Market Drop Wins — Master Log v2

Complete project context · September 2026 Supersedes v1. Self-contained: this document alone provides everything needed to resume work.

Companion artifacts: market-drop-wins-library-v2.md (the strategy library) · market-drop-wins-mvp-plan-v1.md (next phase)

Project Origin

Initial goal was to upgrade my personal knowledge, preparation, and plan to benefit from future stock market drops. Scratch own itch. Specifically deeply research strategies where asymmetric reward to risk payoffs are probable, especially for someone with high financial knowledge and net worth (like me).
With stock markets “highly valued” in Sept 2026, this topic is timely, and a potential opportunity, especially in terms of knowledge and preparation, personally, financial advisors, and other individuals.
Expanded into additional $MART DEBT strategies, which could be a small lead-in offering for SDC target market of financial advisors.
When AI was pushed to find more strategies, especially the “Can’t lose” variety, a large library of strategies was created, categorized in many meaningful dimensions.

1. PROJECT DEFINITION

1.1 What this is

A multi-phase project building a strategy library on how individuals and advisors can benefit from significant market declines, rather than merely survive them.

Significant decline: ≥20% peak-to-trough in a broad equity index
Jurisdictions: Canada primary, United States secondary
Benefit horizon: 10 years default (a parameter, not a constant)
Audience: financial industry professionals primarily (mid-career advisors, dealer/compliance, money managers, investment lenders); retail investors secondary

1.2 Relationship to Smart Debt Coach

Market Drop Wins and Smart Debt Coach are not the same asset. They share a methodology — how strategies, tools and offerings get created for advisors and individuals, and the target market. Market Drop Wins is the first iteration of that methodology, built and tested here before being applied to Smart Debt Coach. Some outputs feed directly into Smart Debt Coach.

Consequence: the primary deliverable of each phase is a repeatable process for producing analysis, not the analysis itself. Every method decision below is a reusable asset.

1.3 Product architecture (five layers)

Library — named, categorized strategies
Micro-apps — fast per-individual analysis
Process — education and implementation, advisor-guided
Monitoring — ongoing management
Triggers — utilities that notify when conditions are met

Layers 1–2 are scalable and defensible. Layers 3–5 carry the recurring revenue and the client relationship. Strategies requiring monitoring are strategically more valuable than equally beneficial one-shot strategies, even at lower expected value.

1.4 Named frameworks

Buy More Low — tranched deployment at pre-set decline thresholds. Held for years, considered a unique innovation. Parked — its own large project.
Debt Swap — converting non-deductible to deductible debt. Three members: cash damming (cash-flow based), classic (non-registered cash, no toll), equity (gains toll, shrunk by a decline).
The Worth It filter — screening out efforts whose benefit doesn’t justify the effort. From earlier myBetterRates business/project. The gate most frameworks lack.
F.A.S.T.T. — Focused, Adaptive, Simple, Tailored, Trustworthy. Trustworthy added because Smart Debt strategies are controversial and higher risk. Full definition lives in the Claude Code knowledge base.
Can’t Lose — the settled label for guaranteed-benefit strategies (not “guaranteed”). Also negative risk inside the Smart Debt universe.
Client-First and world class — the foundation, not features.

1.5 Two-tier offering

Core — selected by applicability breadth. Premium — segment playbooks, not “more strategies,” because a playbook maps to a specific high-value client in the advisor’s book while a feature list makes the buyer judge depth. One library, two access levels — parallel documents would diverge within a year.

2. METHODOLOGY (the reusable assets)

2.1 The zero-opportunity-cost filter

For any strategy: does it cost anything while you wait?

Costs something while waiting	Costs nothing while waiting
Idle cash, foregone equity returns, continuous insurance premiums	Unused credit capacity, a written rule, a plan structure, a tax attribute waiting to be created
Evidence actively negative	Evidence neutral-to-positive

Build predominantly from the right column. The Prepare phase is not “raise cash,” it is “install free options.”

2.2 Behavioural primacy

Theory and reason do not govern investor conduct in a crisis. Emotion does, and fear most of all.

The correct optimization target is not the strategy that wins on paper, but the strategy the investor will actually hold through the decline.

Every candidate carries an implicit second rating: adherence probability under fear. A strategy scoring 8/10 on expected value and 3/10 on adherence is worse than one scoring 6 and 9.

