6.3 Pricing Options: Binomial Trees and Black–Scholes
Where does an option's price come from? Not from forecasts of the market, but from what it costs to copy the option with shares and borrowing: one- and many-step binomial trees, the risk-neutral probability, American versus European options, dividends, and the Black–Scholes formula the trees converge to.
📍 The situation
Rohan, a commerce graduate in Ahmedabad who has just finished §6.2, opens the option chain of a Nifty-style index on his broker's app. In our simulated market the index is at 22,000. A call with strike 22,500 expiring in 30 days is quoted at ₹193 (premiums are quoted per unit of the index, so one index point is ₹1 per unit); the 22,000 call costs ₹406 and the 22,000 put ₹301. Rohan is convinced the index will rise — he would put the chance of that at 60% — and asks the obvious questions. Who decided that the 22,500 call is worth ₹193? Would the price be higher if everyone agreed with him that the market is likely to go up? Why does the same call cost more when the market is nervous, and less as expiry approaches?
The surprising answer, and the subject of this section, is that an option's fair price does not depend on anyone's forecast of the direction of the market. It depends on what it costs to copy the option's payoff by holding the right number of shares and borrowing the right amount — a replicating portfolio. We build that idea up in a one-step binomial tree, extend it to many steps (where American options and dividends fit in naturally), and then let the number of steps grow until the tree turns into the famous Black–Scholes formula, built on the geometric Brownian motion of §3.3. Along the way probability enters in an unexpected form — the risk-neutral probability, which is a pricing device, not a forecast.
All index levels, share prices, option premiums, interest rates, dividend yields, volatilities and lot sizes on this page are simulated or illustrative, chosen to make the arithmetic realistic — they are not market data, forecasts or recommendations. The companies are hypothetical. Exchange contract specifications (lot sizes, expiry days, settlement style) and charges change; check the current NSE circulars and contract specifications. Nothing here is investment advice: trading options can lose money quickly.
One Step: Replication and the Risk-Neutral Probability
Start with the simplest possible market. Over one period a share can move to only two prices — up or down. In this toy world an option's payoff is a known amount in each of the two states, and so is the payoff of any combination of shares and cash. Two unknowns (how many shares, how much cash), two states: we can always solve for a combination that pays exactly what the option pays. If two things pay the same in every state, they must cost the same today — otherwise someone could buy the cheap one, sell the dear one and pocket the difference with no risk.
An arbitrage is a trading strategy that costs nothing (or brings in money) today, can never lose money, and has a positive chance of making money. Pricing models assume that markets leave no arbitrage; a consequence is the law of one price: two portfolios with the same payoff in every possible state must have the same price today.
A replicating portfolio for an option is a combination of the underlying asset and risk-free borrowing or lending whose value at expiry equals the option's payoff in every state. By the law of one price, the option is worth exactly what the replicating portfolio costs.
A share costs $S$ today. After a period of length $\Delta t$ (in years) its price is either $Su$ ("up") or $Sd$ ("down"), with $0\lt d\lt u$. Money can be borrowed or lent at the continuously compounded risk-free rate $r$, so ₹1 grows to $e^{r\Delta t}$ in the period. The model has no arbitrage exactly when $$d\lt e^{r\Delta t}\lt u.$$ An option on the share pays $V_u$ in the up state and $V_d$ in the down state (for a call with strike $K$: $V_u=\max(Su-K,0)$, $V_d=\max(Sd-K,0)$).
Why the condition. If $e^{r\Delta t}\ge u$, the share can never beat the bank: sell the share short, deposit the money, and at the end you have $Se^{r\Delta t}\ge Su\ge$ whatever you must pay back — riskless profit in the down state. If $e^{r\Delta t}\le d$, borrow and buy the share instead. Only when the bank rate lies strictly between the two outcomes is the share genuinely risky.
