Part XIII · Discipline · Chapter 13F
Setup cards
One-page references for each setup and each pattern to avoid
Chapter 13E reasoned through ten decisions from end to end. This chapter compresses that reasoning into setup cards, one-page references in the form you need at the moment of execution. Each gives a strategy's edge, the regime it needs, its size and the rule that ends it. The cards cover the swing setups of Part XI, the allocatable strategies of Part X, the regime filters and the patterns to avoid.
- The four swing setups (Chapter 11H), the composite signal (Chapter 11G), stage analysis (Chapter 11E), and RSI, IBS and ADX (Chapter 11C).
- The edges and their evidence (Chapters 10B to 10I), with the five categories and the evidence tiers T1 to T3 (Chapter 10A).
- The Dealer Mechanics Lab's evidence tiers A to C (Chapter 11B), and the conviction score (Chapter 13B).
- The regime filters (Chapter 10H).
- Risk against notional, the loss budget and kill rules at three levels (Chapter 13C), and the stress scenarios S1 to S4 (Chapters 1D and 13G).
1 · What a setup card is
A setup card compresses one strategy into the form needed for live execution: its edge source, the regime it requires, its size, the rule that ends it and the ways it commonly fails. The cards are designed to be printed, taped next to your monitor or copied into a trading journal. Their value lies in being binding at the moment you are tempted to override them.
A strategy you understand in a calm evening can still be traded badly at 10 in the morning, when the price is moving and a story is forming in your head. A card moves every decision you can make in advance to the evening. At the screen you check conditions and follow rules, and anything the card does not allow waits until you can think about it calmly.
This chapter merges two sets of cards. The Dealer Mechanics Lab wrote cards for its four swing setups, two risk overlays and seven patterns to avoid. The Edge Book wrote cards for ten allocatable strategies, four regime filters and seven strategies it rejects. Both sets share one principle, discretion at observation, mechanism at execution. You use judgment to decide what you are seeing. Once you decide to act, the card constrains how.
A setup card is a treatment protocol. The protocol states the eligibility criteria, the dose, the schedule, the dose modifications for toxicity and the stopping rules, and it is written before any patient is in the room. At the bedside you judge whether the patient meets the criteria. You do not rewrite the dose because this patient seems different. The card's regime row is the eligibility, its sizing rule the dose, its kill rule the stopping rule, and its "skip when" row the contraindications.
2 · How to read a card
The two sources use different fields, but the fields answer the same questions. The table sets them side by side.
| The question | Swing setup card | Allocation card |
|---|---|---|
| Why does it pay? | Edge source | Edge source and category |
| How good is the evidence? | Evidence tier, A to C | Evidence tier, T1 to T3 |
| When may you trade it? | Regime | Regime required |
| In what? | Vehicle | Instrument |
| When do you get in? | Entry trigger | Entry |
| How big? | Sizing rule | Sizing cap |
| When do you get out? | Profit target and time stop | Exit |
| What ends it? | Invalidation | Kill rule |
| What goes wrong? | Common failure | Hidden exposure and stress loss |
| When do you stand aside? | Skip when | Not to trade when |
| What should it earn, and how sure are you? | Not on the card | Expected return and confidence |
The two evidence scales differ. The Edge Book's tiers, T1 to T3, run from peer-reviewed, replicated, multi-decade evidence to thin or decayed evidence (Chapter 10A). The Lab's tiers, A to C, grade the kind of source (Chapter 11B). Tier A is peer-reviewed work, replicated over decades, or official exchange data. Tier B is vendor statistics or backtests with credible methods but no peer review, and Tier C is practitioner pattern language that you must formalize yourself.
Swing cards state their targets in units of R, the dollar loss you take if the stop is hit. A target of 2R is a gain of twice that, and a trade's R-multiple is its profit or loss divided by R (Chapter 11H). A time stop closes the trade after a set number of trading days, whatever its profit or loss.
3 · Swing setup cards
These four cards come from the Dealer Mechanics Lab and summarize the setups of Chapter 11H. Each is a long trade for a positive regime, held for days, in the discretionary sleeve of the portfolio.
