up state - option value in down state) / (stock price in up state - stock price in down state)

If the replicating portfolio and the option did not cost the same today, a trader could buy the cheaper one, sell the more expensive one, and lock in a riskless profit. Ruling out that arbitrage pins down the option price.

Risk-Neutral Probability

Solving the replication problem algebraically produces a shortcut. The option value equals a probability-weighted average of its two future values, discounted at the risk-free rate:

Option value = e^(-r × Δt) × [p × V_up + (1 - p) × V_down]

where

p = (e^(r × Δt) - d) / (u - d)

Here p is the risk-neutral probability of an up move. It is the probability that makes the underlying’s expected return equal the risk-free rate. It is not an estimate of how likely the price is to rise in the real world. Real-world probabilities, analyst views and expected returns do not appear anywhere in the pricing. Two traders who disagree completely about direction will still agree on the binomial price if they agree on u, d, r and the time step.

For assets that pay a continuous yield, the growth term changes. With a dividend yield q, use e^((r - q) × Δt). For currency options, the foreign interest rate plays the role of the yield. With domestic rate r_d and foreign rate r_f, the risk-neutral probability becomes (e^((r_d - r_f) × Δt) - d) / (u - d).

Choosing the Up and Down Factors

The CRR version links the step sizes to the asset’s volatility σ:

• u = e^(σ × √Δt)
• d = 1 / u

Setting d as the reciprocal of u makes the tree recombine and keeps it centered on today’s price. Other parameterizations exist. The Jarrow-Rudd tree, for instance, sets the risk-neutral probability to one half and adjusts u and d instead. Different choices give slightly different prices for a small number of steps, and they converge as the tree gets finer.

What “e” Means in the Formulas

The letter e (about 2.71828) is the constant used in continuous compounding, so e^(-r × Δt) is the one-step discount factor. Some textbook versions use simple compounding, 1 / (1 + r × Δt), instead; the logic is the same.

Step-by-Step: Pricing an Option With the Binomial Model

1. Gather inputs: current price of the underlying, strike, time to expiration, risk-free rate, volatility, any dividend yield or foreign rate, and whether the option is European or American.
2. Choose the number of steps n and compute Δt = T / n.
3. Compute u, d and p from the formulas above, plus the per-step discount factor e^(-r × Δt).
4. Build the price tree forward from today to expiration.
5. Calculate payoffs at the terminal nodes: max(S - K, 0) for a call, max(K - S, 0) for a put.
6. Work backward one step at a time. At each node, the continuation value is the discounted risk-neutral average of the two nodes that follow it.
7. For American options, compare the continuation value with the immediate exercise value at each node and keep the larger one.
8. Read the value at the first node. That is the option’s model price today.

One-Step Binomial Model Example

This hypothetical example uses round numbers to show the mechanics.

• Stock price today: $100
• In one month, the stock is either $110 (up) or $90 (down), so u = 1.10 and d = 0.90
• Call option: strike $100, expiring in one month
• Risk-free rate: 3% per year, continuously compounded, so Δt = 1/12 ≈ 0.0833

At expiration the call is worth $10 in the up state (110 - 100) and $0 in the down state.

Replication approach. Delta = (10 - 0) / (110 - 90) = 0.5. Build a portfolio that buys half a share and sells one call:

• Value in the up state: $55 - $10 = $45
• Value in the down state: $45 - $0 = $45

The portfolio is worth $45 either way, so it is riskless and must earn the risk-free rate. Its cost today, $50 minus the call price, must equal $45 discounted for one month:

Call price = $50 - $45 × e^(-0.03 × 0.0833) ≈ $50 - $44.89 = $5.11

Risk-neutral approach. p = (e^(0.0025) - 0.90) / (1.10 - 0.90) ≈ 0.5125. Call price = e^(-0.0025) × (0.5125 × $10 + 0.4875 × $0) ≈ $5.11

Both routes give the same answer, as they must. The risk-neutral formula is simply the replication argument rearranged.