In the one-step binomial model with $d\lt e^{r\Delta t}\lt u$, an option paying $V_u$ or $V_d$ is replicated by holding $$\Delta=\frac{V_u-V_d}{Su-Sd}\ \text{shares}$$ and a risk-free position of $B=e^{-r\Delta t}\,(V_d-\Delta\,Sd)$ (negative $B$ means borrowing). Its no-arbitrage price is $V=\Delta S+B$, which can be rewritten as $$V=e^{-r\Delta t}\big[p^*V_u+(1-p^*)V_d\big],\qquad p^*=\frac{e^{r\Delta t}-d}{u-d}.$$ The no-arbitrage condition is equivalent to $0\lt p^*\lt1$.
Derivation. The portfolio pays $\Delta Su+Be^{r\Delta t}$ or $\Delta Sd+Be^{r\Delta t}$. Setting these equal to $V_u$ and $V_d$ and subtracting gives $\Delta(Su-Sd)=V_u-V_d$; substituting back gives $B$. Then $$V=\Delta S+B=\frac{V_u-V_d}{u-d}+e^{-r\Delta t}\Big(V_d-\frac{(V_u-V_d)d}{u-d}\Big)=e^{-r\Delta t}\Big[\frac{e^{r\Delta t}-d}{u-d}V_u+\frac{u-e^{r\Delta t}}{u-d}V_d\Big],$$ and the two fractions are $p^*$ and $1-p^*$. Finally $p^*\gt0\iff e^{r\Delta t}\gt d$ and $p^*\lt1\iff e^{r\Delta t}\lt u$. ∎
Look at what is missing from the formula: the real-world probability that the share goes up. It never entered the calculation. The replication argument works whatever that probability is, because the replicating portfolio pays off correctly in both states. The number $p^*$ looks like a probability and behaves like one, so it has a name.
In the binomial model the number $p^*=\dfrac{e^{r\Delta t}-d}{u-d}$ is called the risk-neutral probability of the up move (the probability measure it defines is the risk-neutral or pricing measure, and expectations under it are written $E^*$). It is not a forecast: it is the probability that would make every asset earn exactly the risk-free rate on average, which is why prices can be computed as discounted expected payoffs under it.
- Under $p^*$ the share earns the risk-free rate on average: $E^*[S_{\Delta t}]=p^*Su+(1-p^*)Sd=Se^{r\Delta t}$.
- The no-arbitrage price of any payoff is its risk-neutral expected value discounted at the risk-free rate: $V=e^{-r\Delta t}E^*[V_{\Delta t}]$.
- The real-world probability $p$ and the share's expected return $\mu$ do not appear in the price. (They are already reflected in today's share price $S$.)
Why (1). $p^*Su+(1-p^*)Sd$ $=S\big[d+p^*(u-d)\big]$ $=S\big[d+e^{r\Delta t}-d\big]=Se^{r\Delta t}$. So "risk-neutral" means: compute as if investors did not demand any extra return for risk. That is a calculation trick that gives the correct replication price; it is not a claim that investors really are indifferent to risk. ∎
Many Steps: Backward Induction, American Options and Dividends
Two prices in three months is not realistic. But chain many short steps together and the set of possible prices quickly becomes rich: after $n$ steps there are $n+1$ possible prices, and their spread grows like a random walk (§3.3). The pricing idea stays exactly the same — apply the one-step rule at every node, working backwards from expiry.
Divide the time to expiry $T$ into $n$ steps of length $\Delta t=T/n$. At each step the price is multiplied by $u$ or by $d$. Because an up-then-down move ends where a down-then-up move does, the tree recombines: after $i$ steps the possible prices are $Su^jd^{\,i-j}$, $j=0,1,\dots,i$.
The Cox–Ross–Rubinstein (CRR) choice ties the tree to the volatility $\sigma$ of §3.1: $$u=e^{\sigma\sqrt{\Delta t}},\qquad d=\frac1u=e^{-\sigma\sqrt{\Delta t}},\qquad p^*=\frac{e^{r\Delta t}-d}{u-d}.$$ Then each step's log return is $\pm\sigma\sqrt{\Delta t}$, so its variance is $\sigma^2\Delta t$ — exactly the variance of a GBM log return over $\Delta t$.