Trend pullback and reclaim
| Field | Rule |
|---|---|
| Edge source | Mean reversion within a Stage 2 trend, with support from dealer hedging at the moving averages |
| Evidence tier | Tier A on the card: Connors' RSI(2), Jegadeesh-Titman and the dealer-mechanics framework. Chapter 11H grades the RSI(2) parameterization Tier B |
| Regime | A composite signal of +30 or more. Stage 2 confirmed, with the 30-bar simple moving average (SMA) rising and ADX between 20 and 30. Positive gamma exposure (GEX) |
| Vehicle | XSP options, preferred for tax, or SPY and QQQ options. Shares for liquid single names |
| Entry trigger | RSI(2) below 10, or IBS below 0.20, on a pullback to the 10-bar or 21-bar exponential moving average (EMA), with the reclaim of the EMA confirmed intraday |
| Sizing rule | Risk 1% to 1.5% of the portfolio per trade, with (Chapter 13C) |
| Profit target | 1.5R to 2R, or a close above the prior 5-day high, whichever comes first |
| Time stop | 5 trading days |
| Invalidation | A close below the entry day's low, or the 30-bar SMA's slope turning negative |
| Common failure | The setup fires with the composite below +20, as the regime weakens, and the price rolls over instead of bouncing |
| Skip when | The VIX is above 25, the term structure is backwardated, ADX is above 30, or the composite is below +20 |
Compression breakout
| Field | Rule |
|---|---|
| Edge source | Range expansion after a contraction in volatility, with a tailwind from dealer hedging in a positive-GEX regime |
| Evidence tier | Tier B to C: practitioner support for the volatility contraction pattern (VCP), from Minervini |
| Regime | A composite of +30 or more. ADX falling toward 15, volume contracting, and positive GEX |
| Vehicle | Shares directly, or a bull call spread for defined risk |
| Entry trigger | The price breaks above the range high, or above a VCP pivot, on volume of at least 130% of its 20-day average |
| Sizing rule | Risk 1% of the portfolio per trade, with the entry slightly above the breakout level |
| Profit target | The range's width added above the breakout level, or 2R, whichever comes first |
| Time stop | 8 trading days |
| Invalidation | A close back inside the prior range within 2 sessions |
| Common failure | Fakeouts on weak volume |
| Skip when | Breakout volume below 100% of the 20-day average, a weakening composite, or a VIX above 22 |
Undercut and reclaim
| Field | Rule |
|---|---|
| Edge source | Stop-runs at well-known technical levels, such as the 200-day moving average (MA) and prior swing lows |
| Evidence tier | Tier B: Minervini, and the literature on short covering |
| Regime | A composite of +10 or more, since the setup is less regime-sensitive than the Trend Pullback |
| Vehicle | Shares directly, or a bull call spread |
| Entry trigger | The intraday low penetrates a major technical level, and the price closes back above it by the end of the day (EOD) |
| Sizing rule | Risk 0.5% to 1% of the portfolio per trade |
| Profit target | The next overhead resistance level, or 1.5R, whichever comes first |
| Time stop | 3 trading days |
| Invalidation | A close below the undercut low, which rejects the reclaim |
| Common failure | Not on the Lab's card. Chapter 11H names repeated undercuts, as the level becomes a magnet for further declines, and skips a second undercut of the same level within 5 sessions |
| Skip when | The VIX is above 22, when stop-runs turn into continuations of a bear flag, or the composite is below 0 |
A bear flag is the downward mirror of the bull flag of Chapter 11D: a sharp fall, then a brief and shallow drift upward, with a break below the flag's low continuing the fall.
IBS and RSI mean reversion
| Field | Rule |
|---|---|
| Edge source | Intraday positioning, measured by IBS, combined with a multi-day oscillator, RSI(2) |
| Evidence tier | Tier A on the card: Connors' documented IBS and RSI strategies, replicated over several decades |
| Regime | Stage 2 confirmed, a composite of +20 or more, and positive GEX above the flip |
| Vehicle | QQQ or SPY shares, for the cleanest execution |
| Entry trigger | IBS below 0.20 and RSI(2) below 15, both at the close |
| Sizing rule | Risk 1% of the portfolio, with the stop at the intraday low. The tight stop permits a larger position |
| Profit target | The first close above the entry day's high, or 1R, whichever comes first |
| Time stop | 2 trading days |
| Invalidation | A close below the intraday low of the entry day |
| Common failure | Not on the Lab's card. Chapter 11H names continuation lower when the regime is genuinely deteriorating, with the composite already weakening |
| Skip when | The composite is below +15, ADX is above 30, or the VIX has risen 5 points or more in the last 3 days |
Where the cards and the chapters disagree
The cards summarize Chapter 11H, but in four places they give different numbers from that chapter or from the kill rules of Chapter 13C:
| Setup | The card | Chapter 11H | Chapter 13C's kill rules |
|---|---|---|---|
| Trend Pullback | Skip above a VIX of 25; evidence Tier A | Requires a VIX below 22 and contango; evidence Tier B | A close below the entry day's low, or 5 days |
| Compression Breakout | Skip above a VIX of 22; time stop 8 days | Requires a VIX below 18; time stop 8 days | A close back inside the range, or 7 days |
| Undercut and Reclaim | Skip above a VIX of 22 | No VIX condition | A close below the undercut low, or 3 days |
| IBS and RSI Mean Reversion | Time stop 2 days | Requires a VIX below 22; time stop 2 days | A reclaim of the 21-EMA, or 3 days |
Where they differ, take the stricter reading. Chapter 11H's regime conditions define the setup, so a trade that fails one of them is not the setup at all, and the card's skip row adds vetoes on top. For the time stop, the shorter one ends a trade that has not paid within its usual horizon.
Step 1: check the card. The composite reads +42, SPX is in Stage 2 with ADX at 24, GEX is positive, the VIX is 16 in contango, and RSI(2) closed at 7 on a pullback to the 21-EMA. Every regime condition holds on the card and in Chapter 11H, and no skip condition applies.
Step 2: set the risk. A $500,000 portfolio risking 1% has a budget of $5,000. XSP is at 600, and the stop below the entry day's low is at 595, $5 away.
Step 3: size the calls. You buy XSP calls with a delta of 0.60, so at the stop each loses about , or $300 a contract with the multiplier of 100. For a move this small, delta is a close guide, and for a long call it errs on the safe side, since gamma slows the call's fall. The rule allows contracts, which carry of delta exposure, 115% of the portfolio.
Step 4: write the exits. R is . A target of 1.5R needs the calls to gain $4.50 each, about points on XSP, so 607.50; 2R needs 610. Close at whichever comes first of the target, a close above the prior 5-day high, a close below the entry day's low, or 5 trading days.
The figure runs five of this chapter's cards against one day's readings, starting from those of the worked Trend Pullback above. Drag the VIX past 22 and all four swing cards close at once. Then push VIX/VIX3M above 1.00, and the Trend Pullback stops along with Card 02, the VRP harvest card below.
Trend Pullback: trade on its trigger. Compression Breakout: skip (fails on ADX). Undercut and Reclaim: trade on its trigger. IBS and RSI: trade on its trigger. Card 02: sell at full size.
