The Binomial Options Pricing Model

The binomial model builds a tree of possible future stock prices. It prices options by working backwards through that tree from expiration to today. It's slower than Black-Scholes, more flexible, and the correct model when early exercise actually matters.

Where It Came From

Black-Scholes was published in 1973. It was elegant, fast, and immediately useful — but it had one significant limitation: it couldn't handle options that could be exercised before expiration.

In 1979, John Cox, Stephen Ross, and Mark Rubinstein published a paper introducing the binomial model — sometimes called CRR after their initials. Their goal wasn't to replace Black-Scholes. It was to build a model that could:

1. Price American-style options correctly
2. Be understood intuitively without advanced calculus
3. Converge to the Black-Scholes price as the number of time steps increases

They succeeded on all three counts. The binomial model is often taught before Black-Scholes in derivatives courses precisely because the intuition is clearer — no stochastic calculus required.

A detail worth knowing: Mark Rubinstein later spent years developing the implied binomial tree, a more sophisticated version that could capture the volatility smile that plain Black-Scholes ignores. Cox went on to contribute foundational work in interest rate modeling. Ross is best known for developing Arbitrage Pricing Theory, a major framework in asset pricing separate from options entirely.

How the Tree Works

At each time step, the model assumes the stock can do one of two things: move up by a factor u, or move down by a factor d. The probability of moving up is p, and the probability of moving down is 1 − p.

These three values — u, d, and p — are derived from the stock's volatility and the length of each time step. They're not forecasts. They're parameters chosen so that the tree's statistical properties match the volatility input.

Step 1 — Build the stock price tree. Starting from today's stock price, calculate every possible future price at each node. With 3 time steps, there are 8 possible paths and 4 possible ending prices. With 500 steps, there are far more — but a computer handles that instantly.

Step 2 — Calculate option values at expiration. At the final nodes (expiration), the option value is simply its intrinsic value: max(0, stock price − strike) for a call, max(0, strike − stock price) for a put.

Step 3 — Work backwards. At each node before expiration, calculate the option's value as the discounted expected value of the two nodes it can lead to. But also check: is it worth more to exercise right now than to hold? If yes, the option value at that node equals its intrinsic value — early exercise is optimal.

This backwards check at every node is what allows the binomial model to correctly price American options. Black-Scholes skips this check entirely.

Accuracy vs Speed

The binomial model's accuracy depends on the number of steps. With 5 steps, prices are rough. With 500 steps, the prices converge very close to the exact theoretical value.

The tradeoff is computation time. For a single option, 500-step binomial is fast — milliseconds on modern hardware. But OpCalc computes P/L across 40+ strikes, 300 price scenarios, and 12 time series simultaneously. Running a full binomial tree for each calculation point would be significantly slower than Black-Scholes, which is a single formula evaluation.

For the trades most retail options traders make — short-dated, near-the-money options on non-dividend stocks — the binomial model and Black-Scholes produce prices that differ by less than one cent. The computational cost isn't justified by the accuracy gain.

When Binomial Is Meaningfully Better

American puts deep in the money near expiration. If you're holding a put that's significantly in the money with just days left, the interest you'd earn on the strike price received from early exercise may exceed the remaining time value. The binomial model correctly identifies this and prices the put at intrinsic value. Black-Scholes will slightly overprice it.

Call options on stocks paying discrete dividends. If a stock pays a quarterly dividend and the ex-dividend date falls within your option's life, early exercise of a deep in-the-money call to capture the dividend can be rational. The binomial model handles this correctly when dividend dates are built into the tree. Black-Scholes requires a separate adjustment.

LEAPS and long-dated options. Over a one-to-three year horizon, the early exercise premium accumulates and the dividend effect compounds. Binomial pricing is more appropriate, and the performance cost is justified because you're computing for a single position, not a live calculator updating in real time.

The Convergence Property

As the number of binomial steps approaches infinity, the binomial price converges to the Black-Scholes price — for European options. This isn't a coincidence. Cox, Ross, and Rubinstein designed it this way. The two models are built on the same underlying assumptions about stock price dynamics; they just arrive at the answer by different paths.

For American options, the binomial price converges to the correct American option price, which is always at least as high as the Black-Scholes European price. The difference is the early exercise premium.

This convergence property is why "Black-Scholes or binomial?" is often the wrong question. They're the same model at the limit, with different handling of early exercise. The right question is: does early exercise matter for the specific option you're pricing?

Where Each Model Belongs

OpCalc uses Black-Scholes. See why we made that choice and where the limitations apply.