Part XIII · Discipline · Chapter 13A

The retail haircut

Why paper results shrink in a real account

1 · The problem is rarely the edge

The earlier parts of this book were about understanding the mechanism. This part is about the operational discipline that turns that understanding into compound returns. Its framework is built for swing trading: holding positions in SPX, SPY, QQQ and the Mag-7 stocks (Chapter 2D lists them) for 2 to 20 trading days. That horizon is where most of the framework lives. It is also where the dealer mechanics of Parts VI to IX have their sharpest practical use.

The retail trader's problem is usually not a lack of edge, since most have access to genuinely tradeable setups. The problem is the gap between paper performance and live performance, a gap dominated by behavioral and operational factors. A strategy with a paper Sharpe ratio of 0.80 routinely shows 0.20 to 0.40 in a retail account. The gap does not mean that academics are wrong, or that your backtest is. Backtests sometimes are wrong, but assume yours is not. The gap exists because real implementation adds friction at several layers. Skipping that analysis is the single biggest cause of disappointment when traders try to replicate published strategies.

This part closes the gap in eight chapters, beginning with this one, which measures it as the retail haircut: the share of a paper Sharpe ratio lost on the way to your account. Chapter 13B builds the conviction score, $E \times M \times R \times X \times S$, a test of whether a setup deserves capital. Chapter 13C runs every trade through a five-step pipeline, from observation to invalidation, and sets kill rules at three levels. Chapters 13D and 13E reason through ten swing trades and ten portfolio decisions from end to end. Chapter 13F compresses each setup onto a one-page card, and Chapters 13G and 13H manage the whole portfolio and its exposure to volatility.

2 · Two sleeves: mechanical and discretionary

The rules of this part govern one portion of your portfolio, not all of it, so the split comes first. A mechanical sleeve is a rules-only sleeve that executes predefined entries and exits without discretionary override. Its defining property is that no human judgment intervenes between signal and execution. Four examples show the range:

• A systematic program of SPX cash-secured puts, written on a schedule.
• A 200-day moving-average crossover allocation, which holds the index while its price is above its 200-day moving average and holds cash while it is below.
• A Connors RSI(2) mean-reversion bot on QQQ, a program that buys short pullbacks by rule (Chapter 11C defines the RSI).
• A fixed monthly iron condor sale at 30 days to expiry (DTE).

The two traditions of this book disagree about the last example. The Dealer Mechanics Lab teaches the condor (Chapter 12C), while the Edge Book rejects running condors systematically in a retail account (Chapter 10I). Either way, the example shows what "mechanical" means.

The discretionary sleeve is where you choose each trade by judgment inside the framework. It holds the swing trades of Part XI and the whole tactical layer of this book. The split matters because a mechanical sleeve provides baseline market exposure that keeps compounding while discretionary judgment is suspended. Judgment is suspended during the drawdown timeouts of Chapter 13C, personal or health events, regime confusion, or plain loss-of-conviction phases.

If the mechanical sleeve is the portfolio's heartbeat, the discretionary sleeve is its thinking layer. A trader who runs only discretionary positions has no baseline. If discretion fails, as it sometimes does, the whole portfolio sits in cash earning Treasury yields.

A reasonable structure for a retail swing-trading account is 60% to 80% mechanical and 20% to 40% discretionary. Weight the mechanical sleeve toward systematic equity exposure, through factor funds or buy-and-hold, plus modest systematic premium selling. The stopping rules of this part apply to the discretionary sleeve only. They include halting all trading after a 25% drawdown (Chapter 13C) and letting the conviction score govern entries (Chapter 13B). Mechanical sleeves run regardless.

3 · Two routes from paper to account

Every strategy degrades through layers of friction, each one compounding on the one before. Which layers apply depends on where the paper Sharpe ratio came from. The Dealer Mechanics Lab follows a swing strategy that you build and backtest yourself. The Edge Book follows a published academic strategy on its way into an index, then a fund, then your account.

