Money-Maker Plan — Kimi IDE

Source: Moonshot Kimi (IDE agent) · Generated 2026-05-11 · Most aggressive of the 5 plans

Headline (brutal summary)

The skill is already live at .claude/skills/money-maker-ready/SKILL.md (shipped earlier today); ran end-to-end.

Bottom line: stop backtesting everything. EQUITY and COMMODITY edges are real. Isolate them, size small, and start collecting live track record while you fix FOREX in parallel.

Current state per asset class

Class	n	WR	PF	Verdict
COMMODITY	816	48.7%	2.08	🟢 Best edge — ready for small size
EQUITY	428	52.8%	1.42	🟡 Tier 3 — viable, size small
CRYPTO	8,166	44.8%	1.26	🟡 Dispersion play — elite pockets inside mediocre aggregate
FOREX	1,249	45.6%	0.28	🔴 Portfolio killer — hard stop
ETF	88	53.4%	1.20	⚪ Thin sample — observe only
BOND	18	55.6%	1.72	⚪ Too thin

Overall: 41.7% WR, PF 1.11, Sharpe 1.41 annualized — but that hides the fact that FOREX and a few toxic systems are dragging the entire book down.

The 3 things blocking real money (fixable today)

kimi_signal_tracking is live with -954% PnL and 995% MDD — quarantine it now.
FOREX PF = 0.28 — for every $1 risked you get back $0.28. Exclude until PF > 0.8.
crypto_soc_* variants are backtest-overfitted — they show 66% WR in backtests but 32% WR live (-33.6% decay). Block surgically.

What's actually ready NOW

If you exclude FOREX and the toxic systems, you already have a tradable book in EQUITY + COMMODITY:

COMMODITY cot_positioning_CT_locked LONG: 89.8% WR, PF 13.1 (n=49)
EQUITY rs-breakout-scout LONG: 77.8% WR, PF 6.7 (n=18)
EQUITY Breakout Momentum LONG: 57.9% WR, PF 1.53 (n=38)
System-level: aggregated_picks (77.3% WR, PF 6.42), claude_gainer_st (78.5% WR, PF 6.12)

Recommendation: start a real-money pilot in EQUITY + COMMODITY only, using those proven systems, at 1% risk per trade. That's not "backtesting forever" — that's putting skin in the game where you already have edge.

30-day plan

This Week (P0 + P1)

☐ Blacklist kimi_signal_tracking, crypto_soc_*, and FOREX from real-money flow
☐ Verify the 680% max drawdown figure is using capped PnL (not raw compounding)
☐ Cap quan_engine to 12% of CRYPTO volume (it's an 18% drag at PF 0.70)
☐ Promote st_fear_greed_contrarian (94% WR) to High-Conviction gating
☐ Add walkforward OOS for EQUITY + COMMODITY

Next 2 Weeks (P2)

☐ Wire real CFTC COT data to validate the 89.8% WR COMMODITY pocket
☐ Expand ETF universe (XLF, XLE, XLK) to reach n=120
☐ Build per-symbol PF drill-down in the audit UI

Month 2 (P3)

☐ Integrate Riskfolio-Lib for CVaR portfolio sizing
☐ Rebuild FOREX from scratch (new TP/SL caps, session timing, kill dead strategies)

Go / No-Go gate

Before you deploy $1 of real capital, these 7 checks must pass:

☐ FOREX excluded or PF > 0.8
☐ kimi_signal_tracking blacklisted
☐ crypto_soc_* blocked
☐ Max drawdown calc verified as capped
☐ Walkforward exists for EQUITY + COMMODITY
☐ Rolling 30d Sharpe positive for 7 consecutive days
☐ You personally review and approve the P0 list

Current score: 0/7 — but items 1-4 are 1-hour fixes. You can be at 4/7 by end of day.

Artifacts created (Kimi-side)

reports/money_maker_ready_2026-05-11T2040Z.md — full 11-section audit with all data, sources, and commands
updates/2026-05-11-money-maker-ready-plan.md — fix documentation per AGENTS.md
.claude/skills/money-maker-ready/SKILL.md — repeatable skill for future runs