🪦 RETIRED 2026-09-27 — Desk Blend no longer trades or updates; this page is frozen as a record. Why: It was built from The Book + Futbot + Gringotts (both of the first two now retired) and carried a hidden ~60% of its risk in Bitcoin. HOUSE v2.9 replaces it as the desk's combined book. (All code, history and logs are kept.)
DeskThe ArkAtlasMapIBKR FuturesDesk BlendR&D

Meta-portfolio · the desk's uncorrelated engines, combined

Desk Blend — 40 / 40 / 20

A virtual portfolio holding 40% The Book · 40% Futbot · 20% Gringotts, rebalanced quarterly. The best risk-adjusted thing on the desk — combining low-correlation engines beats any single bot. Should make ~24%/yr at a desk-best Sharpe 1.67.

Paper overlay · places no orders · moves no money

At retirement — the money & P&L as last updated

What a $500k allocation split 40/40/20 across your three real bots would be worth — live dollars, tracked forward. No money moved.

Daily P&L vs on-track baseline

All three bots + the blend — growth of $100 (backtest, 2012–2026, log scale)

The blend (thick green) rides between its parts: near Gringotts' return with a fraction of its drawdown, smoother than any single bot. Log scale so all four stay visible.
The Book Futbot Gringotts Desk Blend (40/40/20)

Backtest scorecard

The blend beats every single bot on Sharpe. ⚠ Per-bot figures here are over the blend’s 2012–2026 window — different from each bot’s own page.

Why it works — correlations

The Book & Gringotts are nearly uncorrelated (BTC vs. equities). That's the free lunch the blend harvests.

Low correlation = the pieces zig when the others zag → the blend's Sharpe () exceeds the best single bot's.

The allocation

Rebalance quarterly back to 40/40/20 — trims whatever ran up (usually BTC after a spike), tops up the laggards. That mechanical sell-high/buy-low is what harvests the diversification.