In an $n$-step tree with $d\lt e^{r\Delta t}\lt u$, a European option with payoff $f(S_T)$ is priced by backward induction: at expiry the node values are $f\big(Su^jd^{\,n-j}\big)$, and at every earlier node $$V=e^{-r\Delta t}\big[p^*V_{\text{up}}+(1-p^*)V_{\text{down}}\big].$$ Equivalently, in one line, $$V_0=e^{-rT}\sum_{j=0}^{n}\binom{n}{j}(p^*)^j(1-p^*)^{n-j}\,f\big(Su^jd^{\,n-j}\big).$$ At each node the replicating portfolio holds $\Delta=\dfrac{V_{\text{up}}-V_{\text{down}}}{S_{\text{up}}-S_{\text{down}}}$ shares, and $\Delta$ changes from node to node: replication is dynamic.
Why. Over the last step each node is a one-step tree, so Theorem 6.11 prices it. Those values are then the "payoffs" of the one-step trees one step earlier, and so on back to today. Each path with $j$ ups and $n-j$ downs has risk-neutral probability $(p^*)^j(1-p^*)^{n-j}$, there are $\binom nj$ such paths, and the discount factors multiply to $e^{-rT}$ — this is the binomial distribution, hence the name. ∎
- Choose $n$ and $\Delta t=T/n$; compute $u$, $d$ (CRR: $e^{\pm\sigma\sqrt{\Delta t}}$, or as given), the growth factor $e^{r\Delta t}$ (use $e^{(r-q)\Delta t}$ for a dividend yield $q$, see below) and $p^*$. Check $0\lt p^*\lt1$.
- Fill in the share prices $Su^jd^{\,i-j}$ at every node.
- At expiry write the payoff: $\max(S-K,0)$ for a call, $\max(K-S,0)$ for a put.
- Step back one column at a time: each node gets $e^{-r\Delta t}[p^*V_{\text{up}}+(1-p^*)V_{\text{down}}]$.
- American option: at each node also compute the value of exercising now, and keep the larger of the two. Mark the nodes where exercise wins.
- Check: European calls and puts on the same tree must satisfy put–call parity $C-P=S-Ke^{-rT}$.
European and American. A European option can be exercised only at expiry; an American option at any time up to expiry. On NSE, index options are European-style and cash-settled; stock options are also European-style, and settled by delivery of the shares (check the exchange's current contract specifications). American options still matter: they are common on other exchanges, early-exercise features appear in many contracts, and the comparison explains why calls and puts behave differently.
Let $r\gt0$.
- An American option is worth at least as much as the European option with the same terms, and at least its exercise value now.
- On a share that pays no dividends before expiry, it is never optimal to exercise an American call early, so the American and European calls have the same price.
- An American put may be worth exercising early when it is deep in the money, so it can be worth strictly more than the European put. (Dividends can make early exercise of a call worthwhile just before the share goes ex-dividend.)
Why (2). A European call is worth at least $S-Ke^{-rT}$: the portfolio "call plus $Ke^{-rT}$ in the bank" is worth $\max(S_T,K)\ge S_T$ at expiry, so it costs at least the share. Since $Ke^{-rT}\lt K$, the call's value exceeds $S-K$, the amount early exercise would give. Selling the call is always better than exercising it, so the right to exercise early is worth nothing. For a put the comparison reverses: exercising a deep in-the-money put gives $K$ now, which can be invested at $r$; waiting can add only a little more, since the share cannot fall below zero. ∎
Dividends (brief). A share price falls by roughly the dividend when it goes ex-dividend, and the option holder does not receive the dividend — so dividends lower call values and raise put values. Two standard adjustments cover most cases. For an index with a continuous dividend yield $q$ (a reasonable model for an index of many companies paying at different times), the share itself returns $r-q$ in price under the pricing measure, so the tree uses $p^*=\dfrac{e^{(r-q)\Delta t}-d}{u-d}$ while discounting is still at $r$. For a share with a known cash dividend $D$ paid at time $t_D\lt T$, the simplest approach is to price as if the share were worth $S-De^{-rt_D}$ today (the "escrowed dividend" method). This is consistent with the cost-of-carry futures price $F=Se^{(r-q)T}$ of §6.1.