A strategy you backtest yourself

For a strategy of your own, the Dealer Mechanics Lab names four layers:

1. Paper backtest. Baseline performance on historical data, as a backtest engine produces it. Fills are at the mid-quote, the midpoint of the bid and the ask, with no costs beyond a token commission, no taxes and idealized entries.
2. Out-of-sample. The same strategy and parameters on data they have never seen, which is where overfitting shows (Chapter 3A). A paper Sharpe ratio of 1.0 from in-sample optimization typically delivers 0.5 to 0.7 out-of-sample on truly fresh data. The gap is the deflation factor of López de Prado's work on the deflated Sharpe ratio (Chapter 1D).
3. Live execution. Real bid-ask spreads on entry and exit, real slippage whenever your size exceeds the depth at the top of the book, and real timing. You do not always see a setup form in real time as cleanly as it appears in the backtest. Real-world execution typically costs another 20% to 30% of the Sharpe ratio.
4. After tax, after behavior. Tax friction depends on the instrument: gains on SPY and QQQ held a year or less are taxed as ordinary income. SPX, XSP and ES futures (Chapter 2A) get the 60/40 treatment of Section 1256 contracts instead (Chapter 2E). Behavior adds early exits, missed setups, inconsistent position sizes and the urge to revenge-trade after a loss.

Behavioral friction is the largest single source of loss in the chain, and the one most retail traders deny exists.

A strategy you read about

For a published strategy, the Edge Book counts the layers from the academic paper to your account:

An investable index follows published rules that a fund can actually track, with real rebalancing dates, real prices and limited capacity. The PUT index sells SPX puts by rule (Chapter 10C). The BTOP50 is a live composite of large trend-following funds (Chapter 10D), and MSCI Quality is the index that the quality fund QUAL tracks (Chapter 10F). The live layer wraps such strategies in exchange-traded funds: PUTW writes puts, DBMF and KMLM follow trends, and MTUM holds momentum stocks.

Why each layer takes a haircut

Academic to investable index. Academic factor portfolios assume what no real index can keep: daily rebalancing, mid-quote pricing with no bid-ask spread, and no transaction costs, or unrealistically low ones. They also assume no borrow costs for short sales, or implausibly easy borrowing, and infinite capacity with no market impact. Daily rebalancing alone is impractical at scale.

Real-world index construction replaces these idealizations with monthly or quarterly rebalancing, real execution prices, capacity limits, and defined methodologies that may not implement the theoretical signal optimally. The typical haircut is 15% to 30%: larger for strategies that depend on micro-cap exposure, daily rebalancing or dispersed positions, and smaller for liquid index strategies.

One methodology choice can cost more than that: the academic betting against beta (BAB) portfolio has a BAB Sharpe ratio of 0.78 (Frazzini and Pedersen 2014). A value-weighted version, which weights each stock by its market capitalization, earns 0.49 (Novy-Marx and Velikov 2022). That is a haircut of 37% from a single choice.

Investable index to ETF or fund. Wrapping an index in a tradable product adds four frictions:

• An expense ratio, typically 40 to 100 bps a year for factor ETFs and 60 to 100 bps for managed futures funds. The broad factor ETFs of Chapter 10F charge less, 0.15% for QUAL and for MTUM.
• Tracking error from cash drag, the timing of dividend reinvestment, and sampling.
• Manager discretion in active funds, or methodology drift in passive ones.
• Tax inefficiency, which depends on the wrapper's structure.

The typical haircut is 10% to 25%, larger for active funds and complex strategies and smaller for vanilla index-tracking ETFs. The put-writing record shows it: the PUT index's Sharpe ratio of about 0.55 to 0.60 becomes a PUTW Sharpe ratio of about 0.30 to 0.45 over 2016 to 2024. PUTW also changed its strategy slightly in 2023, moving to weekly options. Before then it tracked the PUT index more closely. That wrapper drift is a haircut beyond the expense ratio.