Two-Step Binomial Option Pricing Model: European vs. American Put

Extending the same hypothetical inputs to two monthly steps shows what the binomial model adds. Each step uses u = 1.10, d = 0.90, p ≈ 0.5125 and a one-month discount factor of about 0.9975. The strike is $100.

Price tree:

• Today: $100
• After one month: $110 or $90
• After two months: $121, $99 or $81

European call. Terminal payoffs are $21, $0 and $0. Working back, the call is worth about $10.74 at the $110 node and $0 at the $90 node. Its value today is about $5.49.

European put. Terminal payoffs are $0, $1 and $19.

• At the $110 node: 0.9975 × (0.5125 × 0 + 0.4875 × 1) ≈ $0.49
• At the $90 node: 0.9975 × (0.5125 × 1 + 0.4875 × 19) ≈ $9.75
• Today: 0.9975 × (0.5125 × 0.49 + 0.4875 × 9.75) ≈ $4.99

As a check, put-call parity for European options requires C - P = S - K × e^(-r × T). Here that is 5.49 - 4.99 = 0.50, and 100 - 100 × e^(-0.005) ≈ 0.50. The tree is internally consistent.

American put. At the $90 node, holding the put is worth about $9.75, but exercising immediately pays $100 - $90 = $10. The American holder exercises, so that node’s value becomes $10. Working back to today:

0.9975 × (0.5125 × 0.49 + 0.4875 × 10) ≈ $5.11

The American put is worth about $0.12 more than the European put in this example. That difference is the value of the right to exercise early, and the tree shows exactly where that right gets used: at the $90 node after one month. A closed-form European formula cannot show this node-by-node decision, which is the main reason practitioners turn to binomial trees.

Delta at Each Node, and Negative Delta

The tree also yields a delta at every node. At the first node, the European call’s delta is (10.74 - 0) / (110 - 90) ≈ 0.54. The European put’s delta is (0.49 - 9.75) / 20 ≈ -0.46.

A negative delta means the option gains value when the underlying falls. For a put option, that is expected. In replication terms, the put is matched by a short position of about 0.46 shares plus a risk-free deposit. A trader who is long the put and wants to hedge would buy about 0.46 shares per option. Because delta changes from node to node, a binomial hedge is rebalanced at each step, which is the discrete version of delta hedging. The option Greeks explainer covers how delta relates to gamma, theta and vega.

Building a Binomial Options Pricer

The backward-induction algorithm takes only a few lines of code. This Python function prices European or American calls and puts on a CRR tree, with an optional continuous yield q (a dividend yield, or the foreign rate for a currency option):

import math

def binomial_price(S, K, T, r, sigma, n, is_call=True, american=False, q=0.0):
    dt = T / n
    u = math.exp(sigma * math.sqrt(dt))
    d = 1 / u
    disc = math.exp(-r * dt)
    p = (math.exp((r - q) * dt) - d) / (u - d)

# payoffs at expiration; j = number of up moves
    values = []
    for j in range(n + 1):
        ST = S * u**j * d**(n - j)
        values.append(max(ST - K, 0) if is_call else max(K - ST, 0))

# backward induction
    for i in range(n - 1, -1, -1):
        for j in range(i + 1):
            cont = disc * (p * values[j + 1] + (1 - p) * values[j])
            if american:
                Si = S * u**j * d**(i - j)
                exercise = (Si - K) if is_call else (K - Si)
                cont = max(cont, exercise)
            values[j] = cont
    return values[0]

A few practical points apply when you write or use a pricer like this:

• Check that p lies between 0 and 1. If the time step is too coarse relative to the rate and volatility, p can fall outside that range. The tree then implies an arbitrage and its prices are meaningless.
• Use enough steps. Prices oscillate as n changes, often alternating above and below the converged value between odd and even step counts. Averaging two consecutive step counts or using a few hundred steps smooths this out.
• Discrete cash dividends can be modeled by lowering the price at ex-dividend nodes. Subtracting a fixed dollar amount stops the tree from recombining, so implementations often adjust the starting price instead or use a proportional yield.
• Time-varying volatility can be built in, but changing u and d between steps also tends to break recombination and increases the computing load.