From Trees to Black–Scholes
What happens when the steps get shorter and shorter? With the CRR choice, each step's log return is $\pm\sigma\sqrt{\Delta t}$, so the log price after $n$ steps is a sum of $n$ independent small steps — a random walk. By the central limit theorem such a sum is approximately normal, so the share price at expiry becomes lognormal, exactly as in the geometric Brownian motion of §3.3, but with the drift replaced by the risk-free rate. The option price converges as well.
Fix $S,K,T,r,\sigma$ (and $q$). With the CRR tree of $n$ steps, as $n\to\infty$:
- under $p^*$, $\log(S_T/S)$ converges in distribution to $N\big((r-q-\tfrac12\sigma^2)T,\ \sigma^2T\big)$ — the risk-neutral version of GBM, with drift $\mu$ replaced by $r-q$;
- the tree price of a European call or put converges to the Black–Scholes price (Theorem 6.16); the error shrinks roughly like $1/n$ and oscillates as $n$ alternates between odd and even, because the strike sits sometimes on a node and sometimes between nodes;
- the tree price of an American put converges to the American put value, which has no simple closed formula — this is why trees are still used in practice.
Interactive 1 shows the tree and the convergence together. Watch three things: the tree fanning out, the price jumping above and below the Black–Scholes line for small $n$, and — for the American put — the red nodes where exercising early beats holding on.
| $n$ | call | European put | American put |
|---|---|---|---|
| 1 | 13.428 | 5.739 | 5.739 |
| 2 | 11.174 | 3.486 | 4.960 |
| 3 | 12.594 | 4.906 | 5.380 |
| 10 | 11.904 | 4.216 | 5.176 |
| 50 | 12.065 | 4.377 | 5.257 |
| 100 | 12.085 | 4.397 | 5.267 |
| 500 | 12.102 | 4.413 | 5.273 |
| Black–Scholes | 12.106 | 4.417 | — |
Assumptions. The underlying follows geometric Brownian motion with constant volatility $\sigma$ (§3.3); the risk-free rate $r$ and dividend yield $q$ are constant (continuously compounded); trading is continuous and frictionless (no costs or taxes, short selling and borrowing at $r$ allowed); there is no arbitrage; the option is European. Then a call and a put with strike $K$ and time to expiry $T$ (in years) are worth $$C=Se^{-qT}N(d_1)-Ke^{-rT}N(d_2),$$ $$P=Ke^{-rT}N(-d_2)-Se^{-qT}N(-d_1),$$ where $$d_1=\frac{\log(S/K)+\big(r-q+\tfrac12\sigma^2\big)T}{\sigma\sqrt T},\qquad d_2=d_1-\sigma\sqrt T,$$ and $N$ is the standard normal distribution function (written $\Phi$ in §3.3). For a share with no dividends put $q=0$; for a known cash dividend replace $S$ by $S-De^{-rt_D}$ and put $q=0$. The two prices satisfy put–call parity $C-P=Se^{-qT}-Ke^{-rT}$.