4 · Risk overlay cards
Two overlay cards from the Dealer Mechanics Lab change how the swing cards are applied around scheduled events.
| Overlay | Active | Rules |
|---|---|---|
| FOMC week | From 2 trading days before the meeting to 1 day after it (T−2 to T+1) | Reduce all new entries to 50% of their normal size, and widen their stops by 50%. Skip any new entry whose widened stop, 1.5 times the normal distance, would put more than twice the normal portfolio risk on the trade. Look for opportunities in the vanna unwind, 60 to 90 minutes after the announcement |
| Earnings | From 2 trading days before the report to 2 days after it (T−2 to T+2), for any position in the reporting company | Close single-stock positions before the report, or convert them to a defined-risk options structure. Long call spreads and long put spreads are permitted; never hold naked shares or naked options through earnings. Size the position so that the implied move would not produce a loss of more than 2% of the portfolio |
The FOMC rules keep the risk of a new entry below its normal level. Half the size with a stop 1.5 times as far puts of the normal risk on the trade. The third rule vetoes any setup whose natural stop is already so wide that even this overshoots. The vanna unwind is dealers buying back hedges as implied volatility collapses after the announcement (Chapters 5D and 7B). The card looks for it 60 to 90 minutes after the announcement. Chapter 11K opens the window earlier, at 30 to 90 minutes, and enters only once the VIX has fallen more than 2 points within 15 minutes. The card and that chapter also differ before the meeting: the card allows new entries at half size, while the worked FOMC week of Chapter 11K takes none on the eve of the release.
Step 1: the limit. A $300,000 portfolio may lose at most 2% on the report, $6,000.
Step 2: the implied move. The options price a move of 8% for the report, read from the at-the-money straddle (Chapter 11A).
Step 3: the size. A move of 8% costs $6,000 on a position of , so that is the largest position you may hold through the report, 25% of the portfolio. Held as a long call spread, the most you can lose is the spread's cost, so size it so that its cost is no more than $6,000.
5 · Allocatable strategy cards
The Edge Book's ten cards cover the strategies it allows capital: the four foundation premia, four factor tilts and two structural overlays. Every card has the same fields, and a field the Edge Book left blank says so.
Card 01: long equity
| Field | Rule |
|---|---|
| Edge source | Risk premium, Source 1 of Chapter 1G |
| Category | Foundation premium |
| Evidence tier | T1: Mehra and Prescott; Dimson, Marsh and Staunton, across 35 markets |
| Regime required | The default state. Reduce only on extreme valuation combined with weak breadth |
| Instrument | SPY or VOO for the United States, VXUS outside it, VT for the whole world |
| Entry | A lump sum, or dollar-cost averaging into a broad index |
| Exit | No tactical exit; tax-aware rebalancing only |
| Expected return | Real, about 4% to 5% a year over the long run, looking forward; nominal, about 7% to 9% |
| Hidden exposure | A crash beta of 1.0 by definition. A negative real return over several decades is possible, as in Japan from 1989 to 2010 |
| Stress loss | S1: −10% of the allocation. S2: −20%. S3: −35%. S4, a 1987-style gap: −22% overnight |
| Sizing cap | 50% to 65% of the portfolio |
| Kill rule | When 1/CAPE minus the 10-year Treasury yield stays below −1% and 12-month earnings growth is negative, reduce by 25%, not a full exit |
| Confidence | High |
| Not to trade when | You cannot tolerate a drawdown of 35% to 50%, or you need the capital within 5 years |
Dollar-cost averaging means buying a fixed amount at regular intervals rather than all at once. The kill rule compares the equity market's cyclically adjusted earnings yield, one over the CAPE ratio, with the yield on 10-year Treasuries (Chapter 10B). Suppose, hypothetically, that the CAPE ratio is 32 and the 10-year yield 4.5%. The earnings yield is , and the gap is , below −1%. If that gap persists and earnings growth over 12 months turns negative, a $550,000 equity sleeve is cut by 25%, to $412,500. Either condition alone does nothing.
Card 02: VRP harvest
| Field | Rule |
|---|---|
| Edge source | Risk premium, Source 1 |
| Category | Foundation premium |
| Evidence tier | T1: Bondarenko; Carr and Wu; Jurek and Stafford |
| Regime required | VIX/VIX3M below 1.00, 21-day realized volatility below implied, and the TED spread below 100 bps |
| Instrument | XSP options, the mini-SPX, taxed 60/40 as Section 1256 contracts |
| Entry | Sell an at-the-money monthly put with 30 to 45 days to expiry, cash-secured with full collateral against the strike |
| Exit | At 21 days to expiry, close and roll. Or, once the gain on the mark exceeds 50% of the premium, take the profit |
| Expected return | Premium of about 2% to 3% of the strike a month. Net, about 3% to 5% a year after losses, for a realistic retail account |
| Hidden exposure | A tail beta of about 1.0 in the left tail. The skew tax: a skewness of −2.09, against −0.81 for SPX |
| Stress loss | Per at-the-money contract on a 580 strike: S1 −$3,500 on the card, but −$4,100 or worse once gamma is counted, as the example below shows. S2 −$10,000. S3 −$20,000 |
| Sizing cap | 10% to 15% portfolio risk. At most contracts, with set at 5% of the portfolio |
| Kill rule | Layered. VIX/VIX3M above 1.00 for 3 closes: halt new entries. Above 1.05 for 3 closes: cut the size by 50%. 21-day realized volatility above implied: halt new entries. A sleeve drawdown above 20% peak to trough: close entirely and reassess. Re-entry after VIX/VIX3M has closed below 1.00 for 5 days: 25% of the target, then 50%, then 100% over 3 weeks |
| Confidence | High, with discipline |
| Not to trade when | The term structure is in backwardation; the VRP is negative, with realized above implied; a major macro event is imminent, such as an FOMC meeting, a CPI release or a jobs report; or you cannot tolerate 6 months under water |
This card is the VRP harvest of Chapters 10C and 13E, and its kill rule is the phased filter of Chapter 10H. Two of its numbers need a note.
