▲ Work in progress. This site is built and researched in the open: pages, numbers and findings may change or be wrong, and the record has gaps. Not advice. More
The Lab
Ideas traders use, put to the record
← the Lab · strategies tested · write-up
#31 "Shannon's demon": does rebalancing 50/50 harvest volatility? No reliable edge
Past prices; no real orders were placed. Every figure is a share of the money, or a count.
Last checked: 15 October 2026.
Where the idea comes from
- Rebalancing between a coin and cash to profit from its swings is usually called Shannon's demon, after a 1960s talk at MIT by Claude Shannon, as told in William Poundstone's Fortune's Formula (2005). David Luenberger calls the same effect "volatility pumping", and Robert Fernholz the "excess growth rate".
- Its origin, the maths step by step and its catch are on our theory page "Shannon's demon: making money from a price that goes nowhere".
- This page tests it on crypto prices. It repeats a known result; it doesn't discover one.
The idea
Keep half your money in a coin and half in cash. When the coin rises, sell some back to 50/50; when it falls, buy some. If the price swings up and down, this sells high and buys low over and over, and the book grows even if the coin ends where it started. This is called Shannon's demon, after the mathematician Claude Shannon, who described it in a 1960s talk at MIT that he never published, or volatility harvesting. Does it work on crypto, after fees?
How we tested
- Bitcoin, Ether and Solana, October 2022 to September 2026: 48 months, checked every hour.
- Rebalancing rules:
- back to 50/50 when the coin's share drifts past 51%, 52.5%, 55% or 60% (or the same distance under);
- or simply every day, or every week.
- 0.1% fee on every trade, the exchange's standard spot fee.
- Compared with the same 50/50 bought once and never touched, month by month and over the whole 48 months, with our grid bot beside it.
- Fairness: each month was also run 20 times with its hours shuffled, and the rule for success was fixed in advance.
The maths, step by step
1. The demon's step. Start with 0.5 in the coin at 100 and 0.5 in cash.
- The price doubles to 200: the coin is worth 1.0, the cash 0.5. That is a two-thirds share, so sell 0.25 of the coin: 0.75 and 0.75.
- The price falls back to 100: the coin is worth 0.375, the cash 0.75. A one-third share, so buy 0.1875: 0.5625 and 0.5625.
- In all: 1.125, against 1.0 for the book left alone. The 12.5% came from the round trip.
2. When it backfires. If the price doubles and keeps going, the book keeps selling the coin on the way up. It ends with less than the untouched book.
3. Fees. Each rebalance pays 0.1% of what it trades. Tight bands rebalance often and pay more, but each trade is smaller.
What we found
1. Month by month, there was no harvest to speak of.
- Against the untouched 50/50, the rebalanced books did between −0.33% and +0.03% a month, all within one or two standard errors of zero.
- The sign wasn't steady between the two halves.
- No rule passed the test set in advance.
2. Over the whole four years, the coin's path decided it.
| over 48 months | rebalanced | left alone |
|---|---|---|
| Bitcoin | ×2.29 to ×2.32 | ×2.65 |
| Ether | ×1.67 to ×1.76 | ×1.51 |
| Solana | ×2.86 to ×3.10 | ×2.30 |
- Bitcoin mostly rose, so rebalancing sold it too early.
- Ether and Solana went up and came back, so rebalancing won.
- Our shuffled months keep each month's start and end. They can't tell whether that four-year shape was luck.
3. Against the grid. The 50/50 book averaged more a month than the grid (+1.5% to +3.4%, against +0.6% to +1.9%), with deeper worst months. That is mostly because it holds half its money in the coin, against the grid's 18–25%.
What it means
- A 50/50 rebalancing book is a bet that the coin goes up and down rather than one way.
- At normal fees it earns nothing reliable month to month. Over years it wins when the coin round-trips, and loses when it trends.
- Rebalancing at bands is really a grid by another name: each band edge acts as a rung. Like our grid studies, it found no edge in the method itself.
Limits
- Three coins and four years: one long path each, which no test here can check.
- Checked once an hour; each rebalance is done at the hour's close.
Update, 18 October 2026: three charts
Finding 2 as pictures: the 48 months in one, half the money in the coin, rebalanced whenever its share drifted more than 5 points, against the same half-and-half left alone. One chart per coin.
Bitcoin mostly rose, so the book that kept selling it back to half fell behind.
Ether went up and came back, and the rebalanced book ended ahead.
Solana swung hardest, and the rebalanced book gained the most from it.
Credits
- Claude Shannon's rebalancing idea, from a talk he never published. William Poundstone's Fortune's Formula (Hill and Wang, 2005) is the book that made it widely known.
- David G. Luenberger, Investment Science, Oxford University Press, 1998: the same effect, called "volatility pumping".
- E. Robert Fernholz, Stochastic Portfolio Theory, Springer, 2002: the extra growth, called the "excess growth rate".
- R. A. Fisher, The Design of Experiments, Oliver and Boyd, 1935: the permutation test behind the shuffled months.
- Our finding that the result depends on the coin's path repeats the known theory; we tested it on crypto. The bands and controls are our own design.
- Data: Binance's public 1-minute futures prices; Bybit's spot fee.
- PerpGremlin and Claude, "Shannon's demon": does rebalancing 50/50 harvest volatility?, TerminalGremlin, 2026: the tests, the code and the words.
← Pairs tradingMarket making on Bybit →
Educational research from public market data; any trading discussed is paper, never real. Nothing here is advice. The full notice.