Where it comes from (sketch). By Theorem 6.15 the risk-neutral price at expiry is $S_T=Se^{(r-q-\sigma^2/2)T+\sigma\sqrt T\,Z}$ with $Z\sim N(0,1)$, and the price is the discounted risk-neutral expected payoff: $$C=e^{-rT}E^*\big[(S_T-K)^+\big]=e^{-rT}E^*\big[S_T\,\mathbf 1_{\{S_T\gt K\}}\big]-Ke^{-rT}P^*(S_T\gt K).$$ The second probability is the target probability of §3.3 with log drift $r-q-\tfrac12\sigma^2$: $P^*(S_T\gt K)=N(d_2)$. In the first term, writing the expectation as an integral against the normal density and completing the square in the exponent shifts the normal variable by $\sigma\sqrt T$, which turns $d_2$ into $d_2+\sigma\sqrt T=d_1$ and gives $E^*[S_T\mathbf 1_{\{S_T\gt K\}}]=Se^{(r-q)T}N(d_1)$. Multiplying by $e^{-rT}$ gives the formula; the put follows from parity and $1-N(x)=N(-x)$. (Black, Scholes and Merton obtained the same formula from a continuous-time hedging argument — the limit of the replication in our trees.) ∎
Reading the formula. The call is "the share you might receive minus the strike you might pay": $Se^{-qT}N(d_1)$ is the present value of receiving the share when the call finishes in the money, and $Ke^{-rT}N(d_2)$ is the present value of paying the strike, weighted by $N(d_2)=P^*(S_T\gt K)$, the risk-neutral probability of finishing in the money. The number $N(d_1)$ will reappear in §6.4 as the call's delta — the number of shares in the replicating portfolio. Of the six inputs, five can be read off a screen or a contract: $S$, $K$, $T$, $r$ and (approximately) $q$. Only the volatility $\sigma$ is not observable — which is why traders quote options by their implied volatility (§6.4).
| input rises | meaning | call | put |
|---|---|---|---|
| spot $S$ | underlying price | up | down |
| strike $K$ | exercise price | down | up |
| time $T$ | years to expiry (days/365) | up | usually up* |
| volatility $\sigma$ | annual SD of log returns | up | up |
| rate $r$ | continuously compounded risk-free rate | up | down |
| dividend yield $q$ | continuous yield | down | up |
* A deep in-the-money European put can lose value as $T$ grows, because the strike it pays out is received later (Exercise IV.5); for the same reason a deep in-the-money European call on an asset with a large dividend yield can too. The expected return $\mu$ of the underlying is not an input.
- "An option's price is the expected payoff under my probabilities." No — it is the discounted expected payoff under the risk-neutral probability, which is fixed by $u$, $d$ and $r$ (Example 2). Your view of the direction does not change the fair price; it only tells you whether you want to buy.
- "A share with a higher expected return has more valuable calls." The expected return $\mu$ is not an input to Black–Scholes. What matters is volatility, time, rates, dividends and the current price.
- "$N(d_2)$ is the chance my call pays off." It is the risk-neutral probability. The real-world probability differs (it depends on $\mu$), although for short-dated options the two are often close (see the Market case).
- "Black–Scholes predicts option prices." It is a model with one unobservable input. Real markets have fat tails and jumps (§3.1), changing volatility and trading costs; traders use the formula as a translator between prices and implied volatilities (§6.4), not as an oracle.
- Units. $T$ must be in years (30 days is $30/365$, not 30), $\sigma$ and $r$ as decimals per year, and $r$ continuously compounded. A rate quoted with annual compounding $R$ becomes $r=\log(1+R)$.
- "An American option is always worth more." Not for a call on a non-dividend-paying share: early exercise throws away time value, so it equals the European call (Theorem 6.14).
- "A cheap out-of-the-money option is a bargain." It is cheap because, under the pricing measure, it rarely pays. A ₹193 call that needs a 3% rise in 30 days is fairly priced at 14% volatility, not "cheap".
- "The tree's $p^*$ is the market's forecast that the price goes up." It is a device that makes the share earn $r$ on average. It is the same whether the market is optimistic or pessimistic, as long as $u$, $d$ and $r$ are the same.
- Forgetting dividends. A known dividend before expiry lowers calls and raises puts (Example 7); for an index use a dividend yield $q$.
Rohan's 30-day call has strike 22,500; the simulated index is at 22,000; use $r=7\%$, $q=1.2\%$, $\sigma=14\%$ and an illustrative lot of 50 units. He wants to know what he is paying for.