Step 1: the floor. In S1 the index falls 10%, so XSP goes from 580 to 522 and the put is $58 in the money. A premium of 2% to 3% of the strike is $11.60 to $17.40. The put is worth at least its $58 of intrinsic value, so the loss is at least , or $4,060 a contract, and up to $4,640 at the lower premium.
Step 2: where $3,500 comes from. A first-order estimate adds delta and vega. Delta, about −0.5, gives . A one-month at-the-money put on 580 has a vega of about per volatility point, so the VIX's rise of 10 points adds about $670. The total, about $3,570, is close to the card's figure. It leaves out gamma, which is the lesson of Chapter 13E: price a large fall from the option's value at the stressed level.
Step 3: the count. With at 5% of a $1 million portfolio, $50,000, and the S2 loss of $10,000, the cap is contracts. Chapter 13C rounds down to 4 for a buffer, and Chapter 13E's Example 10, with $10,800 a contract, also reaches 4.
Read the card's net return of 3% to 5% a year as a return on the collateral, as Chapter 13E's Example 2 reads its own 2% to 4%. The card's first kill step, halting new entries above 1.00, starts earlier than the single 1.05 trigger of that example.
Card 03: trend following
| Field | Rule |
|---|---|
| Edge source | Mixed: mostly risk premium, some behavioral |
| Category | Foundation premium |
| Evidence tier | T1: Hurst, Ooi and Pedersen, over 140 years; Moskowitz, Ooi and Pedersen |
| Regime required | None: a structural allocation |
| Instrument | DBMF, preferred, or KMLM as the alternative. Not stacked vehicles, unless you deliberately seek leverage |
| Entry | Build over 3 months, not as a lump sum |
| Exit | Hold across cycles. Do not exit automatically during droughts |
| Expected return | A live realized Sharpe ratio of 0.20 to 0.40 for a retail account. About 3% to 6% a year on its own. A larger contribution to the portfolio through its skew in crashes |
| Hidden exposure | Droughts lasting several years: 2011 to 2018 was 7 years. Dispersion between managers of 200 to 400 bps a year |
| Stress loss | S1: −2% to +1%, mostly flat. S2: −2% to +5%, often positive. S3: −5% to +15%, often strongly positive, a crash hedge. In slow drawdowns often positive, as in 8 of the 10 largest drawdowns of a 60/40 portfolio |
| Sizing cap | 10% to 15% of the portfolio |
| Kill rule | DBMF's tracking error to the BTOP50 above 4% a year for 4 quarters, or a drawdown above 25% peak to trough. Not poor performance over 1 year, which happens normally |
| Confidence | High for its role in the portfolio; medium on its own |
| Not to trade when | You cannot tolerate droughts of 3 to 7 years, or you would exit in panic during a choppy regime |
"Stacked vehicles" are funds that use return stacking, holding equity and trend on the same capital, which is leverage (Chapter 10D). Chapter 13E's Example 4 applies this card to an entry during a drought.
Card 04: cross-asset carry
| Field | Rule |
|---|---|
| Edge source | Risk premium, Source 1, with an institutional part |
| Category | Foundation premium, supplementary |
| Evidence tier | T1: Koijen, Moskowitz, Pedersen and Vrugt (2018) |
| Regime required | The TED spread below 200 bps and the MOVE index below 150 |
| Instrument | RSST, a return-stacked product, if available. A self-managed version is complex and not recommended |
| Entry | Not given on the card |
| Exit | Not given on the card |
| Expected return | A live retail Sharpe ratio of 0.15 to 0.30, diversified across currencies, bonds, commodities and equities |
| Hidden exposure | Liquidity events crash carry trades violently, with drawdowns of about 40% in the style of AUDJPY |
| Stress loss | S1: −3% of the allocation. S2: −8%. S3: −15% to −25% |
| Sizing cap | 5% to 10% of the portfolio |
| Kill rule | The TED spread above 200 bps or the MOVE above 150: halt new positions. The carry spread's ratio to its volatility below 0.3: reduce the size |
| Confidence | Medium, since retail access at good quality is limited |
| Not to trade when | In a regime of liquidity stress, or when the only implementation is a self-managed multi-asset book, which is a reason to skip it |
Chapter 10E explains the carry trade and the AUDJPY crash of 2008. Chapter 10D describes RSST differently, as a fund of US stocks and managed futures, so check what a fund holds before you use it for carry.
