July 1, 2026 – July 31, 2026
July 2026 rebalancing bot results: -13.74% vs BTC +7.71%
| Month | Max drawdown (month) | Cumulative at month end | |
|---|---|---|---|
| Rebalancing bot | -13.74% | -17.19% | -11.54% |
| BTC 100% hold | +7.71% | -5.51% | -0.73% |
| QQQ 100% hold | -5.99% | -8.35% | -7.29% |
Versus the same 25/25/50 start with no rebalancing: -0.58 pts
This was a bad month for my rebalancing bot. In July 2026 it returned -13.74%, while bitcoin gained 7.71% and QQQ lost 5.99%. So I trailed bitcoin by a wide margin and did worse than the Nasdaq proxy as well. Since the start of tracking, the bot is at -11.54% cumulative, bitcoin is at -0.73%, and QQQ is at -7.29%. I would like to say the numbers hide a good story, but they don't.
The best day was July 9, when the bot gained 4.48% against 1.75% for bitcoin and 1.81% for QQQ. The worst day was July 28, when the bot lost 6.08% while bitcoin fell 3.24% and QQQ fell 2.3%. I have no recorded notes explaining either move, so I won't invent a story about them. What I can say is that on both days the bot moved by more than either benchmark, in both directions. That fits a portfolio holding a volatile SOXL position next to bitcoin. Over the month there were 13 up days and 18 down days.
On rebalancing: the setup is BTC spot at 25%, a 1x long SOXL perpetual at 25%, and 50% in USDT on Bybit. The bot checks hourly and rebalances when any weight drifts 2 percentage points or more. This month it made 29 trades, 18 buys and 11 sells. Seven were in BTC and 22 were in SOXL. The comparison I care about is against a buy-and-hold portfolio that started at the same 25/25/50 split. Rebalancing came out 0.58 percentage points behind it. That is small next to the overall loss, but it is a loss, and it means the trading did not help this month.
This is where Shannon's demon comes in. The idea is that regularly resetting a portfolio to fixed weights can harvest returns from volatile assets that move back and forth. It works best when prices chop sideways. It works less well when one asset drops hard in a near straight line, because the bot keeps buying more of the thing that is falling. SOXL went from 258.04 to 121.45 over the month, while bitcoin moved from 58,649.1 to 63,270.7. A sharp one-way slide in SOXL is the kind of month where a rebalancing bot's results can lag a plain hold, and that is what the 0.58-point gap shows.
Drawdown was the painful part. The bot's maximum drawdown for the month was -17.19%, against -5.51% for bitcoin and -8.35% for QQQ. Holding half the portfolio in USDT did not protect me as much as I had hoped, because the SOXL position fell so far and the bot added to it along the way. By month end the weights were back near target, with BTC at 24.95%, SOXL at 25.44% and USDT at 49.61%. I also had two deposit days during the month, so the account did not sit untouched.
My honest lesson is that a rebalancing bot does not remove risk. It enforces discipline about weights, but it cannot make a collapsing asset behave. One volatile holding can dominate the result even at a modest 25% weight, and a month like this shows how much of a bot's result comes from what it holds rather than how cleverly it trades. I am not changing anything because of one month. But I will be watching whether the volatility harvesting really pays for itself over a longer stretch, and I will keep reporting the months where it didn't.
Daily entries this month
How rebalancing works: Shannon's Demon →
Methodology
Total equity is logged by the bot every hour at :59. Deposits are removed at the hour they arrive and returns are chained as a time-weighted return (TWR). Daily return compares the last snapshot of the day with the previous day's last snapshot. Benchmarks use Bybit BTCUSDT spot and QQQUSDT perpetual hourly closes.
This site is a personal record, not investment advice. Past performance does not guarantee future results.