- One-step tree. $u=e^{0.14\sqrt{30/365}}=1.04096$: the index ends at 22,901 or 21,134. $p^*=\dfrac{e^{0.058\times30/365}-1/u}{u-1/u}=0.5495$. Only the up state pays (₹401), so $C=e^{-0.07\times30/365}\times0.5495\times401\approx$ ₹219. A crude answer: one step allows only two outcomes.
- Many steps. 2 steps: ₹223.3; 30 steps (one a day): ₹195.0; 100 steps: ₹192.8.
- Black–Scholes. $d_1=-0.4211$, $d_2=-0.4612$; $C\approx$ ₹192.73, matching the quote. One lot costs about ₹9,636.
- What has to happen. Break-even at expiry is $22{,}500+192.73=22{,}692.73$, a rise of 3.15% in 30 days. The risk-neutral probability of finishing in the money is $N(d_2)=32.2\%$, and of passing break-even 25.0%.
- Rohan's view. Suppose he models the index as GBM with an expected return $\mu=12\%$ a year (an assumption, not a forecast). Then the real-world chance of finishing above 22,500 is 36.0%, and above break-even 28.4% — higher than the risk-neutral values, but nowhere near "60% that it goes up". "The index rises" and "the index rises more than 3.15% within 30 days" are very different events.
- What drives the price. At $\sigma=10\%$ the call would cost ₹104; at 20%, ₹335. With 7 days left (same $\sigma$) it would cost ₹29; with 60 days, ₹370.
What he learns. The premium is the market's price of volatility over time, not of direction. Buying the call is a bet that the index will move up by more than about 3% within the month; the most he can lose is the premium plus charges, and in most scenarios he does lose it. Whether that bet makes sense is his decision — the mathematics only makes clear what the bet is.
Exercises
All prices and rates are illustrative. Rates are continuously compounded per year; give prices to two decimals. Use $N(x)$ from a spreadsheet (NORM.S.DIST(x,TRUE)) or Python (scipy.stats.norm.cdf).
- 1 Check that there is no arbitrage and find the risk-neutral probability $p^*$.
- 2 For a call with strike ₹210, find the replicating share holding $\Delta$, the amount borrowed, and the call's price. Confirm the price with $p^*$.
- 3 Price the put with strike ₹210, describe its replicating portfolio, and check put–call parity.
- 4 The call of part 2 is quoted at ₹20. Describe an arbitrage and the riskless profit today per call.
- 5 Explain why, if the risk-free growth factor $e^{r\Delta t}$ were above $u$, there would be an arbitrage in the share itself.
- 6 Suppose the real-world probability of the up move is 0.8. Does the call price change? Compare the expected 6-month return of the share and of the call under this probability.
- 7 A "digital" contract pays ₹100 if the share goes up and nothing otherwise. Price it.
- 1 For $\sigma=25\%$, $r=7\%$ and monthly steps ($\Delta t=1/12$), find the CRR $u$, $d$ and $p^*$.
- 2 A share is at ₹100; in each of two half-year steps it rises 10% or falls 10%; $r=6\%$. Price a 1-year European call with strike ₹100.
- 3 In the tree of part 2, price the European put with strike ₹100 and check parity.
- 4 In the same tree, price the American put. Where is early exercise optimal?
- 5 A three-step tree has $S=$ ₹100, $u=1.1$, $d=0.9$, $\Delta t=0.25$, $r=8\%$. Use the binomial formula of Theorem 6.13 to price a European call with strike ₹100.
- 6 Price the American call in the tree of part 2 and explain the result.
- 7 An index has $\sigma=15\%$, $r=7\%$ and dividend yield $q=1.5\%$. With monthly CRR steps, find $p^*$ with and without the dividend yield. Which way does the yield push call prices?
- 1 A share is at ₹500 (no dividends). Find $d_1$, $d_2$ and the price of a 6-month call with strike ₹520, with $r=7\%$ and $\sigma=30\%$.
- 2 Price the matching put by the formula and by put–call parity.
- 3 The index is at 22,000 ($r=7\%$, $q=1.2\%$). Price a 15-day put with strike 21,500 at $\sigma=15\%$.