Three consequences:

Simplicity and pre-commitment are performance features, not concessions
Peace of mind is a legitimate optimization target. For the financially independent, additional wealth has low marginal utility while drawdown pain is fully felt
Strategies executed by a professional have high adherence despite high complexity — the delegated execution quadrant, and the strongest argument for an advisor-mediated product

2.3 The five-number card

EV alone maximizes terminal wealth, which is not everyone’s objective. Ranking by EV systematically misadvises the segment most worth serving.

Metric	Answers
EV at 10 years	Expected benefit in dollars
Hit rate	Probability benefit exceeds zero
Downside case	Realistic bad outcome (5th percentile)
Effort	Setup hours + ongoing hours/year
Peace-of-mind delta	−2 to +2: does this make a decline easier or harder to live through?

The fifth is the differentiator — every competing framework implicitly rates it zero. Two derived rankings: Wealth rank (EV-weighted) and Peace rank (certainty-equivalent weighted). Divergences are the interesting cases.

EV structure: EV = P(trigger within horizon) × EV(benefit | trigger) − cost of waiting. The first term is a shared parameter to be estimated once and reused everywhere.

2.4 Split confidence

Scale: Low · Medium · High · Very High · Guaranteed (100%)

Mechanism confidence — will it do what it claims, given its precondition? Guaranteed genuinely occurs here.
Outcome confidence — will it deliver net benefit at 10 years? Almost never Guaranteed.

Example: capital loss carryback is Guaranteed/Guaranteed (rare and valuable). Rate hold is Guaranteed/Very High. TFSA room amplification is Guaranteed/High. Post-decline leverage is High/Medium.

Guaranteed-mechanism strategies are the natural shortlist for peace-of-mind optimizers — the only uncertainty is whether the trigger occurs.

2.5 Two categorization axes

Axis 1 — Risk ladder (also the client conversation order, and the Smart Debt playbook): 0. Can’t Lose / negative risk · 1. Low risk · 2. Moderate risk · 3. Higher risk

Start at tier 0 and progress until needs are satisfied or risk tolerance thresholds are met. Most clients never pass tier 1, and saying so is a trust signal.

Axis 2 — Verdict: Recommended · Situational · Not Worth It · Harmful · Avoid (personal overlay, always with its reason).

Not Worth It and Harmful must never share a label. One wastes time; the other costs money.

Placement: the Harmful list belongs in core and arguably the free layer — highest-trust, lowest-cost content, and the best proof that 60+ strategies were examined rather than curated. Not Worth It belongs in premium as an appendix.

2.6 The availability axis

Always (benefit exists any time) · Amplified (larger during a decline) · Drop-dependent (exists only because of the decline).

Marketing line: Most of these you can do today. They just get better when markets fall.

Amplification rule: in a downturn, the identical ask gets a better answer. Hungrier providers improve terms across the whole one-sided-comparison family.

Why it matters commercially: a pure market-drop offering is dormant most of the time. An offering that opens with always-available can’t-lose strategies is not. The decline material becomes the upside rather than the premise.

2.7 Standing rules

Total returns, always. Every equity outcome analysis uses total return. Price-only understates outcomes systematically and more for Canadian investors — TSX yield ~3% vs S&P ~1.5%. Exception: triggers, measured on nominal price index because a trigger must be publicly published and impossible to argue with. Note that a price-index recovery to a prior high takes considerably longer than a total-return recovery.

Big Rocks only, biggest first. Work the highest-impact item, finish it, move. When something new looks appealing, the test is whether it displaces the current biggest rock — not whether it has merit.

The purpose filter: does this lead toward Smart Debt, or away from it? This governs all future scope decisions.

The reduction is the product. Judge the framework on how much it correctly excludes.

2.8 Decline type — a permanent dimension

Type	Characteristics	Examples	Opens	Closes
A. Deflationary	Rates cut, bonds rally, spreads widen, USD strengthens	2008–09, 2020, 1987	Rate channel, refinancing, prescribed-rate loans, CAD cushion	Little
B. Inflationary	Rates rise, bonds fall with stocks	2022	Tax-loss harvesting (bonds too), valuation reset, Roth conversions	The entire rate channel; leverage cost rises; nothing to rebalance into
C. Valuation unwind	Slow, prolonged	2000–02	Tax attributes, compensation resets	Anything needing a V-shape
D. Liquidity shock	Violent, brief	Mar 2020, Oct 1987	Forced-seller liquidity, CEF discounts, extreme premiums	Anything slow; credit freezes

Any borrower strategy depending on falling rates is Type A/D only. Marketing “market drops help borrowers” without this qualifier would have been actively wrong in 2022. Any system that doesn’t handle 2022 explicitly will be tested against it by the first sceptical advisor.

