S.J. Rubin — Bot Development Log

Features shipped

In progress

Planned

Known bugs

Phase 1

Foundation & core architecture

✓ Complete

Mar 2026
Week 1

Core bot architecture

Full Python bot with signal cycle, corroboration filter, risk manager, trade arbiter, position tracker, and exit manager. Connected to Interactive Brokers via ib_insync on paper account port 7497.

main.pybroker.pyrisk_manager.pytrade_arbiter.py

Done

Mar 2026
Week 1

Instrument universe

Energy (USO, UNG, XLE), defense (LMT, RTX, NOC, GD, BA, ITA), safe havens (GLD, TLT), AI/semiconductors (NVDA, TSM, AMD, MSFT, GOOGL), AI energy infrastructure (VST, CEG, EQIX, CCJ, NEE). Each chosen for predictable geopolitical linkage.

config.pystrategy.py

Done

Mar 2026
Week 1

Tiered keyword taxonomy

Tier 1 (2.0× weight), Tier 2 (1.5×), Tier 3 (1.0×). Cross-cluster amplifier fires at 1.25× when two thematic clusters trigger simultaneously. Covers energy, defense, Taiwan/China, POTUS/policy, Israel/Middle East, AI sector.

signals/keywords.py

Done

Mar 2026
Week 1

Polymarket signal (35% weight)

Fetches prediction market YES prices every 15 min. Real money bets weighted as highest-conviction signal. 78 markets monitored across geopolitical themes.

signals/polymarket.py

Done

Mar 2026
Week 1

RSS feed signal (15% weight)

44 live news feeds across Reuters, defense, energy, and geopolitical sources. Refreshed every 2 minutes. VADER sentiment scoring with credibility weighting per source tier.

signals/rss.py

Done

Mar 2026
Week 1

News sentiment signal (20% weight)

AI-scored headline tone across all feeds. Global average sentiment score drives composite signal alongside per-instrument keyword matching.

signals/sentiment.py

Done

Mar 2026
Week 1

Telegram alerts + remote commands

Real-time alerts for every trade, rejection, and error. Remote commands: /status, /positions, /signals, /pause, /resume, /closeall.

telegram_handler.py

Done

Mar 2026
Week 1

SQLite database + dashboard

All signals, decisions, and trade outcomes written to local SQLite. Streamlit dashboard reads from DB, auto-refreshes every 30 seconds. Dark theme.

database.pydashboard.py

Done

Phase 2

Signal expansion & volatility engine

✓ Complete

Mar 2026
Week 2

OPRA options flow signal (15% weight)

Live options chain data from IBKR for LMT, RTX, USO, GLD, NVDA, TSM. Scores based on: unusual call/put volume (>2× 20-day average), put/call ratio direction, and IVR (implied volatility rank vs 52-week history). IVR ≥ 80 caps score at −0.3 as a volatility warning. Returns 0.0 on IBKR disconnection.

signals/options_flow.pydatabase.pydashboard.py

Done

Mar 2026
Week 2

Volatility arbitrage engine — Strategy 2

Three separate vol strategies running in parallel: (1) Oil dispersion — crude IV vs energy equity IV mismatch; (2) EM panic harvesting — sell overpriced puts on EWW/EPOL during temporary shocks; (3) Safe haven skew — GLD/FXY ratio spread when crisis pricing diverges.

vol_strategies/strategy_oil_dispersion.pyiv_fetcher.py

Done

Mar 2026
Week 2

6-signal weight model

Updated from 5 to 6 signals: PM 35%, Sent 20%, RSS 15%, OPRA 15%, Mom 10%, Gov 5%. Weights are starting estimates — optimizer adjusts based on real performance data.

config.pystrategy.py

Done

Mar 2026
Week 2

RSS live cache fix

Dashboard was showing stale zeros for RSS on startup because it read from DB (written every 15 min) instead of the live 2-min cache. Fixed: dashboard now calls fetch_rss_signal() directly for RSS column.

dashboard.py

Fixed

Phase 3

Self-improving AI loop

✓ Complete

Mar 2026
Week 3

Anthropic API optimizer loop

strategy_optimizer.py logs all signal scores, decisions, and trade outcomes to .jsonl files. Periodically sends full history to Claude via Anthropic API. Claude analyzes what's working and recommends updated parameter values (signal weights, profit targets, thresholds). Bot applies changes to config.py automatically.

strategy_optimizer.pyperformance_logger.pyconfig.py

Done

Mar 2026
Week 3

Thematic coverage scanner

Weekly scan: Claude analyzes last 7 days of RSS headlines, identifies thematic gaps in keyword coverage, proposes new keyword clusters and instrument baskets. Proposals written to thematic_proposals log for human review — not auto-applied.