- 4 In part 1, what is the risk-neutral probability that the call finishes in the money?
- 5 Recompute part 1 with $\sigma=60\%$. How much do the call and put change?
- 6 A share at ₹800 will pay a ₹15 dividend in 2 months. Price a 6-month at-the-money call ($K=$ ₹800, $r=7\%$, $\sigma=25\%$) with and without the dividend adjustment.
- 7 A share at ₹1,000 pays no dividend; a 3-month European call with strike ₹900 is quoted at ₹110 ($r=8\%$). Use the bound $C\ge S-Ke^{-rT}$ to show there is an arbitrage, and describe it.
- 1 Explain in three sentences why the expected return of the share does not appear in the Black–Scholes formula.
- 2 A friend says: "I think there is a 60% chance the index rises this month, so the 22,500 call at ₹193 is underpriced." Respond, using the Market case.
- 3 For the one-year option of Example 5 ($S=K=100$, $r=8\%$, $\sigma=20\%$), compute the CRR call prices for $n=1,2,3$ yourself and explain why they oscillate around ₹12.11.
- 4 At the money, an option's value grows roughly like $\sqrt T$. Check this by comparing the 60-day and 15-day at-the-money index calls ($S=K=22{,}000$, $r=7\%$, $q=1.2\%$, $\sigma=14\%$).
- 5 Show that a deep in-the-money European put can be worth less than its exercise value: price a 1-year put with $S=$ ₹50, $K=$ ₹100, $r=8\%$, $\sigma=20\%$. What is the American put worth?
E1. NISM-style Which of the following is not an input to the Black–Scholes price of a European call? (A) volatility of the underlying (B) time to expiry (C) expected return of the underlying (D) risk-free interest rate.
E2. NISM-style A share at ₹100 will be either ₹110 or ₹95 at the end of a short period; ignore interest. The no-arbitrage price of a call with strike ₹100 expiring then is (A) ₹3.33 (B) ₹5.00 (C) ₹6.67 (D) ₹10.00.
E3. NISM-style Other things remaining the same, an increase in the volatility of the underlying will (A) increase call premiums and decrease put premiums (B) increase both call and put premiums (C) decrease both (D) leave at-the-money premiums unchanged.
E4. Case study An index is at 20,000 with $\sigma=18\%$ and $r=7\%$ (ignore dividends). Build a two-step CRR tree with monthly steps. (a) Find $u$, $d$, $p^*$ and the index levels. (b) Price 2-month European and American puts with strike 20,000 and identify any early-exercise node. (c) Price the European call and check parity. (d) The Black–Scholes European put is ₹474.13; why is the tree's value so different, and what would you do about it?
E5. Case study Shares of Tapi Steel Ltd (hypothetical) are at ₹2,400 and will pay a ₹25 dividend in 20 days. Consider 45-day European options with strike ₹2,500, $r=7\%$, $\sigma=32\%$, and an illustrative lot size of 250 shares. (a) Price the call with the escrowed-dividend method. (b) Price the put and check parity. (c) What does one lot of calls cost, and where is break-even at expiry? (d) By how much would ignoring the dividend misprice the call?
E6. Interview A share is at ₹100. Tomorrow it will be ₹120 or ₹80; interest is negligible. What is a call with strike ₹100, expiring tomorrow, worth? Does your answer change if you are told the probability of ₹120 is 90%?
Suggested plan — 2 lectures × 60 min.
- Replication and trees. Open with the 📍 situation and take a class vote: "Should the call cost more if most people expect the market to rise?" Do Example 1 on the board with the two equations, then Theorem 6.11 and the risk-neutral rewriting. Example 2(b)–(c) makes the arbitrage concrete. Move to two steps (Example 3) and the American put (Example 4), using the 🧭 strategy box. Project Interactive 1 with $n=2,3,4$ and the American put to show the red early-exercise nodes.
