Part II · Markets · Chapter 2D

What execution costs

Spreads, slippage and market impact

A strategy with a positive paper Sharpe ratio and unsurprising mechanics can still be a net loser in a retail account. The reason is rarely the model but the gap between the price you saw and the price you got. Across the four-layer haircut framework of Chapter 13A, execution friction (layer 3) typically destroys 20% to 30% of the paper Sharpe ratio even before behavior is considered. For multi-leg options strategies in single-name underlyings, the destruction can exceed 50%.

This chapter consolidates the practical rules of execution, whose mathematics is simple. The discipline to follow them is the hard part.

1 · The bid-ask trap, in numbers

On any options chain for a single-name stock, the "last" price you see is irrelevant. It is the price of the last trade, which may have been seconds or hours ago, at whatever spread happened to exist then. The actionable prices are the bid, what you can sell at right now, and the ask, what you can buy at right now. Between them is the spread, and the spread is where edges go to die.

The last row is 10% to 30% of the premium itself. If you buy at the ask and later sell at the bid while the mid stays put, the round trip costs the whole spread $s$. Measured against an option whose mid is $m$, that is a fraction $s/m$ of what the option is worth: $0.50/5 = 10\%$ at the narrow end and $1.50/5 = 30\%$ at the wide end.

A strategy that harvests the variance risk premium (VRP) sells options priced above the volatility that follows (Chapter 10C). It might collect 80 basis points a year of edge over realized variance, which a single 30% round-trip spread erases. In dollars per spread, suppose you harvest $0.30 of edge and give $0.40 to execution across entry and exit. The net is $0.30 - 0.40 = -\$0.10$ a trade: you lose money holding a "winning" strategy.

This is the edge of Chapter 1C with the cost term made concrete. A trade is worth taking only when its expected gain exceeds its round-trip cost:

where $c_{\text{RT}}$ is everything the round trip pays: spread, slippage, market impact and commissions.

This is not hypothetical: it is the single most common reason retail options strategies underperform their backtests.

2 · The five rules of options execution

Rule 1: the mid is the only fair reference price

Compute the mid, $m = (\text{bid} + \text{ask})/2$, at the moment of decision, and compare any limit order you place with it. "Selling at the mid" means an order at the mid. "Selling at $0.05 better than mid" means an order at the mid plus $0.05. The last price is decorative, however prominently the screen shows it. A theoretical price from a vendor's model is informational, not actionable.

The mid also gives slippage a scale: for a fill at price $P$, with $d = +1$ for a buy and $d = -1$ for a sell, the slippage as a share of the spread is

A fill at the mid scores 0%, and a fill at the far side of the spread scores 50%. A score above 50% means you filled worse than the quote.

Rule 2: marketable limits, not market orders

A market order on a thin option chain at 9:32 AM can fill 30% above the prevailing mid in stress. Always use a limit, as Chapter 2C argued for every order of meaningful size. The limit can be aggressive, at the ask for a fast fill or one penny inside it, but it must exist. The only exception is a clear emergency, and a planned exit is never an emergency.

Rule 3: multi-leg orders go in as multi-leg spreads

Many option positions have several legs, such as a vertical spread: an option bought and another sold on the same underlying and expiry at different strikes. When you place a vertical spread as two separate orders, you leg in: you sell one side, then buy the other. You take execution risk on the timing gap between the fills. Even a 5-second gap during a fast tape can cost 5% to 15% of the spread's credit. Use your broker's multi-leg ticket and price the spread as a single net debit or credit. The broker's market makers either fill it or do not, but you take no legging risk.

Rule 4: avoid the open, the close and the lunch hour

Unless trading those windows is part of your strategy, avoid three parts of the day. The first 15 minutes after the open have the widest spreads of the day, as overnight charm and vanna unwinds clear (Chapter 5D). The last 30 minutes carry mechanical rebalancing that creates price pressure unrelated to your thesis (Chapter 2C, on the closing auction). The middle of the lunch hour, roughly 12:00 to 1:00 PM ET, often has thin liquidity. The cleanest execution windows are generally 10:30 to 11:30 AM ET and 1:30 to 2:30 PM ET.

Rule 5: estimate slippage before entering, not after

Before every click, ask what having to exit at the bid, or buy at the ask, does to your expected profit. Exiting at the unfavorable side instead of the mid costs half the spread on each contract. Compare that with the trade's expected profit:

Here $n$ is the number of contracts and $M$ the contract multiplier, 100 for these options. Above 25%, the trade is execution-marginal: either improve the execution, with tighter strikes or more liquid expirations, or skip the trade.

3 · What execution takes from a year

Rule 5 prices one trade, while a strategy repeats its trades and pays its costs every time. The figure below adds up a year of round trips. In its model, every trade is a round trip that pays, on each leg, half the bid-ask spread and a market impact, plus a commission each way. The year's costs are charged against the capital, which the model takes to be one trade's notional $N$.

Write $s$ for the spread and $i$ for the market impact of one leg, both as fractions of the notional. Write $k$ for the commission per leg and $n$ for the number of round trips a year. One round trip then costs $(s + 2i)N + 2k$, and the year's friction drag on the return is

The bracket is the all-in cost of one round trip as a fraction of the notional, and the drag is that cost times the turnover. The net return is the paper return $R$ minus $D$. The figure scales the paper Sharpe ratio by the same proportion, $\text{SR}_{\text{net}} = \text{SR}\,(R - D)/R$, which assumes the costs leave the volatility unchanged.