3. KEY FINDINGS

3.1 The buy-the-dip critique, and its bounds

AQR (2025): across 60+ years and 196 implementations, holding cash to deploy on declines underperformed passive investing in >60% of cases; average alpha ~0.5%/yr, statistically indistinguishable from chance; 18.7% lower ending wealth than dollar-cost averaging new savings. PWL Capital reached the same conclusion across six country indexes plus MSCI World.

Two bounds: (1) it is an unconditional average — conditioned on the highest-valuation starting points the case is materially weaker, and this should be re-tested on the top valuation quintile; (2) it optimizes terminal wealth. The critique bites only on strategies that cost something while waiting.

3.2 The leverage premise, as finally stated

Adding leverage after a significant decline is almost always lower-risk than adding the same leverage before it. Whether the resulting risk is low enough, and whether remaining risks can be managed responsibly, is a separate and harder question.

Conflating the two is how responsible leveraging guidance gets misread as promotion.

On eliminable risk: most leverage risks can only be managed — market, sequence, carry, income, behavioural. Margin call risk is the significant exception. It can be eliminated outright and for almost all investors it should be. Forced liquidation converts a recoverable drawdown into permanent loss and removes the investor from the recovery the strategy depends on. This is a precondition, not a refinement.

Qualifiers: Shiller CAPE was near 40.6 in September 2026 — 98.8th percentile since 1881; a 20% drop lands around 32, still above the 1929 peak. 2000–02 was −49% and took roughly a decade to recover nominally. Borrowing cost is not guaranteed to fall (2022). Loan structure matters more to outcome than entry price.

3.3 Decline frequency and continuation (S&P 500, closing, price index)

16 bear markets 1929–2026, cross-checked against Yardeni and Hartford Funds.

Depth	Count	Frequency	P(reaches given −20% hit)
−20%	15	1 per 6.5 yrs	100%
−25%	12	1 per 8.1 yrs	80%
−30%	8	1 per 12.1 yrs	53%
−35%	6	1 per 16.2 yrs	40%
−40%	5	1 per 19.4 yrs	33%
−50%	3	1 per 32.3 yrs	20%

Post-WWII (1946–2026, 14 events): −30% is 1 per 13.3 yrs; P(−30% | −20%) = 46%. Median depth −31.1% full, −28.4% post-war.

Threshold conclusions: −20% is well chosen for threshold 1. “Once a decade” interpolates to about −28%; use −30% (round, within sampling error, and −28% is false precision on 16 observations). Crux: given a −20% decline, roughly a 50% chance it reaches −30% — half the time a second tranche never deploys.

P(no ≥20% decline in 10 years) ≈ 21%. Opportunity capital in cash at 3% while equities compound at 8.5% ends at ~59% of what investing would have produced.

⚠ These are S&P 500 price-index numbers. The trigger index is now the investor’s domestic index (TSX for Canadians) and analysis must be total-return. Recomputation is required before any dollar figures ship.

3.4 Tranching: adherence, not alpha

Design	E[deployed]	E[avg discount]
50/50 at −20/−30	76%	23.5%
60/40 at −20/−30	81%	22.6%
40/40/20 at −20/−30/−40	68%	25.1%
100% at −20 (one shot)	100%	20.0%

On expected capital deployed × expected discount, one-shot deployment wins, and more tranches score lower.

Tranching does not maximize expected return. It maximizes the probability that you act at all.

Mechanisms: it makes the first deployment psychologically possible (the reserve’s function is to enable the deployment, not to be deployed); it bounds regret in both directions; it removes the decision from the moment. Roughly four points of efficiency is the price of a plan that executes.

3.5 The sequencing principle

Cash restores the target allocation. Leverage exceeds it.

Deploying cash costs only the opportunity return on cash, already being paid daily. Borrowing costs the after-tax interest rate ongoing, plus a liability, facility risk, servicing obligations and a magnified drawdown. Borrowing while holding idle cash means paying to rent money you already have. Only exception: where the cash carries a tax toll to access — and even then, plan the extraction ahead of the trigger rather than borrow around it.

