Why Leverage Is Not Used by Big Investors

By Mohit Sharma · 13 min read · Mar 2026

The world’s greatest investors – Buffett, Munger, sovereign wealth funds, pension giants – overwhelmingly avoid leverage, and the reasons are mathematical, behavioral, and institutional. The core insight is deceptively simple: leverage increases arithmetic returns linearly but increases variance (and therefore ruin probability) quadratically. Past a certain point, adding leverage actually reduces your compound growth rate even while increasing your expected return – a paradox that most retail traders never grasp.

Warren Buffett has called leverage the primary way smart people go broke. Charlie Munger lists it alongside liquor and drugs as the three paths to ruin. Yet Berkshire Hathaway itself operates at ~1.6x leverage – through insurance float, which is the critical distinction. Float cannot be called, has zero or negative cost, and grows during crises rather than contracting. This is leverage without the kill switch. Traditional margin debt, by contrast, can be called at the worst possible moment, forcing permanent capital destruction during temporary drawdowns.

The historical evidence is overwhelming. LTCM (Nobel laureates, 25:1 leverage, $4.6B lost in 4 months), Archegos ($20B evaporated in 48 hours), Lehman Brothers (30:1, triggered the 2008 global crisis), Jesse Livermore (the greatest speculator ever, died broke) – every case follows the same script: early success breeds overconfidence, leverage increases during good times, and the “impossible” event destroys everything. The mathematical framework of ergodicity economics (Ole Peters, Nature Physics 2019) proves why: in a multiplicative process like investing, the time-average growth rate is always lower than the ensemble average, and leverage widens this gap until your wealth converges to zero with probability 1.

What Buffett and Munger Actually Say About Leverage

Buffett’s definitive statement (1989 Annual Letter): > “A small chance of distress or disgrace cannot, in our view, be offset by a large chance of extra returns.”

Munger’s most famous quote: > “There are three ways to go broke: liquor, ladies, and leverage.”

The key quotes across Buffett’s letters:

The Berkshire Paradox – Buffett DOES Use Leverage

The “Buffett’s Alpha” paper (NBER, Frazzini, Kabiller and Pedersen) proved Berkshire operates at approximately 1.6:1 leverage through insurance float. The paper concluded: “Buffett’s returns appear to be neither luck nor magic, but, rather, reward for the use of leverage combined with a focus on cheap, safe, quality stocks.”

Why float is superior to margin debt:

Berkshire’s float grew from nothing in 1967 to over $160 billion – permanent, free, non-callable capital that functions as equity but appears as a liability on the balance sheet.

Why Institutions Avoid Leverage

Around 80% of sovereign wealth funds use zero leverage. The Norway Government Pension Fund (over $1.7 trillion) caps leverage at 5% of NAV. CalPERS ($495B) only approved 5% strategic leverage in 2022, and it was considered a landmark, controversial decision.

Six reinforcing reasons:

1. Fiduciary duty asymmetry: Nobody sues you for being too conservative. Leverage losses invite immediate litigation. The Prudent Investor Rule requires demonstrating that overall portfolio risk is appropriate – leverage makes this harder.

2. Regulatory constraints: ERISA (US pensions), Ministry of Finance mandates (Norway), Income Tax Act (Canada) all restrict or cap leverage. Post-2022 UK LDI crisis, regulators imposed higher collateral buffers.

3. Liquidity mismatch: Institutions hold illiquid assets (PE, real estate, infrastructure). Leverage requires meeting margin calls with cash. Harvard’s 2008 crisis: $11B in unfunded PE commitments + leveraged derivatives forced them to issue $2.5B in bonds and sell PE at fire-sale prices.

4. Agency problems: Unlike hedge fund managers who get 20% of upside, institutional CIOs have no upside from leverage but face termination and lawsuits from downside. The incentive structure rationally produces conservative behavior.

5. Systemic risk: Institutions are so large that their forced liquidations can destabilize markets. LTCM threatened the US bond market. UK pension LDI leverage nearly collapsed the gilt market (Bank of England emergency intervention required).

6. Canadian exception proves the rule: Canadian pensions (CPPIB, OMERS, OTPP) borrow 15-20% of assets – but they operate under special enabling legislation, have longer-duration liabilities, and the Bank of Canada has flagged this as a potential vulnerability.