strategy_optimizer.py

In progress

Phase 4

Institutional performance tracking

▶ In progress

Phase 4 progress 0 / 5 components shipped

Queued

Trade audit log (append-only)

Immutable per-trade record with: entry/exit timestamps, signal scores at time of decision, which sources agreed, VIX at entry, NAV at entry, exit reason, strategy attribution. Written on entry, updated on exit. Never edited — establishes provable track record.

performance/audit_log.py

Planned

Queued

Performance metrics calculator

Sharpe ratio (annualized, minimum 10 trades), max drawdown, win rate by symbol/strategy/signal source, average hold time (winning vs losing), profit factor, consecutive win/loss streaks, best/worst trade.

performance/metrics.py

Planned

Queued

Live readiness monitor

Tracks rolling 30-day window against go-live criteria: Sharpe ≥ 1.5, max drawdown < 10%, ≥ 20 trades, win rate ≥ 45%, no single day > 3% loss. Sends Telegram alert when all criteria met for 30 consecutive days. Never auto-switches — human confirmation required.

performance/readiness.py

Planned

Queued

Exportable performance report (PDF + CSV)

Generated weekly (Sunday 17:00 ET) and on demand. PDF: summary metrics, last 20 trades table, signal performance breakdown, readiness progress bar. CSV: full audit log with summary row. Both saved to data/reports/ and announced via Telegram.

performance/report.py

Planned

Queued

Performance dashboard tab

New tab in existing Streamlit dashboard showing all metrics, live readiness criteria checklist (Day 0/30), and last 10 closed trades. Read-only.

dashboard.py

Planned

Phase 5

Dual-instance architecture & DevOps

Planned

Live + dev dual instance model

Two parallel bot instances running simultaneously at all times. Live instance trades real capital on IBKR. Dev instance runs paper trading. Dev performance is directly comparable against live at any time — no guessing whether a change improves outcomes.

architectureibkrpaper vs live

Planned

Git branching model + automated promotion

Three-branch structure: main (live production), dev (active iteration), feature/x (isolated per-feature). Promotion path: feature → paper validation → dev → test suite → main. Instant rollback available at the code level at any point.

gitgithubCI/CD

Planned

Daily maintenance window at market close

Bot shuts down at market close each day for scheduled maintenance: automated test suite on pending updates, deploy validated code from dev to live, automated rollback if tests fail, system health checks before next open. If tests fail: operator alerted AND bot auto-reverts simultaneously.

schedulerdeploymentauto-rollback

Planned

Feature toggles — zero-redeployment A/B testing

New logic implemented as on/off switches. Toggle a feature on in the dev paper instance, compare performance against live baseline, flip it off if it underperforms. Enables rapid A/B testing of individual strategy components without touching the live bot.

config.pyfeature flags

Planned

Promotion criteria: dev to live

Before any update is promoted from dev to live it must: outperform the baseline paper account over a validation period (minor changes ~1 day, major features ~5+ days), pass all geopolitical tier stress tests, show no execution speed degradation, and receive explicit operator approval.

deploymentvalidation

Planned

Phase 6

Tiered geopolitical event database Novel IP

Planned

Nine-level severity classification system Novel IP

Historical geopolitical and macroeconomic events classified into three tiers × three sub-tiers = nine severity levels. Each level has corresponding real historical market data. Green 1–3: stable conditions, capital mostly parked in yield instruments. Yellow 1–3: elevated tension, capital shifting toward active deployment. Red 1–3: crisis conditions, full deployment mode. Used to stress-test the bot before any update goes live.

event_database.pystress_tester.py

Planned

Live distribution curve + today's risk marker Novel IP

The nine tiers map to a normal distribution curve. Most days cluster in the middle (Green and low Yellow). True crisis events (Red 3) are rare outliers on the far tail. The system displays where TODAY sits on this curve in real time — a live marker on the distribution showing current geopolitical risk score. This drives capital allocation decisions automatically.

risk_curve.pydashboard Tab 3

Planned

Outlier detection protocol + crisis capital reserve Novel IP

When conditions fall outside the normal distribution beyond Red 3 into true outlier territory, a dedicated outlier protocol fires. System cross-references closest historical analogues and executes with conviction while most participants are paralyzed. A dedicated capital reserve (10–15% of fund) is held specifically for outlier deployment and never touched in normal conditions. Outliers are the highest-earnings-potential events in the system.

outlier_protocol.pycapital_allocator.py

Planned

Self-reinforcing database from live trades Novel IP

Live trading signals that exceed a defined strength threshold — both positive and negative outcomes — are automatically written back into the historical event database. The database grows smarter from real execution data, not just manually curated history. Over time this compounds the accuracy of tier classification and stress-testing.