3.6 Target allocation

Anarkulova, Cederburg and O’Doherty, Beyond the Status Quo: optimal lifecycle weights near one-third domestic, two-thirds international equities, no material fixed income, at every age including retirement. Block bootstrap over developed-market data pre-1900. TDF investors need ~63% more pre-retirement savings; a 10% savings rate matches a balanced investor’s 19.3%.

Three caveats, the last decisive:

Driven heavily by accumulation. Jeske’s analysis: the all-equity advantage during accumulation carries the retirement conclusion, even though 100% equities in retirement is suboptimal under the paper’s own CRRA utility at gamma 3.84.
It is one-third domestic — a TSX-concentrated all-equity portfolio is not what it supports.
It optimizes expected utility of wealth — strong on wealth outcomes, largely silent on peace of mind.

The buffer reframe: several years of expenses in cash is liability matching, well supported against sequence-of-returns risk. And:

The expense buffer is what makes Buy More Low possible. An investor who must sell equities to eat during a decline cannot also be buying. It is infrastructure, not drag.

Dry powder has two components with different rules: the expense buffer (held permanently, correctly bears its cost, never deployed into the strategy) and opportunity capital (deployed on trigger, failure if never deployed). Merging them is how a plan ends up with no buffer at the worst moment.

3.7 The post-drop convexity test — rejected

Thesis: index options have carried permanent negative skew since October 1987 — OTM puts at materially higher IV than equidistant OTM calls, driven by structural crash-protection demand. So the market overcharges for downside convexity and undercharges for upside. After a deep decline, put skew steepens further while the upside wing stays comparatively cheap, yet the forward distribution has become more positively skewed.

Test: four −30% closing crossings in the index-options era (Oct 1987, Sep 2001, Oct 2008, Mar 2020). Oct 2022 never qualified, stopping at −25.4%. Black-Scholes with call-wing IV at 25/30/35%, deliberately low, biasing the test in favour.

Result, 20% OTM 24-month at 30% IV, equal dollars: 1987 = 2.5x · 2001 = 0.0x · 2008 = 0.0x · 2020 = 8.2x · portfolio 4.00x.

Robustness: excluding 2020, 0.30x. Every ex-2020 configuration loses money (0.18x–0.38x across strikes and expiries). Longer expiry made it worse. −40% trigger version: 0.95x.

Verdict: rejected.

Why it failed — the finding worth keeping. The −30% trigger is not the bottom: 2001 fell from 966 to 777; 2008 from 1,057 to 677 within five months. Both expired before vindication. The binding constraint is expiry, not skew.

A non-callable investment loan deployed at −30% has no expiry, no time decay, no roll, and nothing to lose to the passage of time. For post-drop convexity, the absence of an expiry date is worth more than the convexity itself.

Scripts preserved: skew_test.py, skew_test2.py.

3.8 Tail hedging (the Taleb / Universa question)

Four things get called “the Taleb trade”: (A) continuous tail hedge — pays monthly forever; (B) structural mispricing bet (Burry); (C) post-drop convexity — pays once at the trigger, zero carry while waiting; (D) permanent convexity sleeve. Taleb and Spitznagel’s wealth came predominantly from A and B. Market Drop Wins is about C.

The debate: AQR argues systematic put buying costs more than it saves. Universa argues a small allocation raises compound growth by lowering systematic risk. A 2026 backtest on 17 years of SPY options found the disagreement turns substantially on strike selection — AQR used near-ATM; Universa uses deep OTM, which performed materially better.

Critical finding: AQR’s Still Not Cheap and Israelov–Tummala find neither calm markets nor rising volatility have been useful signals for turning a tail hedge on. “Monitor for cheap convexity and alert” is the most appealing product idea in this space and the evidence is against it.

Cheaper substitute: Man Group finds trend-following plus ~25% tail-hedge options nearly replicates the convexity profile, reducing average monthly returns only from ~0.42% to ~0.32%. Puts hedge crashes faster than a signal can react; trend hedges declines lasting long enough to detect.

Instrument detail: SPX (European, cash-settled) prices in dividends; SPY (American) does not — use SPX for long-dated leverage substitution.

3.9 The convexity budget rule

Kelly is the wrong tool: it requires known probabilities (unknowable for rare convex events); overbetting is far worse than underbetting (1.5x Kelly can turn growth negative; above 2x is worse than not betting); it assumes normality.

