The Black-Scholes Equation: What It Does, What It Assumes, and Where It Breaks

Black-Scholes is the most famous equation in finance and one of the most widely misunderstood. It is usually described as a formula that tells you what an option is worth, which makes it sound like a forecasting machine. It is closer to the opposite. The model’s central move is to remove forecasting from the problem entirely.

That is the idea worth carrying away, and this guide builds up to it: what the model actually claims, why the stock’s expected return vanishes from the answer, what implied volatility really means, which assumptions break and how much that costs, and what any of it changes for someone managing their own portfolio. There is a lab at the end where you can price contracts, solve for implied volatility, and watch the model bend.

Start with one coin flip

Skip the calculus for a moment. A stock trades at $100. In one period it will be worth either $120 or $90, and nothing else. A call struck at $100 is therefore worth $20 or $0. Ignore interest for simplicity.

Now build a portfolio that mimics the option: buy $\Delta$ shares and borrow $B$ dollars. For it to match the option in both outcomes:

Subtracting gives $30\Delta = 20$, so $\Delta = 2/3$ and $B = 60$. Buying two-thirds of a share for $66.67 and borrowing $60 costs $6.67, and that portfolio pays exactly what the option pays in both states. So the call must cost $6.67. Any other price lets someone buy the cheap one, sell the dear one, and take the difference with no risk.

Notice what never entered that calculation: the probability of the stock rising. Two investors who disagree completely about the odds still agree on the option’s value, because both can build the same replicating portfolio at the same cost. Black-Scholes is this argument, run continuously.

From hedging to the equation

The continuous version assumes the stock follows geometric Brownian motion:

Hold one option and short $\partial V/\partial S$ shares against it. The random terms cancel instantly, so over the next moment the portfolio is riskless and must therefore earn the risk-free rate. Writing that condition out gives the Black-Scholes equation, here with a continuous dividend yield $q$:

Look at what is missing. The drift $\mu$, the stock’s expected return, does not appear. It is genuinely strange the first time you see it: two people with wildly different views on a stock must still agree on what its options are worth, so long as they agree on volatility. The 1997 Nobel citation named the achievement precisely, awarding the prize “for a new method to determine the value of derivatives.” The method, rather than the formula, was the contribution.¹ Merton extended and generalized the argument in the same year.²

The equation applies to any claim satisfying those market conditions. What makes it a call rather than a put is the payoff you impose at expiry.

The formula, and a worked example

Solving that equation for a European call and put gives:

Take a $100 stock, a $100 strike, one year, 20% volatility, a 5% risk-free rate, and no dividend. Then $d_1 = 0.35$ and $d_2 = 0.15$, giving a call worth $10.45 and a put worth $5.57. Those two prices are locked together by put-call parity:

Here $10.45 minus $5.57 is $4.88, and $100 minus $95.12 is also $4.88. That relationship follows from arbitrage alone and holds with no model at all, which is why it survives everything the rest of this guide throws at Black-Scholes. It also explains why a covered call and a cash-secured put on the same strike are the same economic position wearing different names, a point we work through in the cash-secured put guide.

What N(d₂) is, and what it is not

$N(d_2)$ is the probability the call finishes in the money under the risk-neutral measure. That qualifier carries real weight. Risk-neutral pricing works by pretending every asset drifts at the risk-free rate, which is a computational device rather than a claim about the world. Since stocks are generally expected to return more than cash, $N(d_2)$ typically understates the real-world chance a call ends up in the money.

It also says nothing about whether you profit, because profit depends on the premium you paid. An option can finish in the money and still lose you money. Brokerage tools that label this figure "probability of profit" are describing something the number does not measure.

$N(d_1)$ is not a probability at all under that measure. For a call without dividends it equals delta, and $S_0N(d_1)$ is the present value of receiving the stock only in the states where you exercise.

The Greeks, with units attached

The Greeks are partial derivatives: each one answers how the value moves when one input moves and everything else is frozen. FINRA describes them as measures of an option’s theoretical value computed while "holding all else constant."³ That phrase is the whole caveat: in real markets nothing else stays constant, so the Greeks describe the next small step rather than promising anything about a large one.