Card 05: quality
| Field | Rule |
|---|---|
| Edge source | Mixed: risk premium and behavioral |
| Category | Equity factor tilt |
| Evidence tier | T1: Novy-Marx (2013); Asness, Frazzini and Pedersen (2019) |
| Regime required | None: a structural tilt |
| Instrument | QUAL, the iShares MSCI Quality fund, at 0.15%. COWZ, built on free cash flow, at 0.49%, as the alternative |
| Entry | Replace 8% to 12% of the equity allocation with QUAL |
| Exit | Hold across cycles, with tax-aware rebalancing |
| Expected return | A live retail Sharpe ratio of about 0.30 to 0.45, reaching above the SPX baseline of about 0.4. Excess return of about 50 to 150 bps a year |
| Hidden exposure | An equity beta of about 0.95, so it still loses in crashes. Tracking error to SPX of about 3% to 4% |
| Stress loss | About 90% of SPX's loss in crashes |
| Sizing cap | 8% to 12% of the equity sleeve |
| Kill rule | A 24-month rolling Sharpe ratio below 0: reduce by 50%. Tracking error to the academic factor above 4%: switch vehicles |
| Confidence | High: the most durable factor |
| Not to trade when | You would exit in panic during 2 to 3 years of underperformance |
Card 06: momentum
| Field | Rule |
|---|---|
| Edge source | Behavioral, Source 2 |
| Category | Equity factor tilt |
| Evidence tier | T1 with decay: Jensen, Kelly and Pedersen (2023) find about 47% lost after publication |
| Regime required | No sustained equity drawdown longer than 6 months |
| Instrument | MTUM, the iShares MSCI Momentum fund, at 0.15% |
| Entry | 5% to 10% of the equity allocation |
| Exit | Reduce by 50% during a sustained equity drawdown longer than 6 months |
| Expected return | A live retail Sharpe ratio of about 0.20 to 0.35. Excess return of about 30 to 100 bps a year |
| Hidden exposure | Crash risk in V-shaped reversals: −73% in 3 months in 2009. High turnover, about 80% a year, which is tax-inefficient in a taxable account |
| Stress loss | In a slow drawdown, tracks SPX. In a rally from a V-shaped bottom, an additional 20% to 40% from a crash specific to momentum |
| Sizing cap | 5% to 10% of the equity sleeve |
| Kill rule | After a sustained equity drawdown of 6 months or more, reduce by 50% until the 3-month equity return is positive. Tracking error to UMD above 4%: switch vehicles |
| Confidence | Medium: real but decayed |
| Not to trade when | In a taxable account, if alternatives are available; during a V-shaped reversal out of a bear market |
UMD, for "up minus down", is the academic momentum factor: the return of recent winners minus that of recent losers (Chapter 10F).
Card 07: value
| Field | Rule |
|---|---|
| Edge source | Mixed: behavioral and risk premium |
| Category | Equity factor tilt |
| Evidence tier | T1 with controversy: Arnott, Harvey, Kalesnik and Linnainmaa (2021) |
| Regime required | None: structural, and it needs patience over decades |
| Instrument | RAFI fundamental funds, FNDX and FNDA, or AVUV, small-cap value with a quality screen |
| Entry | 3% to 8% of the equity allocation |
| Exit | Hold through drawdowns lasting several years |
| Expected return | A live retail Sharpe ratio of about 0.10 to 0.30 above SPX. Highly cyclical |
| Hidden exposure | Documented drawdowns of 5 to 10 years; 2010 to 2020 was the worst value drawdown in modern history |
| Stress loss | Not given on the card |
| Sizing cap | 3% to 8% of the equity sleeve |
| Kill rule | A 24-month rolling Sharpe ratio below −0.3: reassess, but do not exit automatically, and do not cut automatically during a drawdown lasting several years |
| Confidence | Medium: real, but punishing along the path |
| Not to trade when | You would exit in panic during a 5-year drawdown, or your horizon is shorter than 20 years |
Card 08: low volatility
| Field | Rule |
|---|---|
| Edge source | Institutional, Source 3, with a behavioral part |
| Category | Equity factor tilt: the long-only version of betting against beta |
| Evidence tier | T1 with a caveat: Novy-Marx and Velikov (2022) reduce the Sharpe ratio of Frazzini and Pedersen's version |
| Regime required | Avoid liquidity stress |
| Instrument | USMV, the iShares minimum-volatility fund, at 0.15%, or SPLV |
| Entry | 3% to 8% of the equity allocation |
| Exit | Hold, and reduce slightly in liquidity stress |
| Expected return | A live retail Sharpe ratio of about 0.25 to 0.40, with lower drawdowns than SPX |
| Hidden exposure | An equity beta still of about 0.6. It underperforms in high-beta rallies, as in 2020 and 2021 |
| Stress loss | Not given on the card |
| Sizing cap | 3% to 8% of the equity sleeve |
| Kill rule | In liquidity stress, with the TED spread above 200 or the MOVE above 150, reduce by 25% |
| Confidence | Medium: real, but smaller than the headline Sharpe ratio of Frazzini and Pedersen |
| Not to trade when | Never in the highly leveraged long-short version, which a retail account cannot size safely |
The four factor cards add up to a sizeable share of the equity sleeve. Their ranges sum to at the low end and at the high end. Chapter 10F puts the tilts together at 25% to 40% of the sleeve, and Chapter 10A at 15% to 30%. On a $550,000 sleeve, 19% to 38% is $104,500 to $209,000 in factor funds, with the rest in the broad index.
Card 09: FOMC cycle overlay
| Field | Rule |
|---|---|
| Edge source | Institutional, Source 3: the Fed's reaction function |
| Category | Structural overlay |
| Evidence tier | T2: Cieslak and colleagues (2019), with thin out-of-sample replication after 2019 |
| Regime required | A normal FOMC schedule, not an emergency intervention |
| Instrument | Time discretionary equity entries to the start of even weeks. Do not rotate SPY, which is tax-inefficient |
| Entry | Modulate, do not rotate. In a tax-deferred account, an optional rotation between 1.0 and 0.75 times equity |
| Exit | Not given on the card |
| Expected return | An addition of 0.10 to 0.30 to the Sharpe ratio, with caveats. Possibly decayed after 2015 |
| Hidden exposure | Decay, if the Fed's transparency keeps improving |
| Stress loss | Not given: a modulation only |
| Sizing cap | Modulation only, not a new sleeve |
| Kill rule | If the 2-year rolling return of the odd weeks reaches or exceeds that of the even weeks, suspend. If implementation costs exceed 30 bps a year, skip |
| Confidence | Low to medium |
| Not to trade when | In a taxable account, where tax acceleration kills a small alpha; when the cycle is disrupted, as in the hiking cycle of 2022 |
The card allows an optional rotation in a tax-deferred account, which Chapter 13E's Example 5 advises against even there, and Chapter 10G's rule cuts equity by 20% to 30% in odd weeks. Chapter 13E sets the three readings side by side.
