Transaction Costs: The Hidden Tax on Every Trade

Every time you buy or sell stock to rebalance your hedge, you pay a little toll. The bid-offer spread. The commission. The market impact. These are transaction costs, and they are the silent killer of options hedging strategies.

Chapter 48 is about what happens when you stop pretending these costs do not exist. The results are surprising: option pricing becomes nonlinear, long and short positions get different values, and portfolio effects matter in ways they never did before.

Why Costs Break Everything

Remember from the last chapter: Black-Scholes requires continuous hedging. That means infinite rebalancing. And with any non-zero cost per trade, infinite rebalancing means infinite total costs. Your entire option premium, and then some, gets eaten alive.

Obviously nobody hedges continuously. But even discrete hedging generates real costs. And those costs depend on who you are. A market maker with tiny spreads pays less than a retail trader with fat commissions. So unlike in the Black-Scholes world, there is no single “correct” option value. The value depends on the investor. This is a completely new concept.

Even more interesting: the same investor will have different values for a long position and a short position in the same option. Transaction costs always drain money from hedgers. Long position? Costs reduce its value below Black-Scholes. Short position? Costs increase what you need to charge above Black-Scholes. The asymmetry is built into the math.

The Leland Model

Hayne Leland started the field of transaction cost modeling in 1985. His approach is elegant and practical.

Assume you rehedge at fixed time intervals. At each rebalance, you pay a cost proportional to the value of stock traded: cost = kappa * |shares traded| * stock price. Here kappa is your personal cost parameter, maybe 0.1% for a market maker, maybe 0.5% for a normal investor.

Leland showed that for a long vanilla call or put, the effect of transaction costs is equivalent to using a modified volatility in the Black-Scholes formula:

sigma_long^2 = sigma^2 * (1 - K)

For a short position:

sigma_short^2 = sigma^2 * (1 + K)

Where K is a non-dimensional parameter:

K = kappa * sqrt(2/pi) / (sigma * sqrt(delta_t))

Long positions use a lower volatility (worth less), short positions use a higher volatility (cost more to maintain). Simple and intuitive.

The parameter K is the key number. It depends on costs (kappa), volatility (sigma), and the hedging interval (delta_t). If K is large, costs dominate and you are rehedging too often. If K is small, costs are negligible and maybe you should hedge more frequently to reduce risk. K gives you a way to compare apples to apples across different stocks and strategies.

The Hoggard-Whalley-Wilmott Model

Leland’s trick only works for single vanilla options because it assumes gamma is always positive. But what about option portfolios where gamma can change sign? A bull spread has positive gamma in some regions and negative gamma in others. A butterfly spread flips gamma sign multiple times.

This is where the Hoggard-Whalley-Wilmott model comes in, and things get really interesting.

The pricing equation becomes:

Black-Scholes terms + cost correction involving |gamma| = 0

Notice the absolute value sign around gamma. That modulus sign makes the equation nonlinear. And nonlinearity changes everything.

In a linear world, the value of a portfolio equals the sum of the values of its parts. Not anymore. Consider this extreme example from the chapter: you hold a long call and a short call with the same strike and expiry. Net position is zero, worth nothing. But if you hedge each leg separately, you pay transaction costs on both sides. The payoffs still cancel at expiry, but you have a negative balance from all the hedging costs.

The smart move is to recognize the cancellation and not hedge at all. Zero cost, zero value. But if you treat the two legs independently, you lose money for no reason. This is why portfolio-level thinking matters with transaction costs.

Economies of Scale and Marginal Costs

This leads to a practical question: what happens when you add a new option to an existing portfolio?

If the new option has gamma of the same sign as your portfolio, adding it makes the cost situation worse. More gamma means more rehedging, means more costs.

But if the new option has gamma of the opposite sign, it partially cancels the portfolio gamma. Less rehedging needed. Lower costs. You might even be willing to pay more than the Black-Scholes value for a contract that reduces your portfolio gamma, because the savings on hedging costs more than compensate.

