Headline execution
Delay-0
Signal computed after the US market close; the resulting allocation is executed at the next market open.
Public proof lab
Return earns attention. Risk earns trust.
Inspect the Delay-0 next-open headline and the adjacent Delay-1 case that tests a reaction one trading session later. Every figure is simulated / hypothetical—not a live track record.
No return-only ranking · exact windows shown
01 · Timing policy
Three clocks, kept deliberately separate
Delay-0 and Delay-1 test whether historical results survive a next-open reaction or a reaction one trading session later. The separate 14-day clock controls when free users receive a real forward signal.
One-session-late robustness
Delay-1
Signal computed after the US market close; execution is delayed by one additional trading session and occurs at the following market open.
Free signal publication
14 calendar days
Forward observations are released 14 calendar days after computation. This embargo is separate from execution-delay assumptions.
Inspect an execution definition
Signal computed after the US market close; the resulting allocation is executed at the next market open.
Delay-1 is a robustness check: it asks whether historical performance and drawdown remain credible if someone reacts or rebalances one trading session later. It is unrelated to the 14-calendar-day publication delay applied to free signals.
02 · Beta evidence
Zeus Beta — Diversified portfolio signal
Research candidate — public validation in progress. It remains visibly separate from the version-controlled Alpha model and is judged against its own matched benchmark.
Delay-0 headline case
Zeus Beta — Diversified portfolio signal
Simulated / hypothetical research. Delay-0 means execution at the next market open after the signal is computed. It is not the separate public publication embargo.
Research candidate — public validation in progressSource stale · 2026-07-01
Current public state unavailable. Evidence source is 20 calendar days old, beyond the 7-day freshness limit. Current public state is unavailable until the canonical audit refreshes. Historical research evidence remains visible with its exact source date.
— Zeus Beta — Diversified portfolio signal┄ SPMO | Buy & Hold
Canonical chart and metric source through 2026-07-01. Both are simulated / hypothetical. The 14-day publication embargo applies separately to forward-journal observations.
Delay-0 · headline
Signal computed after the US market close; the resulting allocation is executed at the next market open.
CAGR
29.8%
Max drawdown
-17.0%
Calmar
1.76
2018-01-02 → 2026-07-01
Delay-1 · one-session-late robustness
Signal computed after the US market close; execution is delayed by one additional trading session and occurs at the following market open.
CAGR
29.0%
Max drawdown
-17.5%
Calmar
1.66
2018-01-02 → 2026-07-01
Delay-1 shows what the same historical rules would have produced if the reaction or rebalance happened one trading session later. It tests timing robustness; it is not the 14-day delay applied to free forward signals.
Matched benchmark · Delay-0 window
SPMO | Buy & Hold
CAGR
20.1%
Max drawdown
-30.9%
Calmar
0.65
2018-01-02 → 2026-07-01
Expert assumptions and provenance
- Accounting
- total_return_adjusted_ohlc
- Costs
- Fee 0.10% · borrow 5.00% · cash yield 0.00%
- Model version
- multi-sleeve-market-confirmed-riskoff-v1
- Evidence generated
- 2026-07-16T16:16:37.000Z
Data through 2026-07-01. Public observations have a separate 14-calendar-day embargo. Source digest: 7995bc969d71a8010857abcf888891deee00ebe3e5e4c7f4caa165316f5f4f5f
Later-start test · Delay-0
Re-runs the same model from later dates. Each result stays beside CAGR, max drawdown and Calmar for its own exact window.
| Window | CAGR | Max DD | Calmar |
|---|---|---|---|
| 2020-01-02 → 2026-07-01 | 34.5% | -17.0% | 2.03 |
| 2022-01-03 → 2026-07-01 | 33.3% | -17.0% | 1.96 |
| 2024-01-02 → 2026-07-01 | 54.6% | -17.0% | 3.22 |
Top-winner removal · Delay-0 diagnostic
Removes the largest winning trades to expose concentration. The canonical artifact exports this diagnostic for the Delay-0 headline case.
| Removed | CAGR | Max DD | Calmar |
|---|---|---|---|
| Top 1 | 20.7% | -18.4% | 1.12 |
| Top 3 | 8.2% | -18.4% | 0.45 |
| Top 5 | 2.5% | -12.6% | 0.20 |
03 · Alpha evidence
Zeus Alpha — Advanced leveraged signal
Version-controlled Trail3 baseline · model tqqq-trail3-frozen-v1. The experimental rearm variant is excluded. Alpha is paired with its leveraged benchmark rather than ranked against Beta.