Card 10: end-of-quarter rebalancing
| Field | Rule |
|---|---|
| Edge source | Institutional, Source 3: flow driven by mandates |
| Category | Structural overlay |
| Evidence tier | T2: Etula and colleagues (2020); NBER Working Paper 33554 |
| Regime required | A divergence of more than 1.5% between SPX and bonds during the quarter |
| Instrument | ES and ZN futures, taxed under Section 1256 |
| Entry | Direction: if stocks outperformed bonds, short ES and long ZN; if stocks underperformed, long ES and short ZN. This is the sign correction from the first version of the framework, so do not reverse it. Enter 5 trading days before the quarter ends |
| Exit | 1 to 3 trading days into the new quarter |
| Expected return | About 17 bps of next-day equity effect at an extreme divergence. Untested live for retail |
| Hidden exposure | A recent paper, with uncertain retail replication |
| Stress loss | Not given on the card |
| Sizing cap | 0.5% of the portfolio at risk per leg for the first 8 quarters |
| Kill rule | After 4 quarters, a hit rate below 50% or a negative P&L: suspend. A quarterly divergence below 1.5%: skip the trade |
| Confidence | Low to medium: T2, and replication is needed |
| Not to trade when | The quarter's divergence is below 1.5%, or your own P&L record is shorter than 4 quarters and negative |
The direction rules of the end-of-quarter (EOQ) trade are called out because an earlier version of the framework had the sign backward, and a reviewer caught it (Chapter 10G). A reversed sign does not merely fail to earn: it takes the side of the flow it meant to front-run. The card keeps the size small for 8 quarters, while Chapter 13E's Example 6 lets it grow after 4 good ones.
6 · Filter cards
A filter card is not a strategy. It holds no capital, and it sizes the cards above up or down, as Chapter 10H explains. The four filters are the Edge Book's.
Card F1: the VIX term structure
| Field | Rule |
|---|---|
| Filter | The ratio of the VIX to the VIX3M |
| Use | Modulate the size of the short volatility sleeve |
| Rule, layered | Above 1.00 for 3 closes: halt new short volatility entries. Above 1.05 for 3 closes: halt all put sales and reduce existing positions by 50%. Above 1.15 for 3 closes: close non-essential short volatility. Below 1.00 for 5 closes: re-enter in phases, 25%, 50% and 100% over 3 weeks |
| Do not | Buy VXX or VIX futures because of this filter |
Card F2: realized against implied volatility
| Field | Rule |
|---|---|
| Filter | 21-day realized volatility against current implied volatility |
| Use | Confirm the short volatility regime |
| Rule | Realized below implied: the regime is acceptable for selling premium. Realized above implied: the VRP is negative, so halt new entries. Realized above implied plus 5 volatility points: close existing positions if they are oversized |
| Do not | Treat it as a standalone signal. Always combine it with the term structure |
Card F3: the liquidity composite
| Field | Rule |
|---|---|
| Filter | The TED spread, the MOVE index and the high-yield option-adjusted spread (HY OAS), combined |
| Use | Reduce all leveraged exposure |
| Rule | The TED spread above 200 bps, the MOVE above 150, or HY OAS above 600 bps: reduce all leveraged exposure by 50% and halt new positions |
| Normalization | Wait for a sustained recovery before re-entering |
Card F4: the sign of GEX, at low confidence
| Field | Rule |
|---|---|
| Filter | The direction of gamma exposure (GEX), positive or negative |
| Use | Soft regime context only |
| Rule | Negative GEX: expect higher realized volatility, and trim put-writing by 10% to 15%. Positive GEX: expect compressed volatility, and size normally |
| Do not | Trade specific GEX levels, such as a call wall or a put wall, or size positions by the magnitude of GEX |
| Note | A Cboe study of 2025 puts 0DTE market-maker (MM) hedging at only about 0.2% of liquidity. Use the sign as a supplementary signal only |
Chapter 10H states two of these rules a little differently. Its liquidity rule also cuts short volatility by 50% and equity beta by 25%. It gives the sign of GEX only 10% to 15% of one filter's weight, and ignores it when it disagrees with the other filters.
7 · Do-not-trade cards
These cards exist to inoculate you against narrative-driven retail bait. Both sources wrote them, and where both rejected the same strategy, the two cards are merged here. The Edge Book's cards are X1 to X7, and the Dealer Mechanics Lab's three others follow as X8 to X10.
Card X1: the drift before FOMC announcements
| Field | Content |
|---|---|
| Why you might be tempted | Lucca and Moench (2015) documented an average return of 49 bps before FOMC announcements, from 1994 to 2011 |
| Why it is dead | Kurov, Halova, Wolfe and Gilbert (2021) found the effect disappeared after 2015. The cause was the Fed's transparency reforms: the dot plot in 2014 and forward guidance |
| Lesson | Edges with a clear mechanism can decay suddenly when policy responds, or when the institutional context beneath them changes |
| Verdict | Do not trade |
The pre-FOMC drift is Chapter 10I's first decayed edge. The Lab's card rejects it too, with the same two papers.
Card X2: S&P 500 inclusion
| Field | Content |
|---|---|
| Why you might be tempted | A historical inclusion premium of 7% or more (Shleifer 1986) |
| Why it is dead | Greenwood and Sammon (2024) find it under 1% after 2010. Index funds got smarter, and arbitrageurs ate the obvious mispricing |
| Lesson | Sophisticated capital can eat an institutional edge |
| Verdict | Do not trade |
Card X3: naive earnings short volatility on single names
| Field | Content |
|---|---|
| Why you might be tempted | The IV crush after earnings takes 30 volatility points or more, and premiums look "rich" 1 to 2 days before |
| Why it is wrong | Adams and Neururer (2020) find a negative mean despite a positive median at retail size. The structural negative skew of a single stock's earnings exceeds the premium collected: the skew tax exceeds the premium harvested |
| Institutional variant | The cross-sectional version of Cao, Han, Tong and Zhan (2022), across 200 or more names, works. A retail account cannot replicate its infrastructure |
| Lesson | A win rate is not an expected value (EV). Structural negative skew dominates at single-name retail size |
| Verdict | Do not trade it systematically. Acceptable: a small calendar spread on a rare opportunity |
Chapter 13E's Example 8 works this card through an NVDA report, where the expected value comes out at about −$4 per option sold.