For the example above, the call carries:

Read those magnitudes against each other and the practical lesson arrives: a one-point move in volatility is worth more than half a dollar move in the stock. Volatility drives this position about as hard as direction does, and you are buying both whether you meant to or not.

You can be right about the stock and still lose

Here is the scenario that explains more retail option losses than any argument about probability. Buy a 30-day call struck at $105 on a $100 stock, with implied volatility at 45% because earnings are coming. At a 4.5% rate the model prices it at $3.27.

Earnings arrive. The stock rises to $102, exactly the direction you predicted. The uncertainty that justified 45% volatility has now resolved, so implied volatility falls to 25%. Five days have passed. The same contract is now worth $1.57.

The stock went up 2% and the option lost 52%. Push the stock all the way to $105, a 5% gain, and the position is still down 11%. To break even after that collapse in volatility, the stock needed to reach roughly $105.68, a 5.7% move. You were right about direction and it did not matter, because you were also implicitly betting that uncertainty would stay high, and it did not.

This is what the Greeks were telling you in advance. Buying a high-volatility option before a scheduled event means paying for volatility that is about to disappear. The event being obvious is exactly why it was already in the price.

Implied volatility runs the formula backwards

In practice nobody knows future volatility, but everybody can see the premium. So traders invert the problem: given the market price, what volatility would the model need to produce it? That number is implied volatility, and it has no closed-form solution, so it has to be searched for numerically.

Implied volatility is a restatement of the price in different units. It is informative, and the research supports that. Christensen and Prabhala found implied volatility outperformed past realized volatility as a forecast of future volatility in S&P 100 options,⁴ and Jiang and Tian found a model-free version "subsumes all information contained in the Black-Scholes implied volatility and past realized volatility" for S&P 500 options.⁵

Being informative is different from being a forecast. Implied volatility also contains what people will pay to transfer risk, which is why it typically sits above the volatility that subsequently shows up.

The tell that the model is incomplete is that implied volatility is not one number. Options on the same stock, expiring on the same day, trade at different implied volatilities depending on strike. If the model were literally true this could not happen. Rubinstein documented these strike and maturity biases and built implied binomial trees to recover the non-lognormal distributions the market was actually pricing.⁶ For equity indexes, downside puts persistently cost more volatility than equivalent upside calls, which is the market saying crashes are more likely and more painful than a lognormal bell curve allows.

That skew has a birthday. Rubinstein tracked how far implied volatilities spread across strikes on S&P 500 options and found the gap across a roughly nine-percent band around the money widened from about 1.5 percentage points in 1986 to about 6.5 percentage points by 1992, as low-strike options grew persistently more expensive than high-strike ones. He attributed the change to October 1987 and named the resulting pattern “crash-o-phobia.” Before the crash the model’s flat-volatility assumption was roughly true in the data. Afterwards it never was again, because the market had learned something the model does not contain.

Every assumption, and whether it survives

A model being unrealistic is not the same as it being useless. The questions worth asking are which assumptions matter for the contract in front of you, how large the resulting error is, and whether you are using the model as a rough benchmark or as a hedging instruction.

Where it breaks, and what the breaks taught us

Volatility moves. Heston made variance itself a mean-reverting random process correlated with returns, which reproduces both the smile and the tendency for volatility to spike as prices fall.⁷ Bakshi, Cao and Chen tested the alternatives head to head on S&P 500 options and concluded that “incorporating stochastic volatility and jumps is important for pricing and internal consistency. But for hedging, modeling stochastic volatility alone yields the best performance.” The best model, in other words, depends on the job.⁸

Prices jump. Merton extended pricing to processes with discontinuities and named the casualty exactly: “The critical assumption required for such a strategy to be feasible, is that the underlying stock return dynamics can be described by a stochastic process with a continuous sample path.” If the price can gap, the hedge cannot follow it, and the risk cannot be fully removed.⁹ Jump risk is unhedgeable in principle, not merely inconvenient.

Hedging costs money. Leland showed that transaction costs break the continuous-rebalancing argument outright, since infinitely frequent trading would cost infinitely much.¹⁰ Real replication is approximate and the approximation has a price.