Delay-0 headline case
Zeus Alpha — Advanced leveraged signal
Simulated / hypothetical research. Delay-0 means execution at the next market open after the signal is computed. It is not the separate public publication embargo.
Version-controlled Trail3 baselineSource stale · 2026-07-01
Current public state unavailable. Evidence source is 20 calendar days old, beyond the 7-day freshness limit. Current public state is unavailable until the canonical audit refreshes. Historical research evidence remains visible with its exact source date.
— Zeus Alpha — Advanced leveraged signal┄ TQQQ | Buy & Hold
Canonical chart and metric source through 2026-07-01. Both are simulated / hypothetical. The 14-day publication embargo applies separately to forward-journal observations.
Delay-0 · headline
Signal computed after the US market close; the resulting allocation is executed at the next market open.
CAGR
44.7%
Max drawdown
-33.8%
Calmar
1.32
2015-01-02 → 2026-07-01
Delay-1 · one-session-late robustness
Signal computed after the US market close; execution is delayed by one additional trading session and occurs at the following market open.
CAGR
33.3%
Max drawdown
-38.9%
Calmar
0.86
2015-01-02 → 2026-07-01
Delay-1 shows what the same historical rules would have produced if the reaction or rebalance happened one trading session later. It tests timing robustness; it is not the 14-day delay applied to free forward signals.
Matched benchmark · Delay-0 window
TQQQ | Buy & Hold
CAGR
38.1%
Max drawdown
-81.7%
Calmar
0.47
2015-01-02 → 2026-07-01
Expert assumptions and provenance
- Accounting
- total_return_adjusted_ohlc
- Costs
- Fee 0.10% · borrow 5.00% · cash yield 0.00%
- Model version
- tqqq-trail3-frozen-v1
- Evidence generated
- 2026-07-03T06:07:49.000Z
Data through 2026-07-01. Public observations have a separate 14-calendar-day embargo. Source digest: df2d92bd07835b5a305e2e467729daa404fa0afa7c66559cbdb43612ce683ed6
Later-start test · Delay-0
Re-runs the same model from later dates. Each result stays beside CAGR, max drawdown and Calmar for its own exact window.
| Window | CAGR | Max DD | Calmar |
|---|---|---|---|
| 2020-01-02 → 2026-07-01 | 67.8% | -33.8% | 2.00 |
| 2022-01-03 → 2026-07-01 | 55.0% | -32.8% | 1.68 |
| 2024-01-02 → 2026-07-01 | 62.6% | -24.8% | 2.52 |
Top-winner removal · Delay-0 diagnostic
Removes the largest winning trades to expose concentration. The canonical artifact exports this diagnostic for the Delay-0 headline case.
| Removed | CAGR | Max DD | Calmar |
|---|---|---|---|
| Top 1 | 34.7% | -33.8% | 1.03 |
| Top 3 | 22.8% | -33.8% | 0.67 |
| Top 5 | 14.2% | -33.8% | 0.42 |
04 · Reading the audit
What these tests can—and cannot—tell you
Useful questions
- ·Does the result remain understandable after one more trading-session delay?
- ·Does a later start radically change the relationship between growth and drawdown?
- ·How dependent is the outcome on a small number of winning trades?
- ·Are the strategy and benchmark measured over exactly the same dates?
Hard limits
A backtest does not establish future performance, tradability at a displayed price, personal suitability or a live operating record. It can still contain selection bias, data error and assumptions that differ from your implementation.
Forward results live only in the journal and begin at day zero. They are never backfilled from these hypothetical series.
Source artifacts and freshness policy
The page is generated from versioned evidence snapshots. Each snapshot contains its model version, data-through date, source artifact hashes and aggregate source-run digest. When the source is stale, the current public state fails closed and is shown as unavailable; historical research evidence retains its exact date.
Beta source date: 2026-07-01. Alpha source date: 2026-07-01.