ETF or fund to your account. For an allocation held as ETFs and left alone, this layer is small. For a strategy you run yourself, it is large, and it has five parts:

• Slippage. Your entries and exits never match the index's exactly, so even small slippage compounds over hundreds of trades.
• Taxes. XSP options get Section 1256 treatment, with gains taxed 60% as long-term capital gain (LTCG) and 40% as short-term (STCG). SPY and other ETFs get the standard treatment. In a taxable account an ETF's monthly distributions create ordinary-income tax events, and rebalancing realizes capital gains. For an active strategy in a taxable account, the tax drag can be 100 to 200 bps a year.
• Behavior. Investors trail the funds they own by 100 to 200 bps a year, most of all in volatile, complex, non-intuitive strategies.
• Sizing. Real users under-size on entry, testing a strategy small first, and over-size after wins, wishing they had allocated more.
• Errors. Missed rebalances, trading errors and timing mistakes.

The typical haircut for this layer is 10% to 30%: smaller for ETF-only allocations, larger for self-managed options strategies.

4 · The haircut equation

Each layer multiplies whatever the layer before it left. Write $\text{SR}_{\text{paper}}$ for the Sharpe ratio on paper, and $h_i$ for the share of the Sharpe ratio that survives layer $i$, a number between 0 and 1. The Sharpe ratio you realize in your account is

The product $h_1 h_2 h_3 h_4$, also written $\prod_{i=1}^{4} h_i$, is the share of the paper Sharpe ratio that survives. One minus the product is the total drag, the share that is lost.

Typical values

The Dealer Mechanics Lab gives typical factors for a competent retail swing trader who operates with discipline:

Their typical product is 0.35 to 0.58, while multiplying the ends of the four ranges gives 0.33 at worst and 0.65 at best. So a paper Sharpe ratio of 1.0 typically delivers a user Sharpe ratio of 0.4 to 0.6 in practice. This is the median outcome, not a pessimistic projection. The Dealer Mechanics Lab also states the equation's typical product more broadly, as 0.30 to 0.60, so that 30% to 60% of the paper Sharpe ratio survives. At the start of its chapter it puts the realized range for a paper 1.0 at 0.3 to 0.5. The product of its own factors, 0.35 to 0.58, lies between those two statements.

The execution factor of 0.80 to 0.90 is a cut of 10% to 20%, milder than the 20% to 30% of the layer's own description. The table's values assume a competent, disciplined trader. Until your own record shows that you are one, plan with the larger cut.

The Edge Book splits its last layer in two and gives its own four factors:

For most strategies the product falls between 0.30 and 0.60, although the ends of these ranges multiply to 0.27 and 0.73.

Use the equation explicitly: when you read a paper claiming a Sharpe ratio of 0.8, multiply by 0.40 as your default expectation, which gives 0.32. If the strategy is unusually clean, with low complexity, an ETF-implementable form and good tax efficiency, use a higher multiplier of 0.55 to 0.70. If it is unusually friction-heavy, with active timing, several legs or other complexity, use 0.25 to 0.35. The Dealer Mechanics Lab states the same rule from its side: when a strategy description claims 0.85, your default expectation is 0.30 to 0.40 in your account.

Which layer costs the most

A product is hard to split into parts, but its logarithm is not. Taking natural logarithms of (13A.1) turns the chain into a sum (Chapter 1A):

Since each $\ln h_i$ is negative, its share of the total is that layer's share of the drag, which makes the layers comparable however many there are.

The figure draws the chain as bars, one for the Sharpe ratio left after each layer. Its sliders set the paper Sharpe ratio and the four factors. The starting values are those of the example: a paper 1.0 with factors of 0.80, 0.85, 0.85 and 0.80. Readouts give the user Sharpe ratio, the total drag and the binding layer, the one with the largest drag. At the start, out-of-sample decay and behavior tie at 20%, and the figure names the first. The dashed line is buy-and-hold SPX, which the figure puts at a user Sharpe ratio of roughly 0.40 after retail-typical tax and behavior haircuts. Set the paper Sharpe ratio to 0.85, a typical claim in a strategy description, and watch the last bar fall below the line.

paper Sharpe

1.00

h₁ · out-of-sample

0.80

h₂ · execution

0.85

h₃ · tax

0.85

h₄ · behavior

0.80

user Sharpe = 0.462

total drag = 53.8%

binding = out-of-sample (20%)

Above the buy-and-hold baseline. Edge is plausibly real.

