Results
Three kinds of figures, kept apart: my eToro account, the shadow books my software calculates without capital, and a simulation of the rule on history. Only the first is money.
My eToro account
The account starts on 9 Dec 2026. Its figures appear here month by month from the first full month, January 2027, next to the base mix.
Shadow books since the forward start
My software books AT-UT and its base mix AT0-UT every trading day exactly as registered, from the eToro daily closes of the five ETFs, with costs, but without capital and without orders. They show how the sealed rule behaves going forward.
No booked days yet.
Simulation 1991–2026, not a track record
Simulation: the registered rule computed on history. No real money was invested.
In the simulation 1991–2026 the rule's worst crisis decline was −14.9 %, the base mix's −51.7 % and the US equity market's −50.4 %. The price: it earned less per year than the fully invested base mix and lagged after sudden rebounds.
| AT-UT | Base mix AT0-UT | US equity market | |
|---|---|---|---|
| Worst decline in a crisis window | −14.9 % (Dot-com, Apr 2000–Oct 2002) | −51.7 % (Financial crisis, Nov 2007–Feb 2009) | −50.4 % (Financial crisis, Nov 2007–Feb 2009) |
| Largest decline 1991–2026 | −18.3 % | −54.8 % | −54.6 % |
| Return per year, 1991–2026 | 7.6 % | 10.3 % | 11.2 % |
| Volatility, annualised | 7.2 % | 14.1 % | 18.2 % |
| Worst week | −8.4 % | −16.8 % | −18.0 % |
| Sharpe ratio | 1.05 | 0.77 | 0.68 |
- AT-UT
- Base mix AT0-UT
- US equity market
| Window | AT-UT | Base mix AT0-UT | US equity market |
|---|---|---|---|
| Russia/LTCM, Aug 1998 | −5.1 % | −15.8 % | −15.7 % |
| Dot-com, Apr 2000–Oct 2002 | −14.9 % | −46.0 % | −47.8 % |
| Financial crisis, Nov 2007–Feb 2009 | −10.6 % | −51.7 % | −50.4 % |
| Euro crisis and US rating, Aug 2011 | −4.7 % | −12.8 % | −14.2 % |
| China and oil, Aug 2015–Feb 2016 | −5.9 % | −15.0 % | −14.4 % |
| Fourth-quarter sell-off, Oct–Dec 2018 | −5.3 % | −17.1 % | −20.1 % |
| Covid, Feb–Mar 2020 | −13.8 % | −31.1 % | −34.2 % |
| Covid, 19 Feb–30 Apr 2020 | −13.8 % | −31.1 % | −34.2 % |
| Rate turn, Jan–Oct 2022 | −9.1 % | −26.0 % | −25.4 % |
| Rate shock, May–Jun 2022 | −0.7 % | −12.5 % | −15.2 % |
| Nov 2022 | −0.4 % | −2.8 % | −4.0 % |
| Yen carry unwind, 15 Jul–31 Aug 2024 | −6.0 % | −7.5 % | −8.9 % |
| Tariff shock, 19 Feb–4 May 2025 | −6.4 % | −14.8 % | −19.6 % |
| Tariff shock, Apr 2025 | −2.4 % | −9.9 % | −12.4 % |
| Precious metals, 28 Jan–31 Mar 2026 | −5.4 % | −9.2 % | −9.4 % |
| Oil, 1 Apr–15 Jul 2026 | −3.5 % | −3.6 % | −4.5 % |
Method and source
Stand-in history from 3 Jul 1991 to 31 Aug 2026, 8,853 trading days: daily equity market returns for the USA, Europe and Japan and one-month Treasury bill returns from the Kenneth R. French Data Library, emerging markets from its monthly series spread evenly over the trading days, all in US dollars. The registered rule runs on it through the same code path with costs; its cash earns nothing, as in the account without interest. 16 crisis windows were fixed before the computation. Source: the result file docs/validation/ergebnisse/produkt-2.json, whose SHA-256 and timestamps are in the register under Produkt 2; its rows for AT-UT equal those of the study docs/validation/ergebnisse/at-u-varianten.json.
Drawbacks
- Return per year in the simulation 1991–2026: 7.6 % for the rule against 10.3 % for the fully invested base mix.
- It lags after sudden rebounds; choppy sideways markets cost a little through switching.
- It cannot prevent a sharp drop within a single week: its worst week in the simulation was −8.4 %.
- A simulation knows the history it runs on; the forward results above are the test that counts.
Simulated results do not guarantee future results.