The transferable finding: 0.3x Kelly reduces the chance of an 80% drawdown from ~1-in-5 to ~1-in-213 while retaining ~half the growth.

Size by loss tolerance, not by estimated edge.

Five rules: (1) a single number written in advance — what you’d genuinely shrug at, halved; (2) distinguish per-event from annual (a one-time 5% budget at a ~12-year trigger vs 0.75%/yr costing ~7.2% per decade with certainty); (3) cap the lifetime total, not just the bet; (4) no averaging down, no re-upping within a cycle — maximum loss is only defined if it cannot be topped up; (5) written before the trigger.

On Bitcoin: selecting category leaders is a defensible network-effect filter, but leaders still fail and leadership is clearer in hindsight. It is a sizing rule, not a selection rule — a risk-management insight, not an alpha insight. And it is a permanent sleeve strategy, not a market-drop strategy.

The framing that must lead (F.A.S.T.T., trustworthy binding): This is an amount you should expect to lose entirely, most of the time. Here is how to size it so losing it doesn’t matter. Here is what has to be true for it to pay. Here is how often that has been true, and how weak the evidence is.

3.10 The Burry pattern

Not a strategy — a search process. He found a market where insurance was priced by sellers using a model he believed wrong: small known capped cost, enormous payoff if right. Monitorable as a watchlist question: where is insurance currently sold cheaply by parties whose model assumes something that may not hold? Indicators: credit spreads vs fundamentals and leverage; private credit marks vs public; concentration in index products; correlation assumptions in structured products. Highest payoff, lowest confidence. Watchlist only.

4. COMPETITIVE LANDSCAPE

4.1 Opportunity scanning / “tax alpha” — real, funded, entirely American

Holistiplan is the category leader. Launched 2019; OCR-scans a tax return in under 60 seconds against a strategy library and produces a branded client-ready report. 30,000+ subscribers; ~39% market share in the 2026 T3/Inside Information survey; first for five consecutive years in T3 and Kitces surveys. Expanded into estate document review and insurance analysis. A few hundred dollars a month, often bundled.

Competitors: FP Alpha, Corvee, TaxPlanIQ, Covisum Tax Clarity, Income Lab, RightCapital Tax Analyzer, Nitrogen AI Tax Center.

Gaps: US only · input is a filed return (a historical document) · tax only, no debt or leverage · annual trigger, no event monitoring · no market-drop dimension · no can’t-lose / risk-ladder organization.

Key observation: a tax return cannot see most of this library — no mortgage renewal date, HELOC spread, RESP grant room, TFSA successor designation, RRIF age election, DTC eligibility, or vehicle loan terms.

4.2 Market-drop offerings — thin and non-commercial

Advisory blog posts titled “Bear Market Playbook” listing four to six items. Asset manager content (MFS, AssetMark) on client communication and retention. Behavioural coaching kits — Don Connelly’s Volatility & Bear Market Tool Kit at $149 is representative.

Not found: any systematic library, Canadian-specific treatment, analytical tooling, trigger monitoring, treatment of borrowing as a decline opportunity, risk-laddered structure, or documented negative results. The material is overwhelmingly behavioural — how to stop clients selling — rather than strategic.

4.3 The positioning gap

A Canadian, risk-laddered strategy library that starts from can’t-lose, includes debt and leverage, is driven by event triggers rather than an annual return, and is supported by analysis tools.

Nothing found occupies it. The US analogue proves the model works commercially while leaving the Canadian market, the debt dimension, the trigger dimension and the market-drop dimension unserved.

Caveats: web search, not a market study — bank- and dealer-internal tools would not surface. And Holistiplan’s dominance is a warning as well as validation.

5. TALBOT’S PERSONAL APPLICATION

5.1 Situation

Financially independent; not optimizing for wealth; optimizing for peace of mind. High financial knowledge; accepts moderate risk especially where reward is asymmetric; acts only on high-confidence strategies.

Zero debt of any kind. HELOC in place with several hundred thousand of capacity. Dry powder in three pockets — corporate, personal, spouse’s — with more than enough to fill TFSA balances at any moment. Several hundred thousand of unused TFSA room deliberately preserved for market-drop deployment. Significant RRSP assets both sides. Hundreds of thousands in a personal corporation now functioning as a holding company. Personal holdings between spouses approximately balanced. Wife salaried at 45% marginal; he draws from the corporation at 38%. Recently reduced equity allocation, realizing roughly $20,000 of capital gains.

