Since the start date: up or down?
+6.8%
2026-08-25 → 2026-10-04 · net of trading fees · funding not included
These figures are net of trading fees but do not include funding, which is shown separately below. The denominator is declared capital, not everything sitting in the account — the account carries additional margin, so measured against all funds in it the percentages are roughly 5/6 of those shown. They are checked and released week by week by our strategy team; we display them and do not assemble them ourselves. They are from our own validation account — not any customer's return, and not a promise of what you would get. The series carries a check value from our strategy team (first characters below). The public verification tool is not live yet; until it is, this is a record, not a protection, and we do not claim otherwise. Check value: 4a0b09c609b43c7c…
Funding is not included in the series above. This period it went against us, about −0.25% of declared capital. We say it precisely because it went against us.
On 4 day(s) this period a protective rule held back part of what the system would otherwise have done that day. That is a self-imposed limit within the design, not a fault.
Two of those days, 1 September and 3 September, were days on which what was held back was exactly the part of the portfolio meant to act as a counterweight. So “within the design” does not mean “no effect”: those two days lifted the published return of the August 2026 issue by 0.36 percentage points, about 18% of that issue’s whole return. That extra return was not skill; it came from carrying less of the risk the design meant us to carry on those days. We have not changed the curve and have not removed those two days, because removing them would make the curve look better, not truer.
This issue in one screen
- This is ETS's second monthly report and the first full one: all of September falls after the start date. Same format as last issue.
- The figures above come only from the NAV series released by our strategy team, counted from 2026-08-25. The methodology has not changed.
- Two days in last issue's window needed one more sentence; it is written under the headline figures. Those two days lifted the return, we say why, and we have not changed the curve.
- We reconciled September's ledger against the exchange's own records: trading fees match to the cent; the gap is almost entirely funding payments, which the ledger does not record, and it makes the ledger read slightly optimistic. Funding is not counted in the NAV; the page's methodology already says so.
One bar per week. Green is a gain, red a loss. We do not publish daily figures: day-level data can be reverse-engineered into what we hold.
Versus simply holding BTC
Same window, two questions: how much was made, and how deep the worst fall was. BTC is measured on daily closes; our side follows the released-series basis of our strategy team. This block keeps the same format every month, rising months and falling months alike; the conclusion is yours to draw.
Over this window BTC +9.5%, we +6.8%; deepest fall BTC −6.9%, ours −4.4%.
Solid line is us; dashed line is buying BTC on the same day and doing nothing.
What happened before the start date
2026-07-30 is the first day this strategy ran live, not the day the account was opened: before that, the account ran a different strategy of ours that we have since retired. That stretch is not part of this record, and we do not present its figures here. Data before the start date comes from our own validation account — not any customer's money, and not a preview of what a customer would get. Before 25 August the account was not running at its own specification, so what it produced cannot represent the strategy, and none of it is counted. Within that stretch, over the 11 trading days from 2026-08-07 to 08-17, one of the system's self-protection mechanisms was in a triggered state: positions could be reduced but not added to, and actual exposure ran at roughly thirty percent of design. That stretch does not represent the strategy's designed behaviour, so we exclude it in full and count it in no statistic. 25 August is the first day it ran at specification, so the figures start there.
What the excluded stretch made or lost on its own, we publish as well — nothing hidden:
What the excluded stretch made or lost on its own, published too.
These three stretches differ from the series above in two ways. First, the denominator: they run on the declared capital of that time, which was smaller, so the same gain or loss looks larger here. Second, the basis: they are measured on what the exchange actually charged, funding included, while the series after the start date does not include funding. The two sets of percentages are therefore not comparable — do not subtract one from the other.
2026-07-30 → 2026-08-06 · net for the stretch −0.5% · not counted · (The strategy team issued this reason in Chinese; the English version is still to be delivered. We do not machine-translate wording we were authorised to publish verbatim — see the Chinese edition for the full text.)
2026-08-07 → 2026-08-17 · net for the stretch −0.7% · not counted · (The strategy team issued this reason in Chinese; the English version is still to be delivered. We do not machine-translate wording we were authorised to publish verbatim — see the Chinese edition for the full text.)
2026-08-18 → 2026-08-24 · net for the stretch +6.0% · not counted · (The strategy team issued this reason in Chinese; the English version is still to be delivered. We do not machine-translate wording we were authorised to publish verbatim — see the Chinese edition for the full text.)