American options exercise early. The formula assumes you wait until expiry. Cox, Ross and Rubinstein’s binomial tree handles early exercise by checking, at every node, whether exercising beats holding, and it converges to Black-Scholes as the steps shrink.¹¹ For path-dependent American problems, Longstaff and Schwartz use least-squares regression inside a simulation to estimate the value of continuing.¹²

Why "implied volatility exceeds realized" is not free money

Options do tend to be priced above the volatility that later shows up, and this gets misread as a standing invitation to sell them. The academic name for the gap is the variance risk premium, and the evidence is real: Coval and Shumway found that zero-beta at-the-money straddles lost roughly three percent per week,¹³ and Bakshi and Kapadia found delta-hedged option positions "underperform zero" on average, with the shortfall larger when volatility is high.¹⁴ Carr and Wu built the method for measuring the premium directly across indexes and individual stocks.¹⁵

Insurance premiums also exceed average claims. That is what compensates the insurer for absorbing losses that arrive all at once, at the worst possible moment. Selling volatility earns a premium for the same reason and carries the same shape of risk: many small gains punctuated by rare large losses, which arrive precisely when everything else you own is falling. A high win rate is a feature of the payoff, not evidence of an edge.

What this means if you manage your own money

The most common way a DIY investor misuses this model is to put historical volatility into it, get a value above the market premium, and conclude the option is cheap. The market price reflects forward-looking volatility, the skew, jump risk, dividends, borrow costs, liquidity, and a risk premium. A gap between your estimate and the quote is nearly always a difference of assumptions rather than a mispricing you can harvest.

The evidence on individual investors trading options is not encouraging. Bauer, Cosemans and Eichholtz, studying a large Dutch brokerage, found most investors "incur substantial losses on their option investments, which are much larger than the losses from equity trading," attributing it to poor market timing and high trading costs, with gambling and entertainment ranking above hedging as stated motives.¹⁶ We cover the broader retail record and the payoff-versus-expected-value confusion in the options risk and reward guide.

So the honest recommendation is narrow. Learn this model to understand what you are being sold, not to price contracts better than the firms quoting them. That understanding pays off in specific ways: recognizing that a covered-call fund is selling your upside, that a high distribution rate is not a return, that a protective put is insurance with a premium you can estimate, and that a cheap-looking weekly option is cheap because it will probably expire worthless. Those judgments improve decisions you are already making. Beating a market maker on price is not a realistic goal, and the model itself explains why: their volatility surface is calibrated to live order flow, and yours is a guess.

The lab

Price a contract, solve for the volatility hiding inside a quote, and watch the assumptions bend. The heatmap is the one to sit with: it shows the region where the stock rises and the option still loses.

There is a standalone version of this lab if you want to bookmark it on its own.

Frequently asked questions

Does Black-Scholes still work, given everything it gets wrong?

It works as a benchmark and a common language rather than as a truth machine. Traders quote each other in implied volatility precisely because the formula gives everyone a shared way to translate prices across strikes, expiries, and underlyings. Its errors are systematic enough to be informative, which is how the smile became evidence about crash risk rather than merely a flaw.

Why does the stock’s expected return disappear?

Because the option can be rebuilt from the stock and cash. If you can replicate the payoff, the option must cost what the replication costs, and that cost does not depend on where you think the stock is going. Two investors with opposite views on the company still agree on the option price if they agree on volatility. This is also why volatility, rather than direction, is the input worth arguing about.

Can I use it to find mispriced options?

Realistically, no. Feeding in your own volatility estimate and comparing to the market tells you how your assumption differs from the market’s, which is useful self-knowledge and not an arbitrage. To profit you would need to be right by more than the bid-ask spread, fees, hedging slippage, and model error combined, against counterparties whose entire business is this calculation.

Is implied volatility a prediction?

Partly. It carries genuine information about future volatility, and it also carries the price of risk transfer, so it usually sits above what subsequently materializes. Treat it as the market’s price for uncertainty rather than its forecast of uncertainty.