5.2 Personal strategy — concluded

A simple system that triggers redeployment of cash back to target equity allocation at a −20% drop, and at a −30% drop draws on established and confirmed credit capacity to leverage an amount to be determined. The leverage amount is ballparked against what was done previously; that is sufficient for now.

Design decisions settled:

#	Decision	Settled
1	Measured from	Closing price, prior all-time high. Outcome analysis uses total return.
2	Index	Investor’s own domestic index. TSX for Canadians.
3	Currency	Investor’s local currency.
4	If a threshold never fires	Accept it — including threshold 1.
5	Both fire quickly	Deploy both.
6	Re-arming	Yes, on recovery to a new all-time high.
7	Nominal or real	Nominal price index change.
8	Tranche 1 source	Cash, sized to restore target allocation.
9	Tranche 2 source	Leverage, justified and sized independently.

Personal findings: the HELOC is capacity but the freezable kind — no market-value margin call (the lender looks at the house, not the portfolio) but typically demand-callable and limit-reducible, with falling home values the trigger. Consequence bounded to tranche 2, which fires ~half the time. TFSA should receive highest-conviction long equity, not asymmetric bets — losses are permanently unusable and destroy irreplaceable room. LEAPS deferred; the Ayres–Nalebuff lifecycle argument rests on young investors’ unrealized human capital and is silent on a financially independent investor. Corporate equity meltdown dropped from personal scope.

5.3 The only Stage 1 artifact: the Monitor Market utility

To be built in Claude Code using the existing custom notification utility.

Function: daily check of closing levels for the Canadian broad equity index and optionally the US index; notify when a drawdown threshold is crossed.

Requirements: daily close only, no intraday or streaming · drawdown from prior all-time closing high, nominal price index · thresholds −20% and −30%, configurable · fires once per threshold per cycle · re-arms on recovery to a new all-time closing high · notification via the existing utility · persist the running all-time high and each threshold’s armed/fired state between runs · log every daily reading for audit and future back-testing.

Out of scope v1: total-return series, portfolio-specific drawdown, position sizing, recommendation logic, any UI beyond the notification.

Build so a second rule type can be added without rework — the same daily loop serves deadline monitoring.

6. PARKED SUB-PROJECTS

Each is real and valuable. Each was parked deliberately to preserve focus. None should be resumed without displacing the current Big Rock.

Item	Why parked	State on resumption
Buy More Low (full implementation)	Its own large project; many dozens of hours already invested elsewhere	Thresholds and seven design decisions settled (§3.3, §5.2). Needs TSX total-return data, tranche sizing, LevPro scenarios.
SD-13 — the insurable mortgage tier	Became a focus deviation from both Market Drop Wins and Smart Debt. Also has no market-drop connection	Two decisions identified: (A) whether to buy the insured rate — answered no, a ~2.80% premium to save ~0.15pp has roughly a 20-year payback; (B) 25-yr insurable vs 30-yr uninsured — net 5-year cost of the 30-yr default ≈ $7,790 on a $600k mortgage, pure tier misclassification ≈ $5,738. Broker review one-pager drafted. Belongs to the Smart Debt mortgage suite, not here.
LEV-11 corporate equity meltdown	Niche; multiple cash sources exist to fund any leveraging	Mechanism sound, unmodelled. Premium tier PB-2 anchor.
LEAPS / options routes	Complex, ongoing monitoring, small market, not needed	Library candidates.
Post-drop convexity (AC-1)	Tested and rejected — see §3.7	Retained as a documented negative result. Do not revisit without real option surfaces and a larger sample.
The Can’t Lose supplement (non-debt)	Outgrew the project; off-purpose	Frozen. Lead magnet or mid-tier only, drawing on Financial Freedom Without Sacrifice material and the b-Book format. No further development.
Mortgage Decision Suite / broker channel	Different channel, buyer, sales cycle and compliance environment	10 strategies grouped. Let broker friends react; treat as a later decision.

7. ERRORS MADE AND CORRECTED

Preserved deliberately. A methodology that catches its own errors is more trustworthy than one reporting only successes — and being able to show a client you tested your own most appealing idea and rejected it is worth more than a backtest that worked.

