Copy trading on hold; return target withdrawn

Scope: the Binance DEMO and Hyperliquid PAPER accounts from their flat start on 27 September 2026 to 06:40 UTC on 31 October 2026, the findings of the day-30 review dated 29 October, and the decision taken on them. Both accounts trade simulated money. The rule is unchanged.

Live record

• Return. The Binance DEMO account is down 7.49% and the Hyperliquid PAPER account is down 7.56% since 27 September.
• Round trips. Each account has completed 17 round trips, from opening fill to final exit, and all 17 closed at a loss. Four positions are open on each account.
• Current figures. The track record page states the live record at every build, with each fill.

Day-30 review

• Replay. The deployed rule was re-run bar by bar on production Binance candles over 27 September to 29 October, 193 four-hour bars from a flat start. It selected the same coins on every bar and produced the same 15 round trips as both accounts, in the same order. From 7 October every order matched in direction and size, 24 of 24 on each account. Live minus replay at the last bar was −0.57 percentage points on Binance and −0.52 on Hyperliquid, inside the −1.5-point line fixed in advance, and fees, funding and fill prices account for the gap. About 91% of the loss is the rule's own market loss.
• Size of the loss. The return at day 32, −6.4%, was at about the 6th percentile of flat-start backtest windows of the same length since 2021, and about the 9th percentile of windows since 2024.
• Losing run. At the review, 15 round trips in a row had lost on each account; the count is now 17. The longest losing run in the continuous backtest since 2021 is 18. A run of 15 or more occurred in about 2% of 32-day backtest windows.
• Backtest edge. In the review's backtest (4-hour returns, the resize rule as widened on 14 October, data to 29 October), the Sharpe ratio net of costs and funding was 2.1 for 2021–2023, 1.0 for 2024–2026 and 0.7 for 2025–2026. The monthly returns in the published tear sheet give 2.1, 0.8 and 0.7 for the same periods, and those are the figures quoted elsewhere on this site. All of them are hypothetical. The decline since 2024 carries more weight in this decision than one month of live results.
• Bitcoin filter. Seven of the 15 round trips were short positions closed between 24 and 28 October; Bitcoin crossed its 50-day average seven times between 24 and 29 October. Eight alternative filter rules, declared before the runs, each had a lower backtest Sharpe ratio over 2021–2026 than the current rule. None was adopted.

What one month can show

The standard error of an annualised Sharpe ratio estimated from 32 days of returns is about 3.4, so the live record can neither confirm nor reject the strategy. Separating a Sharpe ratio of 1.4 from zero would take about two to three years of live results. One month does test execution, and execution matched the rule.

Decision

• Testing continues on the DEMO and PAPER accounts, with the rule unchanged.
• The return target is withdrawn. The site now states the backtest by period, marked hypothetical. No return is targeted or promised.
• Copy trading is closed. It will not open before the live record covers at least 90 days and passes its review.

Retire limits

Two limits are fixed in advance. If either is reached, the bot stops opening new trades; exits and the stop orders on open positions continue to work, and the owner decides whether the strategy continues.

• 19 losing round trips in a row, one more than the longest losing run in the continuous backtest since 2021 (flat-start 32-day windows reached 20).
• Equity 12.5% below the starting balance.

Next reviews

Day 60 on 26 November 2026 and day 90 on 26 December 2026. Each review is published in this journal.