Infrastructure · Ops Console

Team & Agents

Status of every automated piece of the system. Read-only for now — no remote triggers yet.

1 running1 needs attention0 not built
Run Research Agent

No scheduled automation yet (HF Jobs needs prepaid compute credits) — run it locally whenever you want new hypotheses proposed and screened.

source .venv/bin/activate && python scripts/research_agent.py \ --csv-root data/raw/forex/extracted \ --hf-state-repo KUTech/trading-llm-agent-state \ --num-proposals 3

Requires OPENROUTER_API_KEY in a local .env file. Results push automatically to KUTech/trading-llm-agent-state on Hugging Face — this dashboard reflects them next time it's rebuilt.

Jolash Reasoning Model

Fine-tuned chat/reasoning assistant (Qwen2.5-0.5B LoRA)

Needs attentionlast run Jul 6, 12:14 AM
07/06, 00:14

Compared fine-tuned vs base model output — fine-tune produced unresolved <think> traces instead of direct answers. Needs dataset rework or retraining.

07/05, 21:14

Completed training run on Colab (Qwen2.5-0.5B, 5000 examples, 157 steps) and pushed adapter to KUTech/trading-llm.

Testing Engine

Deterministic backtest pipeline (features -> model -> honest metrics)

Donelast run Jul 5, 11:10 PM
07/05, 23:10

Ran EURUSD 4h-direction backtest with realistic spread costs — confirmed no exploitable edge (Sharpe -0.86).

07/05, 22:40

Trained baseline HistGradientBoostingClassifier — validation AUC 0.513, at chance level.

Research Agent

Proposes specific, testable trading hypotheses via LLM API (OpenRouter), auto-screened

Idlelast run Jul 9, 10:00 AM
07/09, 10:00

Proposed 4 new combos -- EURUSD 1h h=12, USDCHF 4h h=24, AUDUSD 1D h=12 cleared the cheap screen (NZDUSD 4h h=12 rejected). Second, independent USDCHF signal is notable given USDCHF is already the one confirmed candidate.

07/06, 18:10

Proposed USDCAD 1D horizon=24 and AUDUSD 4h horizon=12 -- USDCAD cleared the cheap screen with a notably strong AUC=0.6223 (small daily sample, needs full validation before trusting).

07/06, 17:55

Proposed EURUSD 4h horizon=12 and USDJPY 1D horizon=24 via OpenRouter (gpt-4o-mini) -- both auto cheap-screened, state persisted to KUTech/trading-llm-agent-state.

USDCHF Paper Trading

Live hourly signal + vol-target sizing + drawdown guard against an OANDA practice account (launchd, real order flow, no real capital)

Runninglast run Jul 13, 4:02 AM
07/13, 04:02

Held +1000 units long (proba_up 0.575, vol_scale 1.000, drawdown guard untripped).

07/13, 02:02

Flipped flat -> long (proba_up 0.553 crossed the 0.55 confidence threshold).

07/12, 21:02

First live decision after launch -- position_sizing.py (vol-target scale + drawdown guard) wired in mid-session.

Central Bank Sentiment Fetcher

LLM-scores Fed/SNB policy-statement hawkishness (2015-2025) into a candidate carry-adjacent feature

Donelast run Jul 12, 11:12 PM
07/13, 04:00

Full validation on USDCHF 1D h=20: Sharpe 0.44 (2024-25) vs. -0.66 (2021-22) -- REJECTED, failed robustness. Same pattern as the raw carry features.

07/12, 23:12

Fetched and LLM-scored Fed + SNB statements 2015-2025 -> data/sentiment/USD.csv, CHF.csv.