The interest-only vs amortized loan model was fundamentally invalid. It compared equal principal instead of equal after-tax cash flow. The primary controllable input to a leveraged strategy is the cash flow going into it. At 5% and a 38% marginal rate, interest-only costs ~3.1 cents of after-tax cash flow per dollar borrowed versus ~12 cents for a 10-year amortization, so equal cash flow supports a loan several times larger. Apples-to-apples is not really possible; the question becomes “what leverage amount is responsible,” which is a LevPro question. Result discarded.
“Many users” was the wrong bar for building tools. The criterion is analytical rigour, not user count. Build the analysis with assumptions exposed; skip the polish.
The TFSA funding constraint was over-weighted. With ample cash in three pockets, wrapper location is not a deployment bottleneck.
LEV-7 was initially downgraded on opportunity-cost grounds — applying a terminal-wealth objective to someone who does not share it.
Preparation steps were miscategorized as strategies. HELOC review, dry powder split and target allocation are prerequisites.
The post-drop convexity thesis was mine, and it was wrong. Tested and rejected.
DS-2 Classic Debt Swap was understated — the before-tax rate arbitrage was missed, as was the argument that settles the strategy’s main objection: proceeds could be re-invested in cash-like holdings leaving allocation literally unchanged, and the rate and deductibility benefits still captured in full. DS-2 is not leverage wearing a tax story.
G39 in-kind RRIF withdrawal was oversold as a tax strategy. No withholding applies on the minimum, so it can be satisfied entirely in kind — but the tax outcome is identical to selling and repurchasing. Real benefits are avoided trading costs and spread, no time out of market, and removal of a repurchase decision a frightened retiree often never makes.
SD-13 was allowed to consume disproportionate effort on a strategy with no market-drop connection. The purpose filter existed and was not applied.

8. REFERENCE NUMBERS

Declines (S&P 500, closing, price index, 1929–2026): −20% 1/6.5 yrs · −25% 1/8.1 · −30% 1/12.1 · −35% 1/16.2 · −40% 1/19.4 · −50% 1/32.3. Post-WWII −30% 1/13.3. Median depth −31.1% full, −28.4% post-war. P(−30% | −20%) = 53% full, 46% post-war. P(no −20% in 10 yrs) ≈ 21%.

Valuation: Shiller CAPE ~40.6 September 2026, 98.8th percentile since 1881.

AQR buy-the-dip: 196 implementations, 60+ years, underperformed in >60%; ~0.5%/yr alpha indistinguishable from chance; 18.7% lower ending wealth vs DCA.

Cederburg et al.: one-third domestic / two-thirds international, no fixed income, every age. TDF investors need 63% more savings. 10% rate matches a balanced investor’s 19.3%.

Fractional Kelly: 0.3x cuts an 80%-drawdown chance from ~1-in-5 to ~1-in-213, retaining ~51% of growth. Above 2x Kelly is worse than not betting.

Man Group trend + tail hybrid: 25% tail-option allocation alongside trend cuts average monthly returns from ~0.42% to ~0.32% while nearly replicating put convexity.

Universa: reported 3,612% March 2020 / 4,144% Q1 2020 — on the put allocation, not the total portfolio.

Canadian tax: capital loss carryback three years, carryforward indefinite. Superficial loss 61-day window, 30 days either side of settlement. Prescribed rate at the 1% floor July 2020–June 2022; 3% since Q3 2025. Bank of Canada 3.00% Jan 2025 → 2.25% late 2025, held through 2026.

ESPP: a 15% discount is ~17.6% return on purchase price before any market movement. Reset provisions can re-baseline for up to 27 months.

RDSP: grant matches 300%/200%/100% by family net income; first $500 can attract $1,500. Bond up to $1,000/yr with no contribution. Carry-forward 10 years; annual maxima $10,500 grant / $11,000 bond; lifetime $70,000 / $20,000; claimable to age 49.

Mortgages: IRD reset via blend-and-extend — reported break-fee reductions up to ~90%, individual cases only. Renewal: ~70% of Canadians sign the bank’s first offer; banks price 0.20–0.75% above market; shoppers save 30–60bps. Straight switches exempt from the OSFI stress test since November 2024. Tier pricing: roughly 4.05% insured / 4.20% insurable / 4.40% uninsured, one lender one week.

Options: negative index skew permanent since October 1987. SPX prices in dividends; SPY does not.

Yields: TSX ~3%; S&P 500 ~1.5%.