At 0.85 the starting factors leave 0.393, just below buy-and-hold: a strategy that ends there is paying you to underperform the index. Now set the paper Sharpe ratio to 1.10 and the factors to 0.73, 0.81, 0.69 and 0.67, the ones implied by the swing example later in this chapter. The user bar lands near 0.30, with behavior, at a 33% drag, as the binding layer and tax close behind.

5 · Complexity costs more

The haircut varies with the kind of strategy, as the Edge Book's general rule shows:

The haircut grows with complexity, discretion, self-management and active timing. More legs mean more friction, more decision points mean more behavioral risk, self-management adds operational risk and active timing adds missed trades.

The operational implication is to prefer simple implementations, because every additional leg in an options structure must add enough premium or risk-management value to overcome the extra haircut it brings. Often it does not.

6 · Three strategies, from paper to account

A swing setup on SPY

The Dealer Mechanics Lab's example is a trend-pullback setup of the kind Chapter 11H formalizes.

The paper 1.1 became 0.30 to 0.35, keeping 27% to 32% of itself: a haircut of 68% to 73%. This is normal. Plan for it.

The implied factors are $0.8/1.1 = 0.73$, $0.65/0.8 = 0.81$, $0.45/0.65 = 0.69$, and 0.67 to 0.78 for behavior. The tax factor sits far below the typical 0.85 to 0.95 because the trade is in SPY, where every gain meets the short-term rate. Expressed through XSP instead, the same trade would pay the lower 60/40 blend of Chapter 2E. Tax lowers a Sharpe ratio only to the extent that it takes proportionally more from the average return than from the swings around it. A tax that took the same share of every gain and refunded the same share of every loss would scale both alike and leave the ratio unchanged.

The win rate's fall is worth a second look. Had the average win stayed at 1.4 times the average loss, the expected result per trade would have fallen from $0.67 \times 1.4 - 0.33 = 0.61$ average losses to $0.60 \times 1.4 - 0.40 = 0.44$ (Chapter 1C). That fall of 28%, from seven points of win rate, is close to the Sharpe ratio's fall of 27%, from 1.1 to 0.8.

Put-writing on the PUT index

The Edge Book walks the whole haircut for a VRP harvest, XSP cash-secured monthly puts (Chapter 10C).

Plan for a Sharpe ratio of 0.25 to 0.35 if you allocate to this, and treat anything higher as a pleasant surprise. The haircut is real. The Edge Book's short version of the same chain gives the steps as round haircuts: 10% to 15% for the index, a further 25% to 35% for PUTW, and 10% to 25% more for your account.

Trend following

The third example follows a strategy with far less leverage and complexity than put-writing.

You might expect trend, with less leverage and less complexity, to take a smaller haircut than the VRP. In some ways it does: it carries no risk of margin expansion, gets cleaner tax treatment and has lower path dependency. But its behavioral haircut is large because of its long whipsaw periods. Investors who ride trend through droughts of 4 to 7 years before it pays in a 2008 or a 2022 are rare. Most exit during the drought and miss the payoff.

Trend's value, however, is not in its standalone Sharpe ratio but in how it behaves in crashes. Even a trend Sharpe ratio of 0.20 contributes more to a portfolio's Sharpe ratio than its standalone arithmetic suggests, because of its negative crash beta (Chapter 10D).

7 · The largest layer is you

Of the four layers, the behavioral one, $h_4$, is consistently the largest single source of loss. The Edge Book goes further: the behavioral part of the user-account haircut is often larger than all the other parts combined.

