Portfolio Construction & Risk Management

Gap identified 2026-04-24: The wiki contains many strong investment theses but no framework for translating them into actual portfolio decisions. This page closes that gap. See [[financial-markets]] for individual instruments and [[behavioral-finance]] for the psychology layer.

---

The Gap in Plain Terms

Hakyun can construct an investment thesis well — the cybersecurity paper is evidence. But a thesis answers "what to buy." A portfolio framework answers:

• How much to put in each position
• How much loss is acceptable before exiting
• Whether two positions are actually the same bet in different clothes
• When the thesis is broken, not just temporarily wrong

Without this layer, good analysis produces bad outcomes. A correct thesis entered at 20% of portfolio with no exit rule is still a coin flip on whether it adds value.

---

The Risk Hierarchy (Howard Marks)

Before sizing, understand the three types of risk that matter:

Risk Type	Definition	Common mistake
Permanent loss of capital	Money that doesn't come back	Treating volatility as risk
Volatility	Price moving around	Treating it as permanent loss
Shortfall risk	Not achieving what you need	Ignoring opportunity cost

"Risk is not volatility. Risk is the probability of permanent loss." — Marks

For Hakyun specifically: The most dangerous risk is permanent loss on leveraged positions (options, leveraged ETFs) where time decay or gap-down events eliminate capital. The second most dangerous is shortfall risk — having money in low-returning assets while high-conviction theses play out without you.

---

The Barbell Approach (Taleb / Dalio)

Referenced in the Synapse market wraps but never explained. Two versions:

Taleb's Barbell

Split capital between:

• Safe (85–90%): Treasury bills, cash equivalents, instruments with near-zero downside
• Speculative (10–15%): High-upside asymmetric bets with capped downside (long options, early-stage positions)

Nothing in the middle. The middle — moderate-risk, moderate-return assets — has the worst risk-adjusted return because it provides false security while exposing to permanent loss.

Why it works: The safe end funds the speculative end. Maximum loss is ~15%. Maximum gain is uncapped (options can return 10–100x). Convexity is on your side.

Application to Hakyun's theses:

Allocation	Instruments
Safe (85%)	Singapore T-bills (currently 3–4% annualized), CPF-SA (2.5–5%), SGS bonds, cash
Speculative (15%)	Long-dated calls on CRWD/ZS thesis, URA (uranium ETF), energy sector plays, direct cybersecurity positions

Dalio's All Weather

Designed for all four macro environments:

Environment	Performing assets
Rising growth	Equities, corporate bonds, commodities
Falling growth	Treasury bonds, gold
Rising inflation	Commodities, gold, TIPS, inflation-linked bonds
Falling inflation	Equities, long-duration bonds

All Weather allocation (Dalio's classic):

• 30% Equities
• 40% Long-term bonds
• 15% Intermediate bonds
• 7.5% Gold
• 7.5% Commodities

Warning: This was designed for institutional investors. The bond weighting is dangerous in a rising-rate environment (which Warsh's tenure may produce). Also designed before crypto existed as an asset class.

The Q1 2026 insight: March 2026 proved that the bond allocation fails in an inflationary war shock — bonds and stocks fell simultaneously. Only energy and commodities provided protection. This is the stagflation scenario that the All Weather portfolio historically underperforms.

---

Position Sizing: The Kelly Criterion

Kelly calculates the theoretically optimal bet size given your edge.

Full Kelly formula:

f* = (bp - q) / b

f* = fraction of capital to bet
b  = net odds received (if you bet $1 and win, you get $b back)
p  = probability of winning
q  = probability of losing (1 - p)

Example: If you believe CRWD has 60% probability of 50% upside and 40% probability of 30% downside:

• b = 1.5 (win $1.50 per $1 at risk)
• p = 0.60
• q = 0.40
• f* = (1.5 × 0.60 - 0.40) / 1.5 = (0.90 - 0.40) / 1.5 = 33%

But: Full Kelly is almost always too aggressive for real portfolios. The standard practice is Half Kelly (16.5% in this example) or Quarter Kelly (8.25%).

Why Half Kelly:

• Full Kelly maximizes long-run geometric growth but produces violent drawdowns
• Half Kelly captures ~75% of the long-run growth with ~50% of the drawdown
• Quarter Kelly is appropriate when probability estimates are uncertain (which they always are)

Practical rule: For high-conviction, well-researched theses with asymmetric upside (like ZS with 36% YTD decline + fundamental strength), use Half Kelly. For speculative positions (uranium, hydrogen), use Quarter Kelly or less.

---

Correlation: Are You Actually Diversified?

The single most common error in portfolio construction: owning five "different" things that are all the same bet.

Hakyun's current thesis portfolio (implicit):

Thesis	Real underlying bet
Cybersecurity (CRWD, PANW, ZS)	AI transition + enterprise IT spending
Uranium	Clean energy demand + supply shock
Oil/energy	Geopolitical risk premium + inflation
Bitcoin	Liquidity / debasement / digital asset adoption
Gold	Real yield decline / debasement

Correlation analysis:

• CRWD/PANW/ZS: High correlation with each other (same sector, same thesis drivers). These three don't add diversification — they amplify each other. Treat as one position.
• Bitcoin + Gold (2026 context): Used to be correlated (both = debasement hedge). In Q1 2026, they decoupled — Bitcoin held while gold fell during Warsh shock. The correlation may be breaking down permanently.
• Oil + Bitcoin: Negative correlation in inflation shock — oil is the inflation driver; Bitcoin fell on oil spike initially, then recovered. They're different bets.
• Uranium + Oil: Both energy, but different thesis — uranium = clean energy demand; oil = geopolitical risk. Partial correlation but not high.