How the Binomial Model Is Used

American-style options. Equity options that can be exercised before expiry, especially puts and calls on dividend-paying stocks, are the classic use, because the tree checks the exercise decision at every node.

Hedging and risk management. The deltas from the tree tell a trader how many units of the underlying offset an option position. Repricing the tree under different inputs (higher volatility, a rate change, a price shock) shows how a book of options could respond.

Embedded and corporate options. Convertible bonds, callable bonds, warrants and employee stock options all contain exercise or conversion decisions that fit naturally on a tree.

Real options. In capital budgeting, the same logic applies to decisions such as expanding, delaying or abandoning a project. Each node represents a possible state of the project’s value, and the tree shows where exercising the “option” to act adds value.

Exotic options. With modifications, trees can price some path-dependent products such as barrier options. Products whose payoff depends on the full price path, like Asian options that pay on an average price, need extra state variables at each node or a different technique altogether.

Binomial Model vs. Other Option Pricing Models

Black-Scholes model. The Black-Scholes model gives a closed-form price for European options. It assumes the underlying follows geometric Brownian motion in continuous time, with constant volatility and frictionless markets. The binomial model can be seen as its discrete counterpart: with CRR parameters and a growing number of steps, the binomial price of a European option converges to the Black-Scholes value. The practical difference is early exercise. Black-Scholes has no built-in way to handle it, while the tree does.

Trinomial trees. These allow three moves per step (up, middle, down). They often converge with fewer steps and give more flexibility in fitting the tree to market data.

Monte Carlo simulation. Monte Carlo generates thousands of random price paths and averages the discounted payoffs. It suits path-dependent payoffs such as Asian options and problems with many sources of risk. Early exercise requires extra techniques, such as regression-based methods, because simulation runs forward in time while exercise decisions are found by working backward.

Finite difference methods. These solve the partial differential equation behind option prices on a grid of prices and times. Like trees, they handle American exercise. They also deal well with boundary conditions such as barriers.

What People Confuse the Binomial Model With

• Binomial model vs. binary options. Binary (digital) options pay a fixed amount or nothing. The binomial model is a valuation method that can price many kinds of options, binary payoffs included. The similar names refer to different things.
• Binomial tree vs. decision tree. Corporate decision trees often use subjective probabilities and a risk-adjusted discount rate. A binomial pricing tree uses risk-neutral probabilities and the risk-free rate, so it depends on no-arbitrage logic rather than opinion.
• Model price vs. market price. The tree gives a theoretical value from your inputs. Traders often run the model in reverse, solving for the volatility that makes the model price match the market price (implied volatility).

Limitations of the Binomial Model

Two outcomes per step. A real asset can take any price within a range in any period. A single-step tree with two possible values is a crude picture. Adding steps addresses this, because the distribution of terminal prices becomes richer as n grows, but a coarse tree can miss sharp moves.

Constant volatility in the standard form. CRR trees use one σ for the whole life of the option. Real markets go through volatility spikes, and implied volatility differs across strikes and maturities (the volatility skew). Adapting the tree to match these patterns adds complexity.

Frictionless assumptions. The replication argument ignores transaction costs, bid-ask spreads, taxes and constraints on borrowing or short selling. All of these affect the real cost of hedging and the timing of trades.

Complexity with nonstandard features. Each added feature, such as a path-dependent payoff, a cash dividend or a changing rate, needs extra parameters at each node and may stop the tree from recombining, which raises the computing burden sharply.

The Bottom Line

The binomial model matters most to traders and analysts valuing American options, convertible bonds, employee stock options or real options, where the exercise decision can happen before expiry. Its tree shows where early exercise pays and gives a hedge ratio at every node. Results depend on enough steps and sound inputs, since the model assumes constant volatility and frictionless markets.