What investors actually earn

Morningstar's annual Mind the Gap study has measured the behavioral gap for two decades by comparing two returns on the same funds. A fund's time-weighted return compounds its return period by period, whatever money it held, so it measures what the fund did. Its investors' dollar-weighted return weights each period by the money they actually had in the fund, so it measures what the investors got. Across more than 20 years, the investors have trailed their funds consistently, by 100 to 200 bps a year.

The toy gap is extreme, though its mechanism is the real one: money arrives after gains and leaves after losses. The gap is largest in three kinds of fund:

• Sector funds, where investors chase performance, buying after rallies and selling after drops.
• Complex strategies, which investors do not understand, so they panic at drawdowns.
• High-volatility strategies, which investors enter at peaks and exit at troughs.

By retail standards, premium selling and trend following both qualify as complex and highly volatile. The behavioral gap on them can easily exceed 200 bps a year.

Six ways a swing trader leaks

For swing trading, the behavioral failure modes are six:

1. Position-size inconsistency. Sizing larger after wins and smaller after losses, the reverse of what disciplined trading requires.
2. Stop migration. Moving a stop further away as the price approaches it, "to give the trade more room". This converts small losses into large ones.
3. Premature exits on winners. Closing winners early to "lock in profit" while letting losers run: the classic disposition effect.
4. Revenge trading. Taking marginal setups after a loss to "make it back", with predictable consequences.
5. Setup creep. Lowering the bar for what counts as a valid setup during slow weeks.
6. News chasing. Entering a position on news after the move has already happened, rather than on the geometry of a setup.

Chapter 11N gives remedies for setup creep, stop migration and inconsistent sizing among the failures it catalogs. For revenge trading, Chapter 13F's card adds a day-level defense: after taking a planned loss, stop trading for the rest of the session.

The forced-liquidation cost

Beyond simple panic selling, the behavioral cost includes forced liquidation (Chapter 1E). When portfolio margin requirements expand in stress, undisciplined sizing means you are forced to close positions at the worst time. That is not strictly behavioral, since margin rules are structural. But it is often caused by the behavioral failure to size for stress in advance. A single forced liquidation with the VIX at 30 can wipe out 12 to 18 months of accumulated edge.

The fix is structure

Resolving to behave better than the average investor is not a fix, because most people who make that resolution are wrong. The fix is structural: written rules, entry criteria committed in advance, mechanical position sizing, and kill criteria committed in advance. Discipline exists only when it binds under stress, which it does only if you wrote it down in calm conditions, before the trade.

8 · Plan for the haircut, not against it

The honest framing is uncomfortable. Many sophisticated retail traders spend 100 hours or more implementing a strategy with a paper Sharpe ratio of 0.6, expect to realize 0.5, and actually realize 0.15 after a year. For a swing strategy with a paper Sharpe ratio of 1.0, spending 100 hours on it and realizing 0.35 is the median outcome.

Use the haircut equation to filter strategies before you commit capital. If a strategy's user Sharpe ratio after tax does not beat simply holding SPX, the strategy is not worth its operational complexity. Chapter 1D makes the same point from the other side: the competitor for an active strategy is not zero but the equity risk premium you could hold passively.

The fix is not to find better strategies but to plan for the haircut explicitly and to choose the places where it bites least:

1. Pick strategies whose value does not depend on perfect execution.
2. Pick instruments with favorable tax treatment: XSP, SPX or ES over SPY, for the 60/40 advantage.
3. Pick setups with simple, mechanical entry and exit rules, and use ETF implementations where you can.
4. Commit to position sizing in advance (Chapter 13C).
5. Build behavioral safeguards, the written rules and kill criteria above.
6. Let go of complex strategies that do not survive the haircut.

9 · Where this leads

Chapter 1B gave you the distributions of returns, Chapter 1G the reasons edges exist, and Chapter 1D the reasons the Sharpe ratio misleads for short volatility. This chapter adds realistic retail expectations. Chapter 13B uses all four to formalize the conviction framework, the admission test that any strategy must pass before it gets capital. Without the haircut you would over-allocate to academically attractive strategies that do not survive implementation. The haircut is the gravity that pulls paper Sharpe ratios back to reality.

10 · Check yourself