The hedge fund distinction:

The Mathematical Case Against Leverage

Volatility Drag (Variance Drain)

The fundamental formula:

Geometric Return = Arithmetic Return - (Variance / 2)

With leverage factor L: - Arithmetic return scales linearly: L x mu - Volatility drag scales quadratically: (L^2 x sigma^2) / 2

Real-world proof – leveraged ETFs: - TQQQ (3x Nasdaq) 5-year CAGR: ~22.9%. QQQ CAGR: ~16.1%. If 3x worked: should be ~48%. Actual multiplier: ~1.4x, not 3x. Max drawdown: 80%+ vs QQQ’s ~35%. - COVID crash: S&P 500 dropped 33.9%. UPRO (3x S&P) dropped 75.8%. - Morningstar study (2009-2018): average 2x leveraged ETF returned -11.1% annually while underlying indexes returned +15.7%.

Kelly Criterion

The Kelly formula for optimal leverage:

f* = mu / sigma^2

Where mu = excess return, sigma^2 = variance.

Critical properties: - At exactly Kelly: maximum long-term geometric growth - At 2x Kelly: expected geometric growth rate drops to zero - Above 2x Kelly: wealth converges to zero with probability 1

Most practitioners use 1/4 to 1/2 Kelly because parameter estimation errors compound with leverage. Ed Thorp (who pioneered Kelly in markets) used fractional Kelly throughout his career.

The Asymmetry of Losses

With 2x leverage, a 25% market drop = 50% portfolio loss, requiring 100% to recover. A 50% market drop = 100% loss – total wipeout, no recovery possible.

Ergodicity (Ole Peters, Nature Physics 2019)

This is the deepest argument. Peters proved that investing is a non-ergodic process – the time-average growth rate (what happens to you) differs from the ensemble average (what happens on average across parallel universes).

The coin toss example: - Heads: wealth increases 50%. Tails: wealth decreases 40%. - Expected value per flip: +5% (take this bet!) - Time-average growth rate: -5.1% per flip (you will go broke!) - After 100 flips: mean wealth = $16,000 (driven by rare extreme winners). Median wealth = $0.51. 86% of individuals lost money.

The expected value says “take this bet.” The time-average growth rate says “this bet will ruin you.” You cannot exchange wealth with your parallel-universe selves. You live on one path.

Path Dependency

Leveraged returns depend on the order of returns, not just the average (Avellaneda and Zhang, SIAM 2010):

Path: +10%, then -10%
Unleveraged: $100 -> $110 -> $99 (net: -1%)
2x leveraged: $100 -> $120 -> $96 (net: -4%)
3x leveraged: $100 -> $130 -> $91 (net: -9%)

Both paths have 0% arithmetic average, but leverage creates losses scaling with L^2.

The Graveyard of Leveraged Investors

The universal pattern across every case: 1. Early success breeds overconfidence 2. Leverage increases during good times 3. The “impossible” event hits (Russian default, subprime, single-day crash) 4. Leverage turns temporary drawdown into permanent destruction 5. Liquidity vanishes precisely when needed 6. Counterparty chains create contagion

LTCM in detail: Two Nobel laureates (Scholes, Merton), the most sophisticated models in finance, 40%+ annual returns for 3 years. Then $4.6B lost in 4 months. At peak, $4.7B in equity supporting $124.5B in assets and $1.25 trillion in derivatives. A 3.6% decline in assets = 100% equity destruction.

Archegos: Bill Hwang used total return swaps across multiple prime brokers, none of whom saw the full picture. Grew from $1.5B to $36B equity controlling $160B exposure in one year. When ViacomCBS did a secondary offering, the entire structure collapsed in 48 hours. $20B personal wealth to zero.

Jesse Livermore: The most skilled speculator in history. Made $100M (~$1.5B today) shorting the 1929 crash. Lost it all through leveraged trading by 1934. Filed for bankruptcy three times. Shot himself in a hotel cloakroom in 1940. His suicide note: “My life has been a failure.”

What the 2008 Crisis Proved About Systemic Leverage

In April 2004, the SEC voted to exempt the five largest investment banks from the net capital rule, allowing them to use internal risk models instead of the 12-15:1 leverage cap.

Result: - Lehman Brothers: 30-44:1 leverage. At 30:1, a 3.3% decline in assets wipes out ALL equity. - Bear Stearns: ~33:1 - Merrill Lynch: up to 40:1

They loaded up on mortgage-backed securities. When housing fell 27%, these banks were not just insolvent – they were catastrophically insolvent.