The figure's controls are these:

A verdict turns yellow when the drag passes 20% of the paper return and red past 50%. Start at the defaults, then halve the number of trades.

trades per year

120

notional per trade

$50k

bid-ask spread

4.0 bps

market impact

2.0 bps

commission per trade

$1.00

paper Sharpe

1.00

paper annual return

20%

RT cost = 8.0 bps

RT cost = $40.00

annual friction = $5,040

all-in RT cost = 8 bps

paper return = 20.0%

friction drag = 10.08%

net return = 9.92%

net Sharpe = 0.50

Red. Friction exceeds 50% of the paper return: the strategy is mostly paying execution costs to the wholesaler. Re-evaluate.

Halving the trades to 60 halves the drag to 5.04%, which is 25% of the paper return, and the verdict turns yellow with a net Sharpe ratio of 0.75. At 40 trades the drag is 3.36%, the verdict green and the net Sharpe ratio 0.83. Nothing about the strategy's signal changed: only how often it paid the toll. The costs that look negligible per trade, 8.4 basis points, become half the return at 120 round trips a year.

4 · Market impact: the square-root law

Beyond the spread and commission that every order pays, a large order also moves the price against itself: that is its market impact. A practical model that captures most of the empirical behavior of trade execution is

where $Q$ is the order's size, average daily volume (ADV) is how much of the security trades on a typical day, and $k$ is a constant for the security and the conditions. Slippage scales as the square root of relative order size. Taking out 1% of ADV produces 10 times the slippage of taking out 0.01% of ADV, not 100 times: that is the importance of the square root. This is the Almgren-Chriss "square-root law" of market impact, validated by extensive empirical work.

The square root has two consequences for how cost grows with size. Per share, a bigger order costs more, but less than in proportion: four times the size doubles the slippage. In total dollars, the cost is the slippage times the size, so it grows as $Q^{3/2}$: four times the size costs eight times as much.

For SPY, daily volume is approximately 50 to 80 million shares. A 1,000-share retail trade is 0.001% to 0.002% of ADV, well below the threshold where impact becomes measurable. For a 1-million-share institutional trade, 1% to 2% of ADV, impact is noticeable, perhaps 5 to 15 bps depending on conditions. For a 100-million-share trade, covering days of execution, the price impact is large, and institutional execution algorithms slice the order across hours or days to minimize the total.

The law describes orders large enough to eat into the book over time. The same effect on a small retail buy is usually negligible. SPY's top of book is many millions of shares deep, so a 100-share retail order has zero market impact. For institutional flow that needs to move millions at a time, slippage and market impact are the dominant execution costs. They are why iceberg orders, which show only a small part of their size at a time, VWAP execution algorithms and dark pools exist. This is also why dealer hedging flow (Part VI onward) uses ES futures rather than spot SPY. The futures book is the most capital-efficient place to absorb large hedging size with minimal market impact.

5 · Instrument selection is an execution decision

SPX, SPY and XSP all reference the same underlying but have different execution characteristics:

• SPX has the tightest bid-ask spreads on liquid strikes, measured against the contract's size, and 1256 tax treatment. It is cash-settled, so there is no assignment risk. Its contract is the largest, $100 per index point, or $500,000 of notional at SPX 5,000. It is the best for size and the worst for small accounts.
• XSP has the identical structure to SPX at one tenth of the contract size, with 1256 tax treatment and cash settlement. Its spreads are sometimes wider in proportion than SPX's. It is the best instrument for most retail swing-options traders.
• SPY has the highest liquidity in raw volume and the tightest absolute spreads in dollars. Its tax treatment is ordinary, labeled "1099" in Chapter 2E's tax figure. It is American-style, so there is early-assignment risk on in-the-money short legs near an ex-dividend date. It is the best for share-level execution and tax-disadvantaged for short-term trading.

Chapter 2A found SPY's options slightly tighter than XSP's, at $0.01 to $0.02 against $0.05 to $0.10. That is the same ranking in dollars: SPY wins on the dollar spread, while SPX's spreads are the tightest in proportion to what a contract controls.

For a retail trader running a Trend Pullback or a Compression Breakout setup (Chapter 11H), XSP is the default choice. Against SPY, its marginal spread cost is paid back many times over in tax efficiency (Chapter 2E). Against SPX, its smaller contract size permits cleaner position sizing.

Single-name options

The Mag-7 names, NVDA, AAPL, GOOGL, MSFT, META, AMZN and TSLA, have institutional-grade options liquidity, with bid-ask spreads of $0.05 to $0.15 on near-dated at-the-money strikes. Below the top 50 names, spreads widen dramatically. A small-cap biotech option might have $1.00 spreads on $3.00 premiums. That is untradeable for any strategy whose edge is measured in implied-volatility mispricing rather than in percentage moves of the underlying.

6 · When legging is acceptable

Despite Rule 3, there are two cases where leg-by-leg execution beats a multi-leg order:

1. When the multi-leg order does not fill: on some thinly traded strikes, market makers do not price the spread as a package but do price each leg individually. After 30 seconds with no fill on a multi-leg order at the mid, you can try legging, but accept that your edge has just degraded.
2. When you are closing a position into rapid movement and want to capture one side quickly. Suppose the short call of your iron condor has moved deep in the money and the put side is worthless. Closing the call side first protects against further loss, and the worthless put side can close whenever. This is "scratching the trade": not optimal execution, but valid risk management.

7 · The execution journal

The mark of execution discipline is the post-trade journal, in which you record six things for every trade:

• the mid at decision time, before clicking;
• the limit price submitted;
• the fill price;
• the spread width at decision time;
• the time of day;
• the volume in the contract that day.

Over 50 trades, this dataset reveals your personal slippage profile. Most retail traders, asked to estimate their average slippage, guess 10% to 20% of the spread. The journaled data routinely shows 40% to 60%. The difference is not skill but discipline, and recognizing the gap is the first step to closing it.

8 · Check yourself