This is how economies of scale work in options trading. Big portfolios with naturally offsetting positions spend less on hedging per contract than small portfolios. Market makers know this intuitively, but Wilmott gives you the math to quantify it.

Different Cost Structures

The basic model assumes costs are proportional to the value traded (kappa * |v| * S). But real-world costs are more complicated. There might be:

Fixed costs per transaction (k1): you pay something regardless of trade size. This is like a minimum commission.

Volume-proportional costs (k2 * v): cost depends on number of shares, not their value.

Value-proportional costs (k3 * v * S): the Leland model.

The general model handles all three simultaneously. The pricing equation gets more complex but stays in the same framework. The key insight remains: transaction costs add a term to the Black-Scholes equation that depends on gamma, and that term is nonlinear because of the absolute value sign.

Hedging to a Bandwidth

Instead of rehedging at fixed time intervals, another strategy is to rehedge when the position drifts too far from the ideal. You set a tolerance band around the Black-Scholes delta. As long as your actual hedge ratio stays within that band, you do nothing. When it crosses the boundary, you rebalance back to the Black-Scholes delta.

This bandwidth approach was analyzed by Whalley and Wilmott (1993) and Henrotte (1993). The bandwidth is controlled by a parameter H0 that represents the maximum expected risk you are willing to tolerate.

This strategy is intuitive and practical. Why rehedge if the position is barely out of line? The risk from being slightly mishedged is small, and the cost of correcting it might not be worth it.

The Optimal Strategy: Utility-Based Models

All the models above choose a strategy first and then price the option. What if you flip it around and ask: what is the optimal strategy?

Hodges and Neuberger (1989) tackled this using utility theory. The idea: you have a utility function that describes your risk preferences, and you maximize expected utility while accounting for transaction costs.

The result, later simplified by Whalley and Wilmott using asymptotic analysis (because the full model is too complex to compute), is beautiful:

1. There is an optimal bandwidth around the Black-Scholes delta
2. With proportional costs only, the bandwidth scales as the cube root of the cost parameter
3. When the position drifts outside the band, you rebalance to the edge of the band, not to the center
4. With fixed costs, you rebalance to an optimal interior point

The bandwidth formula is:

H = (3/2 * kappa * S * sigma * |gamma| / gamma_risk)^(1/3)

Where gamma_risk is your risk aversion parameter. Higher risk aversion means tighter bands, more hedging, more costs.

Mohamed (1994) tested this strategy with Monte Carlo simulations and found it to be the most successful of all strategies tested. Theory and practice agree here, which is always nice.

Negative Option Prices and Blow-Up

There are some nasty edge cases. With fixed costs, far-out-of-the-money options can have negative values. Think about it: the option is worth almost nothing but you still pay a fixed cost every time you rehedge. The accumulated costs exceed the option’s value.

The practical solution: stop hedging when the option value would go negative. You effectively abandon the position when it is so far out of the money that hedging costs more than the option is worth. This creates a free boundary problem similar to the American option problem.

Even worse, some transaction cost models can “blow up” near expiry near the strike. The gamma becomes infinite at expiration for at-the-money options, and if the cost term involves gamma raised to a power greater than one, the solution can become infinite. The equation becomes ill-posed.

This is not just a mathematical curiosity. It tells you that some hedging strategies simply cannot be sustained near expiration for at-the-money options. You need to adjust your strategy as the option approaches expiry and the strike.

Key Takeaways

Transaction costs fundamentally change the nature of options pricing. Values become investor-dependent. Long and short positions have different values. Portfolio effects matter through the nonlinearity of the pricing equation.

The single most important parameter is K, the ratio of transaction costs to the volatility-adjusted hedging frequency. Keep this number in mind when designing hedging strategies. If K is small, costs are manageable. If K is large, you are hemorrhaging money on hedging.

And perhaps most importantly: reducing portfolio gamma is the key to reducing transaction costs. This is why market makers constantly look for offsetting positions, and why adding a new trade to your book that reduces gamma can be worth more than its Black-Scholes value.

Previous post: Discrete Hedging: What Happens When You Can’t Hedge Continuously

Next post: Volatility Modeling: The Big Picture