Consequences: - Bear Stearns: sold for $2/share (had traded at $170) - Lehman Brothers: largest bankruptcy in US history ($639B) - AIG: $182B government bailout - US households: lost ~$16 trillion in net worth - Global GDP loss: ~$2 trillion in 2009

Buffett’s Superior Alternative: Getting Leverage Without the Kill Switch

The “Buffett’s Alpha” paper proved Berkshire operates at 1.6:1 leverage – but through insurance float, not margin debt. This distinction is everything:

Float characteristics (from Buffett’s letters): > “Float is money we hold but don’t own. In an insurance operation, float arises because premiums are received before losses are paid.” (1997)

“Since our float has cost us virtually nothing over the years, it has in effect served as equity.” (1995)

The lesson for non-Berkshire investors: The only safe forms of leverage are those that cannot be called during a crisis. This includes: - Insurance float (Berkshire model) - Long-duration, fixed-rate debt at low rates - Embedded leverage through PE allocations (not ideal – Harvard 2008 showed the risks) - Operating leverage (fixed costs in a growing business)

Margin debt, repo, total return swaps, and any callable leverage are fundamentally incompatible with long-term wealth creation because they introduce a kill switch that activates at the worst possible moment.

Nuances and Contradictions

Leverage IS Used by Some Great Investors – With Critical Constraints

• Berkshire Hathaway: 1.6:1 through float (non-callable, free leverage)
• Renaissance Technologies: Medallion fund uses significant leverage but with the most sophisticated risk management in history, extremely short holding periods, and strict position limits
• Canadian pensions: 15-20% leverage under special legislation, with very long-duration liabilities
• Real estate: Most real estate investing uses leverage (mortgages), but this is fixed-rate, long-duration, non-callable debt against a tangible asset

The pattern: leverage works when it is non-callable, long-duration, low-cost, and applied to assets you deeply understand. It destroys when it is callable, short-duration, and applied to volatile or illiquid assets.

Kelly Criterion Suggests SOME Leverage Is Optimal

The math suggests ~1.5-2.2x leverage is optimal for US equities (depending on the time period). But: - Including the Great Depression reduces optimal leverage to ~1.47x - Fat tails systematically reduce optimal leverage vs. normal distribution models - Parameter estimation errors mean you never know your true edge precisely - Practitioners recommend 1/4 to 1/2 Kelly, which for equities = approximately 1x (no leverage)

The Paradox of “Safe” Leveraged Assets

The most dangerous leverage is on assets that appear safe: - LTCM: Treasury arbitrage (spreads “always” converge) - Bear Stearns funds: AAA-rated CDOs - UK pension LDI: Government bonds - Orange County: Government securities and repos

When everyone agrees something is safe, leverage concentrates on it. When the “safe” thing cracks, the leverage unwinds catastrophically. Safety enables the very leverage that makes it unsafe.

India’s F&O Market: The Largest Retail Leverage Experiment in History

• 93% of individual F&O traders lost money (SEBI, FY22-24)
• Rs 1.8 lakh crore in aggregate losses
• 97% of FPI profits from algorithmic trading
• This is a wealth transfer from leveraged retail to unleveraged (or better-leveraged) algos
• Despite consecutive years of losses, 75%+ of losing traders continue

Practical Rules

1. Never use callable leverage for long-term investing. Margin debt, total return swaps, and repo are incompatible with wealth preservation.

2. If you must use leverage, use 1/4 Kelly at most. For equities, this means approximately 1x – which is no leverage at all.

3. The 1% rule protects against leverage addiction. Never risk more than 1% of capital on any single leveraged position.

4. Get leverage through operations, not debt. A business with high operating leverage (fixed costs, scalable revenue) gives you leverage without the kill switch.

5. Float over Margin. If you can access insurance float, rental income, or other non-callable capital, this is the Buffett path. If you can’t, accept unleveraged returns and compound patiently.

6. The boring truth: Getting rich slowly without leverage is the schlep that nobody wants to do. It doesn’t make for dramatic stories. It works.

Code and media leverage scale without blowup risk. Capital leverage has an extinction event built into the tail. The specific knowledge worth having: how to compound wealth without leverage that can kill you.

This article is for general education only. It is not investment advice or a recommendation to buy or sell any security or product. Investments are subject to market risk; returns are not guaranteed.