Card X4: scalping GEX levels
| Field | Content |
|---|---|
| Why you might be tempted | Public GEX models show specific call walls and put walls, with a compelling story about dealer hedging |
| Why it is wrong | Public GEX models depend on their assumptions, and vendors disagree. The explosion of 0DTE options after 2022 changed the structure. A Cboe study of 2025 puts 0DTE market-maker hedging at about 0.2% of SPX daily liquidity, as the Edge Book's card says. The Lab's card says "futures liquidity": that is the base of Cboe's 2023 note, which put the flow at 0.04% to 0.17% of S&P futures liquidity |
| Lesson | A compelling narrative is not evidence quality |
| Verdict | Do not trade specific levels. Acceptable: the directional sign, as low-confidence regime context |
The Cboe study of 2025 is Mandy Xu's "0DTEs Decoded" (May 2, 2025). Because it measures same-day hedging only, using it against walls built mostly from longer-dated open interest extends the measurement. Both traditions reject the scalp, as Chapters 1G, 10A and 10I note. They differ on how much weight the levels deserve for judging a trade's path, and the Lab's other chapters give them more (Chapters 8C and 11A).
Card X5: systematic iron condors
| Field | Content |
|---|---|
| Why you might be tempted | A "defined risk" income strategy in markets that are usually range-bound, with theta accumulating every day |
| Why it is wrong | The CBOE CNDR index returned about 690% from 1986 to 2024, against about 5,500% for the SPX total return. That underperformance over 38 years is decisive. Wing breaks dominate the aggregate P&L |
| Lesson | A defined maximum loss per trade is not a defined aggregate loss |
| Verdict | Do not trade systematically. Acceptable: a small, discretionary condor on an occasional opportunity at extreme implied volatility |
The two totals are compound rates of about 5.5% and 11% a year (Chapter 10I). Chapter 12C teaches the iron condor as a structure chosen for a regime, and the Lab's own card agrees that the systematic version fails.
Card X6: a put spread funding a LEAP
| Field | Content |
|---|---|
| Why you might be tempted | "Short premium funds the long LEAP", and "leveraged equity exposure with a VRP edge" |
| Why it is wrong | Decomposed, it is about 1.4 times SPX with an embedded leverage drag of 50 to 150 bps. The net edge over simple leverage is approximately zero, and the path is worse than that of ES futures |
| Lesson | Decomposition kills false claims of edge |
| Verdict | Do not trade. Use instead ES futures, a 2 times leveraged ETF, or financing through a box spread with SPY |
Chapter 13E's Example 9 takes apart this pairing of a short put spread with a long-dated LEAPS call.
Card X7: VIX exchange-traded products, long or short
| Field | Content |
|---|---|
| Long VIX, such as VXX and UVXY | VXX lost 99.84% over its life, from 2009 to 2024. Roll decay costs 5% to 15% a year in normal times, and much more in extended contango |
| Short VIX, such as the former XIV | XIV was terminated after losing 84% to 96% of its value in a single day in February 2018. SVXY has been de-leveraged, but it is still subject to catastrophic moves |
| Verdict | Do not trade either side |
| Use instead | SPX or XSP options directly, with explicit position management |
Card X8: buying the Friday rip into Monday
| Field | Content |
|---|---|
| Why you might be tempted | The Monday-morning dealer buying that charm sets up over a weekend, as Chapter 5D describes |
| Why it is wrong | The charm-driven Monday-morning flow is roughly 10 to 20 bps on average, too little to justify a swing trade |
| Lesson | Mechanical effects exist but are usually too small to trade as standalone signals |
| Verdict | Do not trade |
Card X9: holding through events at full size
| Field | Content |
|---|---|
| Why you might be tempted | A working swing position, and a scheduled event that may move it further your way |
| Why it is wrong | Event-day implied moves are typically 1.5 to 3 times the normal daily range. A standard swing stop can be gapped through entirely |
| Lesson | A stop caps a loss only if the market trades at the stop, and a gap opens beyond it |
| Verdict | Close before the event, or convert the position to a defined-risk structure |
With a normal daily range of 1%, an event day priced at 1.5% to 3% can open beyond a stop set 1% away, and the loss is then set by the opening price.
Card X10: revenge trading after a loss
| Field | Content |
|---|---|
| Why you might be tempted | The urge to make a loss back at once |
| Why it is wrong | Trades taken after a loss show systematically worse outcomes |
| Lesson | Structure, not willpower, is the defense: the rule of a 30-day cool-off after a 25% drawdown (Chapters 13C and 13G) |
| Verdict | After taking a planned loss, do not trade for the rest of the session |
8 · How to use the cards
Both sources give the same five steps, and the Edge Book adds detail to two of them:
- Print or copy the cards relevant to your portfolio.
- Refer to them before trading, and verify every condition on the card.
- Update them when the regime changes, marking each card "halted" or "active" with the date of the change.
- Review them quarterly, confirming that no kill rule has triggered.
- Recalibrate them once a year against the latest evidence.
The cards are deliberately rigid, and their value lies in being binding when you are tempted to override them.