- Black–Scholes. Run Interactive 1's Play to show convergence and the odd–even zig-zag (Example 5). State Theorem 6.16 with its assumptions; sketch the derivation from §3.3's lognormal target probability. Work Example 6, then project Interactive 2: slide $\sigma$ and the days to expiry, and press Play to watch time value vanish. Finish with the 📈 Market case and Practice corner E2, E6.
Class activity. Pairs build a three-step tree on paper for a call and a put of their choice, check parity, then compare with a classmate's spreadsheet using 100 steps. Discuss why their two-step answers differ from the formula.
Misconceptions to watch for. Using real-world probabilities to price; thinking $p^*$ is a forecast; forgetting to convert days to years; mixing annual and continuous compounding; believing American options are always worth more; treating Black–Scholes as a prediction rather than a model with assumptions.
Suggested homework. I(1–4, 6), II(2–4, 7), III(1, 3, 5, 7), IV(2, 5); Practice corner E4, E5.
Project. Build your own option pricer twice — as a binomial tree and as the Black–Scholes formula — and watch the first converge to the second.
Why it matters. Every option chain, margin calculator and risk report rests on these two calculations. Building them once removes the mystery, and makes it obvious which inputs drive the price.
What to do.
- Inputs in B1:B7: spot (B1), strike (B2), days to expiry (B3), $r$ (B4), $q$ (B5), $\sigma$ (B6), steps $n$ (B7). In B8
=B3/365(years), B9=B8/B7($\Delta t$), B10=EXP(B6*SQRT(B9))($u$), B11=1/B10($d$), B12=(EXP((B4-B5)*B9)-B11)/(B10-B11)($p^*$). - Tree. With $n=10$: in B20:L20 put $j=0,\dots,10$; in B21:L21 the expiry prices
=$B$1*$B$10^B20*$B$11^($B$7-B20); in B22:L22 the payoffs=MAX(B21-$B$2,0). In B23 enter=EXP(-$B$4*$B$9)*($B$12*C22+(1-$B$12)*B22)and copy it across to K23 (one cell fewer) and down, each row one cell shorter; after ten rows, B32 holds the price. - One-line tree. Copy the $j$ values into A40:A50 and the payoffs into B40:B50; then
=EXP(-B4*B8)*SUMPRODUCT(BINOM.DIST(A40:A50,B7,B12,FALSE),B40:B50)gives the same price — this is Theorem 6.13. - Formula.
d1 =(LN(B1/B2)+(B4-B5+B6^2/2)*B8)/(B6*SQRT(B8)),d2 =d1-B6*SQRT(B8), call=B1*EXP(-B5*B8)*NORM.S.DIST(d1,TRUE)-B2*EXP(-B4*B8)*NORM.S.DIST(d2,TRUE), put by parity. - Make a table of tree prices for $n=1,\dots,100$ (Python is easier, below) and chart them against the formula.
Python version: import numpy as np; from scipy.stats import norm; from scipy.special import comb, then j=np.arange(n+1); ST=S*u**j*d**(n-j); C=np.exp(-r*T)*np.sum(comb(n,j)*p**j*(1-p)**(n-j)*np.maximum(ST-K,0)); for an American put use backward induction with V=np.maximum(V, K-S_nodes) at each step.
Hand in: the convergence chart for a call and a put, the number of steps needed to get within ₹0.10 of the formula, and the American–European put difference for $r=4\%$ and $r=10\%$.
Going further. Take a few strikes from the publicly available NSE option chain for the index, use the contract's actual days to expiry, and find by trial and error (or Goal Seek on $\sigma$) the volatility that reproduces each quoted premium — you will have computed implied volatilities, the subject of §6.4.
🧪 Self-check quiz
Q1. In a one-step tree $u=1.1$, $d=0.9$ and $e^{r\Delta t}=1.01$. The risk-neutral probability of the up move is
Q2. In the Black–Scholes formula, $N(d_2)$ is
Q3. For a share that pays no dividends before expiry (and $r\gt0$), an American call is worth
Q4. As the number of steps $n$ in a CRR tree increases, the tree price of a European call
Q5. In a one-step tree, a call option is replicated by