Rule of thumb: If two positions would both get destroyed by the same event (e.g., a severe global recession destroys both equity and oil demand), they're correlated. Don't size both at full weight.

---

Asymmetric Payoff: Why Options Fit the Barbell

The wiki covers options mechanics but not their portfolio function. For thesis-based investing:

Long calls on high-conviction names provide:

• Capped downside: Premium paid (e.g., 2–5% of portfolio)
• Uncapped upside: If thesis plays out faster/harder than expected
• Time as risk, not capital as risk

The ZS case study (April 2026):

• Stock at -36% YTD
• Fundamentals: AI Security ARR hit FY26 target 3 quarters early; ARR +25% YoY
• A long-dated call (12–18 months out) on a narrative reversal:
  • If thesis correct in 12 months: 200–400% on the option
  • If thesis wrong: lose the premium (2–3% of portfolio)
  • Max loss: small. Max gain: 20–40× the cost.

This is the barbell applied to a specific opportunity.

---

Exit Rules: When Is the Thesis Broken?

The most important discipline in the wiki. For each thesis:

Cybersecurity Thesis Exit Criteria

The thesis breaks (not just pauses) if:

1. ARR growth decelerates below 15% for two consecutive quarters at CRWD/ZS — this would signal the AI disruption narrative is actually playing out in fundamentals, not just market sentiment
2. Microsoft achieves genuine share in a security segment through bundling that results in negative net new ARR at a pure-play (not just lower growth, but actual customer losses to MSFT)
3. A genuine frontier offensive AI model deploys capabilities that defenders cannot match — i.e., the "Mythos" scenario actually materializes with a confirmed large-scale breach of a major incumbent's own platform
4. Multiple compression continues to 15x revenue or below on ZS/CRWD — this would mean the market is pricing in secular decline, not just sentiment correction

Uranium Thesis Exit Criteria

1. New mines brought online significantly faster than expected — typically requires 5–10 years, but fast-track permitting could change this
2. Nuclear reactor cancellations accelerate — if countries reverse nuclear commitments (post-Fukushima 2 scenario)
3. Kazatomprom reverses production cuts or geopolitical stabilization restores Niger production

General Exit Discipline

The pre-mortem rule: Before entering any position, write one paragraph: "It is 18 months later and this position has lost 50%. What happened?" This forces you to articulate the bear case before you're emotionally invested.

The thesis journal: Every position should have:

Entry date:
Entry thesis (2 sentences max):
Price targets:
Specific events that would prove me wrong:
Stop-loss level and what it represents:
Re-evaluation date:

---

Leverage and Time Decay: The Lethal Combination

The wiki notes leveraged ETFs (e.g., 2× or 3× ETFs on semiconductors, energy) as "precision instruments, not long-hold vehicles." The mechanism of why:

Volatility decay (beta slippage): A 2× ETF on an index that goes +10%, -10% in two days doesn't return 0%.

• Day 1: $100 × (1 + 2×0.10) = $120
• Day 2: $120 × (1 - 2×0.10) = $120 × 0.80 = $96

The unleveraged index returned 0% ($100 → $110 → $99). The 2× ETF returned -4%. Over months of choppy markets, this decay compounds into significant losses even when the underlying thesis is directionally correct.

Rule: Leveraged ETFs → maximum 1–2 week holding periods for tactical trades. Never hold during earnings, Fed meetings, or known binary events.

---

Practical Framework for Hakyun's Situation

Stage 1 (Now — NTU Student, Limited Capital)

• Priority: Human capital returns > financial capital returns. The marginal return on a new skill or internship experience at 22 far exceeds the marginal return on $5,000 invested.
• Allocation: 80% safe (T-bills, CPF, HYSA), 20% learning capital (small positions in high-conviction theses to build decision-making discipline under real conditions)
• Purpose: Not to make money. To build the decision journal, test thesis construction, feel the psychology of loss.

Stage 2 (First Job, $50k–$200k Capital)

• Implement barbell: 75% safe/income-generating + 25% thesis-driven
• CPF OA → CPFIS for REITs or ETFs; SA → let it compound at 4–5%
• First priority position: diversified low-cost global ETF (VWRA or similar)
• Second: One or two high-conviction thesis positions sized at Quarter Kelly

Stage 3 (CFA Chartered, Asset Management Role)

• Professional-grade portfolio construction with risk budgeting
• Correlation matrices, factor exposure analysis, formal drawdown limits
• More options as hedging tools, not speculative instruments

---

Key Questions to Answer Before Any Investment

1. What is my edge? Information advantage? Time horizon advantage? Structural advantage (Singapore tax treatment)? If I have no edge, I should own an index.
2. What is the downside scenario? Specifically, not abstractly.
3. How much will I regret this loss? The regret test is better than any model.
4. Is this a signal or noise? Is the price move caused by fundamentals or sentiment? (The cybersecurity selloff was sentiment. The SMCI DOJ indictment was fundamentals.)
5. What would change my mind? If I can't answer this, I'm not analyzing — I'm rationalizing.

---

Related Pages

[[financial-markets]] | [[behavioral-finance]] | [[cybersecurity-thesis]] | [[market-intelligence]] | [[energy-commodities]] | [[hakyun-ryu]]