Keep the cards where you place orders, not in a folder. Mark each allocatable card with its current status and the date you last checked its kill rule, and keep the do-not-trade cards in the same place. The temptation they answer arrives as a new story, not as the old name. When a pitch reaches you, find its card before you read the pitch a second time, since most retail trading content promotes the strategies these cards reject (Chapter 10I). Record every card you overrode in your journal (Chapter 13H), with the reason, and review those entries first each quarter.
9 · Rigidity and judgment
Some readers object that trading requires judgment, not rigid rules. Trading at the level of a signal does require interpretation, but trading at the level of a strategy requires written rules. The card format formalizes the strategy level. Within each card, you exercise discretion about whether its conditions apply right now, but the cards themselves are not for daily revision.
The balance is discretion at observation and mechanism at execution. When you see something interesting, you observe and interpret it, which is the observation step of Chapter 13C's pipeline. When you decide to act, the cards constrain how. This is what disciplined trading looks like in operation.
10 · The whole framework on one card
The Dealer Mechanics Lab closes its discipline section with the book in one paragraph. Classify any new "edge" by its source: risk premium, behavioral, institutional or microstructure. Score its conviction, , and apply the four-layer retail haircut. Stress-test it through S1 to S4, specify its kill criterion at all three levels, and then size for survivability with .
If you do this consistently, you filter out about 80% of retail trading content as low-conviction or rejected. The remaining 20% gets sized to anchor a portfolio that lasts decades. The framework's success depends not on finding rare edges but on operating consistently with the edges that are available. Most retail traders have access to the same setups, the same indicators and the same information, so the differences in outcomes come from differences in execution discipline. Everything else in this book, in the Lab's words, was the context that justifies the discipline.
A card is only as current as its last review. A regime can change the meaning of a rule before anyone rewrites it, as the hiking cycle of 2022 may have done to the FOMC cycle. An edge can decay while its card still reads "active", as the drift before FOMC announcements did after 2015. The rigidity that protects you from impulse also carries a stale rule forward, which is why the cards have a quarterly review and an annual recalibration.
- Trade only what has a card, and only when every regime condition on it holds.
- Where two cards or chapters disagree, take the stricter number until you have reviewed the evidence.
- Size from the card's rule, never from how good the setup looks.
- Follow the kill rule the moment it fires, and change a card only at a scheduled review, never during a trade.
- Before any new strategy, look for it among X1 to X10.
Loosening a skip condition because today's setup looks unusually clean, or widening a stop because the price is close to it, turns the card back into an opinion. The second is the stop migration of Chapter 11N. Changes belong to the quarterly review, made in calm conditions and written down. A card edited mid-trade protects nothing.
The cards reject structures, not tickers. A condor sold on QQQ instead of SPX, or earnings premium sold through a new ETF, is the same strategy under a new name. Match a pitch to a card by what it does, sells or depends on, not by what it is called.
- A setup card fixes a strategy's edge, regime, vehicle, entry, size, exits and kill rule in advance, so that nothing is decided at the screen.
- The Lab's four swing cards summarize Chapter 11H; where a card and that chapter differ, take the stricter number.
- The Edge Book's ten allocation cards cover the foundation premia, four factor tilts and two overlays, each with its stress loss and kill rule.
- Filter cards hold no capital and size the other cards, and the ten do-not-trade cards answer the pitches you meet most often.
- Discretion belongs to observation, and mechanism to execution. Cards change only at scheduled reviews.
11 · Check yourself
Question 1
Your readings: VIX/VIX3M at 1.03 for 4 closes, 21-day realized volatility of 15% against implied of 19%, the TED spread at 40 bps and the MOVE at 110. Which of Cards 02, 03 and 04 allow new positions?
Answer
Card 02 does not. Its regime requires VIX/VIX3M below 1.00, and F1 halts new short volatility entries above 1.00 for 3 closes. The ratio has not reached 1.05, so existing puts are not cut. Card 03 has no regime requirement, so trend allocations proceed. Card 04 requires the TED spread below 200 bps and the MOVE below 150, and both hold, so carry may add positions.
Question 2
A friend shows you a year of monthly iron condors that won 11 times out of 12. Which card applies, and what evidence overrides the temptation?
Answer
Card X5. A high win rate is exactly the pattern of a defined-risk short volatility strategy: small wins, occasional wing breaks. The CNDR index returned about 690% from 1986 to 2024 against about 5,500% for the SPX total return, 38 years of underperformance. Defined loss per trade is not defined loss in aggregate.
Question 3
What is the difference between a setup card and a filter card?
Answer
A setup card describes a strategy you allocate capital to: its edge, entry, size, exits and kill rule. A filter card describes a measurement that holds no capital and sizes other strategies up or down, such as the VIX term structure. Trading a filter directly, for example buying VXX because the VIX is in contango, is the classic error the filter cards warn against.
Question 4
What role does the kill rule play on each card?
Answer
It is the pre-committed answer to "what would prove me wrong?", written before the trade in calm conditions. On a swing card it is the invalidation and the time stop. On an allocation card it names a measurable trigger, such as a VIX/VIX3M threshold, a rolling Sharpe ratio or a tracking error, and the action the trigger forces. Its value is that it binds when your judgment is least reliable.
Question 5
Why are Card 10's direction rules called out so prominently?
Answer
An earlier version of the framework had the sign backward. If stocks outperformed bonds, rebalancers sell stocks and buy bonds, so the front-runner goes short ES and long ZN. A reversed sign takes the opposite side of the very flow the trade exists to anticipate, so the error turns a small expected gain into a small expected loss.
Question 6
During FOMC week, your normal Trend Pullback risks 1% of a $400,000 portfolio with a stop $4 below entry on SPY. What size and stop does the overlay allow?
Answer
Normal size is shares. The overlay halves it to 500 shares and widens the stop by 50%, to $6. The risk becomes , 0.75% of the portfolio, below twice the normal risk, so the entry is allowed.