Published on: 2026-09-11
A market can swing violently and still finish exactly where it began. Shannon’s Demon shows how systematic rebalancing through those movements can leave a portfolio with more money than the same starting allocation left untouched. The math works, although real markets impose much tighter limits on the result.
Shannon’s Demon shows how fixed-weight rebalancing can alter compound wealth even when an asset eventually returns to its starting price.
The effect comes from changing the amount of capital exposed before subsequent price moves, rather than volatility generating profit on its own.
Real-world results depend on return patterns, correlation, rebalancing frequency and trading costs.
The thought experiment demonstrates a mathematical possibility, not a guaranteed source of excess return.
Shannon’s Demon is a portfolio-rebalancing thought experiment associated with Claude Shannon, the mathematician whose work laid the foundations of information theory.
The classic setup divides capital equally between a volatile asset and cash. When price movements push the portfolio away from its original 50/50 allocation, capital is shifted between the two holdings to restore the target weights.
A rise in the volatile asset therefore leads to selling some of it. A decline leads to allocating additional capital to it. The rebalancing rule itself requires no directional forecast. Its profitability still depends on what prices do after each adjustment.
The counterintuitive result appears when repeated price reversals change portfolio weight enough that rebalancing alters the compound return.
Consider a portfolio starting with $10,000. Invest half in a volatile asset and keep the other half in cash. Cash earns zero interest, and we ignore trading costs.
The volatile asset falls by 50%, reducing its value from $5,000 to $2,500. Cash remains at $5,000, leaving total wealth of $7,500. The portfolio is then restored to equal weights, placing $3,750 in the asset and $3,750 in cash.
The asset subsequently doubles, returning exactly to its original price. The $3,750 position becomes $7,500, while cash remains unchanged. Total wealth reaches $11,250.
Stage |
Initial 50/50, No Rebalancing |
50/50 Rebalancing |
Starting wealth |
$10,000 |
$10,000 |
After 50% asset decline |
$7,500 |
$7,500 |
After full price recovery |
$10,000 |
$11,250 |
Both portfolios began with the same allocation, and the volatile asset ended at its starting price. Rebalancing left the second portfolio 12.5% higher.
Percentage returns compound multiplicatively. An asset that falls 50% needs a 100% gain to recover. Rebalancing changes how much money is exposed before that recovery occurs. The unchanged portfolio has $2,500 invested when the asset doubles.
The rebalanced portfolio has $3,750 exposed to the same move. The ending asset price can therefore be identical while ending portfolio wealth differs. A flat chart does not necessarily produce a flat wealth path.
The 12.5% gain is deliberately artificial. It comes from an extreme two-move sequence, cost-free trading and a precisely timed rebalance. It demonstrates the mechanism rather than the return you should expect in real markets.
Research produces a far more restrained picture than Shannon’s stylised example.
Evidence |
What It Suggests |
2025 Quantitative Finance study |
Rebalancing premiums were typically below 50 basis points annually under realistic model assumptions |
2026 empirical research |
Rebalancing frequency and cross-asset correlation materially affected results |
Broader rebalancing research |
Mean reversion can favour fixed weights, while momentum can favour allowing winners to run |
The 2025 study used analytical models and Monte Carlo simulations and found that serial correlation in asset returns also influenced the size of the rebalancing premium.
A July 2026 Finance Research Letters study covering more than 24,000 two-fund portfolios likewise found that shorter rebalancing cycles and lower cross-asset correlation were associated with stronger excess rebalancing returns in its sample.
These findings do not make the 12.5% example wrong. The figures measure different things. Shannon’s result comes from one extreme round trip, while empirical and simulated studies examine repeated rebalancing over much longer periods.
Professional portfolio practice also treats rebalancing primarily as a way to maintain intended asset weights and risk exposures. CFA Institute’s 2026 curriculum recognises calendar-based and range-based approaches, with transaction costs, volatility, correlation, momentum, taxation and liquidity influencing how those rules are set.
The evidence supports a conditional conclusion. Rebalancing can affect compound returns, although no fixed premium appears simply because prices are volatile.
These approaches can generate similar-looking trades while following different rules.
Approach |
Trigger |
Main Premise |
Shannon-style rebalancing |
Allocation drift |
Restore target weights |
Price or spread deviation |
Expect reversal |
|
Predetermined price levels |
Capture repeated range movement |
Mean-reversion strategies usually depend on an expectation that a price, spread or valuation relationship will move back toward a reference level. Grid strategies place trades around predetermined price levels and generally benefit from repeated movement through those levels.
Fixed-weight rebalancing acts when portfolio weights move away from target. The latest market move does not have to reverse immediately for the rule to trigger.
Persistent trends expose the difference. Some grid structures can accumulate exposure against a continuing move, while fixed-weight rebalancing may repeatedly trim a winner or add to a falling asset.
Several conditions can weaken or eliminate the apparent rebalancing advantage.
Persistent momentum. Rebalancing repeatedly trims an asset that continues to outperform. Positive serial correlation can therefore favour allowing portfolio weights to drift rather than immediately restoring them.
Permanent deterioration. A lower price does not guarantee eventual recovery. Continually restoring exposure to an impaired asset can compound losses rather than exploit temporary volatility.
Higher co-movement. As holdings follow increasingly similar return paths, the relative price dispersion available for rebalancing generally narrows.
Trading friction. Spreads, commissions, slippage, financing costs, taxes and market impact can consume an advantage that may already be measured in basis points.
Leverage and illiquidity. A position can be liquidated before the strategy's required return path has time to unfold, while thin markets may prevent rebalancing at the assumed price.
Long-run mathematics provide little protection against forced liquidation. These limits are why volatility itself should never be confused with an exploitable edge. Repeated reversals and a persistent one-way move can generate similar measures of volatility while producing very different outcomes for a fixed-weight strategy.
Shannon’s Demon separates two decisions that are often treated as one.
Correctly anticipating where a market eventually finishes does not determine how much capital survives the route. Position size and changes in exposure can materially alter the outcome before the final price is reached.
Strong recent performance can encourage confidence that a winner will continue rising, while sharp weakness can make additional exposure harder to justify emotionally. A predetermined rebalancing rule reduces the latest price narrative's influence on allocation decisions.
A strategy can have favourable long-term mathematics and still become unusable if leverage, drawdowns or liquidity force the position to close first.
Shannon’s Demon therefore offers a broader lesson than “buy low and sell high.” Portfolio mechanics can influence results independently of a directional market forecast.
No. Negative correlation is not required. Lower correlation can create greater relative movement between holdings, although the result also depends on volatility, expected returns and the sequence of those returns.
Yes. The assets do not need to finish unchanged. The round-trip example simply isolates the mechanism. Rebalancing can change portfolio outcomes across many different return paths.
Not necessarily. A return difference from maintaining fixed portfolio weights can come from systematic portfolio construction rather than forecasting skill or security selection.
No. Returning to the starting price creates a clear demonstration because the directional return disappears from the comparison. Rebalancing can also matter when assets finish above or below where they began.
Under Shannon’s constructed conditions, the market can go nowhere and still leave more money behind. In real markets, the answer depends on the return path, portfolio structure and implementation. The enduring value of Shannon’s Demon is not a promise of free profit from volatility. It shows that market direction and portfolio return are separate problems. Getting the destination right does not settle what happens to capital along the way.
