The first time I saw a backtest where a 50/50 mix of a coin and cash finished ahead of the coin itself, I assumed the code was broken. The asset ended the window almost exactly where it started. The cash earned nothing. The blend was up double digits. I spent an evening hunting for the bug and found Claude Shannon instead, who described this exact effect in lectures decades ago. People call it Shannon's Demon, and it is the cleanest demonstration I know that volatility on its own can be harvested for return, on paper anyway.
The toy version goes like this. Take an asset that either doubles or halves each period with equal probability. Buy and hold it and your median outcome is flat forever, because over time every double gets paired with a halving that cancels it. The arithmetic average return per flip is a juicy 25 percent, but the compounded, geometric return is exactly zero, and geometric is the one your account balance actually experiences. Now hold half your money in the asset and half in cash, and rebalance back to 50/50 after every move. On an up flip the portfolio gains 50 percent. On a down flip it loses 25 percent. One up and one down multiply your wealth by 1.5 times 0.75, which is 1.125, and that compounds out to roughly 6 percent of growth per flip, extracted from an asset that goes nowhere and a cash pile that earns nothing.
Nothing shady is happening under the hood. Rebalancing forces you to sell a slice after every double and buy a slice after every halving. It is buy low, sell high, executed mechanically, with the 50/50 target making every decision for you. Shannon reportedly never traded the idea himself, partly because commissions in his day would have eaten the whole edge. Keep that detail in mind, because it turns out to be the most prophetic part of the story.
Where the extra return actually comes from
The general version runs on the gap between arithmetic and geometric returns. A volatile asset compounds more slowly than its average return suggests, and the drag is roughly half its variance. Something with 80 percent annualized volatility, which is an ordinary number in crypto, carries a drag of roughly 32 percent a year. An asset like that can have a solidly positive average return and still grind a buy-and-hold position down over time.
Mixing with cash changes the geometry in your favor. Hold a fraction of your portfolio in the volatile asset and your expected return scales with that fraction, but your variance scales with its square. Cut exposure in half and you keep half the upside while taking only a quarter of the volatility drag. Rebalancing is the maintenance work that keeps this ratio locked in. Skip it and the weights drift, and the advantage quietly leaks away.
For a 50/50 blend of one volatile asset and cash, rebalanced consistently, the bonus works out to roughly the asset's variance divided by eight. At 80 percent annualized vol that is about 8 percent a year. At 50 percent vol it is closer to 3. Those figures describe the growth rate of the blend versus the average growth rate of its two legs, under idealized assumptions, and every word of that qualifier is load-bearing.
One more subtlety before the bad news. Rebalancing improves your median outcome while giving up a little expected value, because the expectation of buy-and-hold is carried by a small number of absurdly lucky paths. You are trading lottery tail for a better typical result. Whether that trade appeals to you depends on how many independent lives you plan to live, and my count is one.
The three things that eat it
Trend is the big one. The math assumes each move is independent of the last, and markets do not always cooperate. When an asset trends, rebalancing becomes a machine for repeatedly doing the wrong thing. In a sustained bull run you sell a slice of the winner at every step up, so you finish holding a fraction of what you started with while buy-and-hold laps you. In a sustained collapse you buy the loser all the way down. Run this against a token that eventually goes to zero and the cash leg gets fed into the fire one rebalance at a time, and you lose far more than half your stack. Crypto produces both regimes, in size, sometimes in the same year, so the clean coin-flip assumption is doing a lot of work.
Costs are the second. Every rebalance pays an exchange fee, crosses a spread, and takes some slippage. The theoretical bonus at crypto volatility is mid single digits per year, which sounds like decent room until you notice that frequent calendar rebalancing generates enormous turnover for tiny incremental benefit. Most of the harvest comes from the large swings, and you can capture those with far fewer trades than a daily schedule implies.
Taxes are the third, and often the fatal one. In many jurisdictions every rebalance out of the appreciated leg is a taxable disposal, typically at short-term rates given how often you are trading. A strategy whose entire edge is a few percent a year has no room for that. If you cannot run it inside a tax-sheltered wrapper or a jurisdiction that treats frequent crypto trades kindly, do the after-tax arithmetic before you admire the backtest, because the backtest does not pay taxes and you do.
How I would size the expectation
If you want a realistic number rather than a story, walk through something like this before committing money.
- Square the asset's annualized volatility and divide by eight. For a 50/50 blend against cash, treat that as the ceiling on the bonus, and expect realized results to come in below it.
- Rebalance on bands rather than a calendar. Trading only when the volatile leg drifts something like 5 to 10 percentage points from target keeps most of the harvest while cutting turnover dramatically.
- Estimate your all-in cost per rebalance, meaning fee plus spread plus slippage, guess how many times a year your bands will realistically trigger, and subtract the product from the ceiling.
- Apply your actual tax rate to the gains the strategy realizes. In a taxable account this step deletes the bonus for a lot of people, and it is better to learn that on paper than in an account statement.
- Only pair legs you would be comfortable holding on their own. The demon adds a modest return to a portfolio you already believe in, and it cannot rescue an asset on its way to zero.
After all the haircuts, my honest estimate for a disciplined band rebalance between a major coin and cash or stables is a bonus in the low single digits per year, with some years negative because the market trended instead of chopping. That sounds underwhelming, and mostly it is. But the rebalance was worth doing anyway, because it keeps your risk roughly constant instead of letting one bull market quietly turn a half position into most of your net worth. Shannon's Demon just means the discipline comes with a small rebate attached, and I will happily take a rebate on something I was going to do regardless.