LTCM Co-founder Victor Hagani: “Taking Risk Is Always a Negative.”
Odds on Open · Watch on YouTube · Generated with SnapSummary · 2026-07-03

Video Summary — “Your trades at LTCM had edge and still blew up” 🎧

Key Themes

Sizing vs. Selection
- Success depends on two decisions: what to invest in and how much to allocate.
- LTCM’s failure: correct selection, wrong sizing (overleveraged). ⚖️
Risk ≠ Free Value
- Expected value and risk-adjusted value differ. Maximizing expected return can imply ruinous sizing.
- Correct sizing is often easier to get right than finding the right investments.
Cost of Risk (a.k.a. internal “fee”) 💸
- Taking risk has an implicit cost that should be treated like a fee you charge yourself.
- Rule-of-thumb: cost ≈ variance (σ^2) or for Kelly-like objective cost ≈ variance/2.
- Example: 15% vol → cost ≈ 2.25%; 30% vol → cost ≈ 9% (risk-averse) or 4.5% (Kelly).
Concentration vs. Lifetime Objective 🎯
- High concentration (big upside) is the right answer to the wrong question if the goal is “become a billionaire.”
- Better objective: maximize smooth, sustainable lifetime spending and giving (risk-adjusted).
- Uncompensated idiosyncratic risk reduces lifetime sustainable spending.

Coin-flip experiment (60% heads)
- Most people (including finance professionals) mis-size bets: too small, too big, or incorrectly adjust after streaks.
- Optimal constant fraction ≈ Kelly (≈20% for this example); participants grasped it when explained.
“Crystal Ball / WSJ front page” trading game 📰
- Players see a redacted historical WSJ front page (one day ahead) and trade on market moves; reveals news interpretation + sizing issues.
- Findings:
  - ~100 business-school players: ~52% directional accuracy, many busted, aggregate ~break-even. Sizing poor and inconsistent.
  - Six veteran macro traders: ~60% hit rate, selective bets, all made money.
  - Open experiment with AIs (Claude, GPT, Gemini, Grok): off-the-shelf Claude performed best among free models; overall AIs too aggressive but sometimes match/beat human players when crystal ball info exists.

Sizing matters as much as selection — always decide how much to risk given expected returns and your risk aversion.
Treat risk as a cost/fee — subtract σ^2 (or σ^2/2 for Kelly-like objective) from expected returns to get true risk-adjusted performance.
Prefer compensated systematic risk; avoid uncompensated idiosyncratic risk.
Young investors: you can take more systematic risk because of human capital, but avoid extreme leverage because of frictions, costs, and psychology.
Non-professionals are unlikely to sustainably beat the market; professionals can.
Primary actionable advice: save consistently (target ~20% of income) — without saving, other planning is moot. 💾

Determine clients’ risk aversion via hypothetical expected equity premium (e.g., “If stocks return +4% over bonds, how much would you allocate?”) to set sizing.
Use diversified, low-cost index building blocks; dynamic asset allocation to adjust exposure by risk environment.
Offer transparent, low-fee (≈12 bps) managed solutions and tools (games, calculators) to educate clients about sizing and human-capital-aware planning.

“It wasn't the selection of the trades, it was the sizing.”
“Getting your sizing right is not a zero-sum activity.”
“Taking risk is always a negative.”
“The right answer to the wrong question” — lots of concentration is right to become a billionaire, but that’s the wrong objective for most people. 🤔

Reframe financial goals: prioritize smooth, sustainable lifetime spending over “get rich” metrics.
Compute a simple risk cost: square your portfolio volatility to approximate the annual cost you should charge yourself.
If unsure how much risk to take, answer: “If equities offered X% premium, how much would I allocate?” and scale exposures accordingly.
Save early and target ~20% of income. 💡

— End of summary —