This execution side has not placed a single real-money order for any customer yet. It runs on us first, for long enough, before it runs for you — the order does not reverse.
Backtest reference
A backtest runs the strategy over past prices from the beginning. It is not live trading; it sits here only to give you a sense of scale, not a promise.
Over the same backtest window, what simply holding BTC would have looked like. Same basis as the block above: BTC on daily closes, both sides annualised before comparing. We draw no conclusion for you here.
Over this backtest window BTC annualises to −26.1% with a deepest fall of −51.2%; ours −1.0% and −22.1%. The window spans 548 days.
This backtest curve runs the same decision logic and the same parameters as the live account, but it does not model two real-world constraints: the exchange's minimum order size, and the size of the account. Live, at the current size, more than half of the adjustments that should happen cannot be placed that day because the computed amount falls below the exchange minimum. The smaller the account, the more this bites. So read this curve as the history of the strategy logic, not the history of this system at your size.
There is an explicit rule for how long live trading must run before it can be compared with the backtest; it will be stated in the report when reached. Until then live and backtest figures stay separate, and we do not put one forward as support for the other.
Of the seven fields, what this issue cannot provide
- Win rate · worst single trade: what gets released is NAV points, not a trade log. The methodology for these two cells (for example "share of positive weeks") is our strategy team's call; when set, we fill them; until then they stay marked "not available".
Cells we cannot fill do not disappear. That is part of the format, not an exception for this issue.
What we promised last issue, and what we delivered
- Let you see your own positions in the console: not delivered yet. It stays on the list below, unchanged.
- Turn the free observation tier into a real entry point: not done. For now our effort goes into making this monthly report itself stand up to checking, and we have not started this line of work. Rather than write another date we cannot meet, we are taking it off the list from this issue; when work on it starts again, it goes back on.
- The four performance figures: the NAV curve and the maximum drawdown are now filled in; win rate and worst single trade are still empty, for the reason in the section above.
What we owe you by the 2026-10 issue
- Let you see your own positions in the console. Same test: log in and see whether your positions are there.
- The win rate and worst single trade cells. Each needs our strategy team's sign-off; if it cannot be settled, the next issue will again say "not available" and why.
Written here so that if we miss it, we cannot pretend we never said it.
Market context
Real public data (Binance spot daily bars, UTC calendar months; peak and trough on daily closes). Numbers only, no cycle call.
Aug was the strongest month so far this year: BTC +25.0%.
Zoom out and BTC is −4.6% year to date. From the Jan high of 96,948 to the Jun low of 58,623 (daily closes) it lost 39.5% along the way.
Market context · this month's numbers
September 2026 (UTC calendar month):
| Month open | Month close | Change | Range (intraday high/low) | Annualised vol | Max drawdown (daily close) | True intraday drawdown | |
|---|---|---|---|---|---|---|---|
| BTC | 78,581.30 | 83,623.60 | +6.4% | 74,967.97 ~ 87,395.67 | 41.4% | −6.9% | −8.9% |
| ETH | 2,467.64 | 2,686.01 | +8.8% | 2,356.41 ~ 2,807.34 | 43.5% | −5.1% | −11.5% |
Methodology, identical every issue: Binance spot daily candles, UTC calendar month, change = month open to month close, volatility = standard deviation of daily log returns × √365. The two drawdown columns use different bases: "daily close" misses the intraday low, so it reads smaller. This month BTC's true intraday drawdown was 1.3× the close-based one; ETH's was 2.3×. We show both; printing only the smaller one would make the market look calmer than it was. Every cell is compared against the data script by a machine gate before publication; a mismatch blocks the release. BTC/ETH only: they are the common reference, and we do not publish what we hold.
The limits you should know
If we talk about the stop, we have to say all of it, or it sounds tighter than it is: the stop line is checked once a day, not in real time; we do not place per-trade stop orders on the exchange; there is no automatic intraday floor; and after you deposit or withdraw, this gate pauses until the starting baseline is reset (today we do that by hand; it is not self-service). These four sentences are not a disclaimer; they are the real shape of the protection.
We do not promise to beat BTC.What we have is a line: if cumulative loss reaches at most 25.0% of declared capital the account stops, earlier when our strategy team's daily line is tighter; checked once a day, not in real time; paused on your first round and after any deposit or withdrawal; a stop is lifted only after our review, never automatically. Details on the risk page. BTC has no such line — however far it falls, you carry all of it.