Key takeaways

• The model prices an option by replicating it from stock and cash, which is why the stock’s expected return does not appear in the answer.
• N(d₂) is the risk-neutral chance of finishing in the money. It is not the real-world chance, and it is not the chance you profit.
• A one-point move in implied volatility was worth more than half a dollar move in the stock in our example: volatility is a primary driver, not a detail.
• A 30-day call bought at 45% volatility lost 52% of its value when the stock rose 2% and volatility fell to 25%. Being right on direction is not sufficient.
• The volatility smile is the market pricing fat tails and crash risk that the original model assumes away.
• Options being priced above subsequently realized volatility is compensation for bearing tail risk, in the same way insurance premiums exceed average claims.
• Learn the model to understand what you are being sold. Use it to price better than market makers and you are competing on their terms.

Related guides

• A 1,000% Upside and a 100% Downside Does Not Make It a Good Bet on payoff versus expected value, and what retail option traders actually earn.
• Covered Calls Are Not Free Income on what selling the right tail costs, in the language of this model.
• Cash-Secured Puts the position put-call parity shows is the covered call in disguise.
• Position Sizing because knowing what a contract is worth settles nothing about how much of it to own.
• The Merton Share the same author’s framework for sizing risk itself.

Sources

1. Black, F., & Scholes, M. (1973). The Pricing of Options and Corporate Liabilities. Journal of Political Economy 81(3), 637–654.
2. Merton, R. C. (1973). Theory of Rational Option Pricing. The Bell Journal of Economics and Management Science 4(1), 141–183.
3. FINRA. Options, investor education. Describes the Greeks as theoretical values computed holding all else constant.
4. Christensen, B. J., & Prabhala, N. R. (1998). The relation between implied and realized volatility. Journal of Financial Economics 50(2), 125–150. S&P 100 index options.
5. Jiang, G. J., & Tian, Y. S. (2005). The Model-Free Implied Volatility and Its Information Content. The Review of Financial Studies 18(4), 1305–1342.
6. Rubinstein, M. (1994). Implied Binomial Trees. The Journal of Finance 49(3), 771–818.
7. Heston, S. L. (1993). A Closed-Form Solution for Options with Stochastic Volatility with Applications to Bond and Currency Options. The Review of Financial Studies 6(2), 327–343.
8. Bakshi, G., Cao, C., & Chen, Z. (1997). Empirical Performance of Alternative Option Pricing Models. The Journal of Finance 52(5), 2003–2049.
9. Merton, R. C. (1976). Option pricing when underlying stock returns are discontinuous. Journal of Financial Economics 3(1–2), 125–144.
10. Leland, H. E. (1985). Option Pricing and Replication with Transactions Costs. The Journal of Finance 40(5), 1283–1301.
11. Cox, J. C., Ross, S. A., & Rubinstein, M. (1979). Option pricing: A simplified approach. Journal of Financial Economics 7(3), 229–263.
12. Longstaff, F. A., & Schwartz, E. S. (2001). Valuing American Options by Simulation: A Simple Least-Squares Approach. The Review of Financial Studies 14(1), 113–147.
13. Coval, J. D., & Shumway, T. (2001). Expected Option Returns. The Journal of Finance 56(3), 983–1009. The three-percent-per-week figure applies to zero-beta straddles.
14. Bakshi, G., & Kapadia, N. (2003). Delta-Hedged Gains and the Negative Market Volatility Risk Premium. The Review of Financial Studies 16(2), 527–566.
15. Carr, P., & Wu, L. (2009). Variance Risk Premiums. The Review of Financial Studies 22(3), 1311–1341.
16. Bauer, R., Cosemans, M., & Eichholtz, P. (2009). Option trading and individual investor performance. Journal of Banking & Finance 33(4), 731–746.
17. The Options Clearing Corporation. Characteristics and Risks of Standardized Options. FINRA Rule 2360(b)(11)(A)(i) requires delivery at or before options account approval.

Editor’s note

Educational content, not investment advice. Option values here come from the Black-Scholes-Merton model with European exercise, constant volatility, a continuous dividend yield, and no transaction costs, so they approximate rather than describe American equity options. Probabilities derived from the model are risk-neutral and represent pricing conventions rather than forecasts. The worked example and the volatility-crush scenario were computed with the same pricing code used in the lab and cross-checked against an independent implementation. Citations verified against publisher metadata as of July 2026.