Atlas

The desk's flagship allocation: The Ark as the dominant core, with Futbot and Gringotts as small uncorrelated satellites. It bears the whole book — more return than the Ark alone, at the same drawdown.

PAPER OVERLAY · PLACES NO ORDERS

🦅 Hawk —

Guardian over the three engines · paper phase · alerts to phone via Pushover · auto-liquidation off until go-live

What Atlas is

70 / 20 / 10 — core & satellites

The Ark · 70%

Levered, gated tech engine + gold + managed-futures. The return core — and the desk's best crash performer.

Futbot · 20%

IBKR futures: Nasdaq / gold / BTC, vol-targeted. A different broker, different instruments — genuinely uncorrelated.

Gringotts · 10%

BTC 44-day trend satellite. Small on purpose — big enough to lift return, too small to sink the ship.

Nothing new traded. All three were live paper bots at the time; Atlas was a virtual view of holding them 70/20/10, maintained with ±10–15% drift bands (~1–3 rebalances a year).

Atlas V2 — the challenger (running alongside)

65 / 15 / 20 — Ark / DBMF / Gringotts

Testing found Futbot is the least-diversifying satellite (~0.56 correlated to the tech core — essentially more tech-trend in a costume). V2 swaps it for a managed-futures sleeve (DBMF), which is genuinely crash-positive — DBMF gained +21.6% in 2022 while tech collapsed — and lifts BTC to 20%. Result: similar return, shallower drawdown, best return-per-drawdown on the desk. It’s running head-to-head with V1 to prove it forward. Still paper — no decision yet.

Backtest metric	V1 · 70/20/10	V2 · 65/15/20

V2 backtest uses a managed-futures proxy over the long window; the live track below uses the real DBMF ETF.

Daily profit — what Atlas should be tracking to

The target line is what Atlas should earn at its honest cross-cycle expectation (~21%/yr — the Ark's survivorship-free 18.8% core plus a haircut satellite premium; the Monte-Carlo forward edge cut ~half for crypto uncertainty) on the launch capital. SPY (~8.6%/yr) is drawn the same way. The green actual track builds day by day from the three live bots — a handful of days in is pure noise; judge it against the target line over months, not days.

Why Atlas over the Ark alone

Backtest — total return,

⚠ These figures are each bot as a SLEEVE over THIS backtest window — they differ from the standalone backtest on each bot’s own page (e.g. The Ark reads 40.4% / −21% on its page, vs its sleeve figure here).

BTC's one-time run removed (drift reset to ~20%/yr, its volatility & crash-timing kept) so the numbers are forward-usable, not backfit.

Strategy	CAGR	Sharpe	max DD	Calmar	Total

Growth of $1 — log scale

Atlas V1 70/20/10Atlas V2 65/15/20 (DBMF)The ArkSPY

Monte Carlo — 5,000 × 10-year paths

The winner, stress-tested

Paired block-bootstrap: the same resampled history handed to both engines, so the comparison is fair.

Across the distribution	Ark	Atlas	SPY

Behavior in the crashes

Peak-to-trough in each stress window. The Ark's internal gold/managed-futures make the core resilient; the small crypto satellites add a touch of tail in a crypto-plus-equity sell-off (2022).

Window	SPY	The Ark	Atlas V1	Atlas V2