event_database.pywrite-back threshold configurable

Planned

Phase 7

Multi-AI benchmarking & collaborative refinement Novel IP

Planned

AI models competing in paper trading simultaneously Novel IP

Multiple AI models — Claude Opus, Claude Sonnet, ChatGPT, Gemini, Perplexity, Grok — run simultaneously in paper trading accounts, each generating signals independently. Performance compared in real time against the live baseline. Answers a question no platform currently asks: which AI model actually makes better trading decisions? Proven with real paper trading data, not benchmarks.

ai_benchmarker.pydashboard Tab 2multi-provider

Planned

Collaborative refinement mode Novel IP

Beyond competition, models operate collaboratively: one model generates a signal hypothesis, another critiques or refines it, the result feeds back to the first for iteration. Creates a cross-model feedback loop that catches blind spots individual models miss. All iterations logged for analysis and contribute to the signal optimization database.

ai_orchestrator.pycollaborative mode toggle

Planned

Secrets-based model-agnostic AI orchestration

API keys for each AI provider managed through Settings tab. System routes signal generation through whichever models are configured. Model-agnostic architecture — new models can be added as they emerge without structural changes.

settings.pydashboard Tab 4

Planned

Phase 8

Capital allocation engine & full dashboard

Planned

Always-deployed capital allocation module

Capital is always in one of three states: earning yield in liquid safe instruments, partially deployed in active positions, or fully deployed on strong signals. Zero idle cash. The allocation engine reads the current position on the geopolitical distribution curve and automatically adjusts the ratio between safe parking and active deployment as conditions shift. Green: majority in T-bills/HYSA/dividend ETFs. Yellow: capital migrating toward active. Red: maximum deployment. Outlier: crisis reserve fires.

capital_allocator.pytier-based migration

Planned

Signal-strength-proportional deployment sizing

Deployment percentage is proportional to signal strength. Weak Green signal → 5% of available capital. Strong Red signal → 50%. Outlier protocol → full crisis reserve. All thresholds configurable in Settings tab.

capital_allocator.pyconfig.py

Planned

Full 4-tab dashboard

Tab 1 — Strategy management: live trading column + dev/paper comparison columns (Strategy X, Y, Experimental). Tab 2 — AI benchmarking: each configured model's paper trading performance vs live baseline, collaborative mode toggle. Tab 3 — Historical curve explorer: interactive normal distribution with all historical events plotted, live today marker, click-through event details, dropdown filter by category. Tab 4 — Settings: IBKR params, AI API keys, risk thresholds, deployment ratios, notification prefs.

dashboard.pyTab 1–4streamlit

Planned

Competitive landscape

Novel features vs existing platforms

Reference

Feature	This system	NexusTrade	QuantConnect	Vortex AI	SentinelXQuant
Tiered geopolitical stress testing (9-level, real historical data) Novel IP	✓	✗	✗	✗	✗
Live distribution curve + today's risk marker Novel IP	✓	✗	✗	✗	✗
Outlier detection + dedicated crisis capital reserve Novel IP	✓	✗	✗	✗	✗
Self-reinforcing historical database from live trades Novel IP	✓	✗	✗	✗	✗
Claude API continuous signal optimization loop Novel IP	✓	✗	✗	✗	✗
Multi-AI benchmarking in live paper trading Novel IP	✓	✗	✗	✗	✗
Cross-model collaborative signal refinement Novel IP	✓	✗	✗	✗	✗
Multi-column live vs paper comparison dashboard Novel IP	✓	✗	✗	✗	✗
Tier-based automatic capital allocation (no idle cash)	✓	✗	✗	✗	✗
Feature toggles for zero-redeployment A/B testing	✓	✗	✗	✗	✗
Backtesting / historical simulation	✓ via stress testing	✓	✓	✓	✓
Paper trading	✓ Multiple simultaneous	✓	✓	✓	✓
Live trading via broker API	✓ IBKR / ib_insync	✓	✓	✓	✓

Active bugs

Known issues under investigation

3 open

Confirmed
Mar 20

Polymarket keyword false matches on sports/entertainment

Bare substring matching causes "war" to hit "Warriors" (NBA), "oil" to hit "Oilers" (NHL), "PLA" to hit "Playboi Carti". All 78 monitored markets return 0.0 score because genuine geopolitical signals are diluted by sports noise. Fix: whole-word regex + sports pre-filter.

signals/polymarket.pyroot cause confirmed

Bug

Confirmed
Mar 20

OPRA Error 10089 — market data subscription mismatch

OPRA subscription purchased on live account, not paper account. Paper account gets Error 10089 on every instrument. Fix: add ib.reqMarketDataType(3) to use delayed data (free, always available, sufficient for 15-min cycle) OR subscribe to OPRA on paper account separately in IBKR.

signals/options_flow.pyIBKR paper account

Bug

Confirmed
Mar 20

Unawaited coroutines — accountSummaryAsync + qualifyContractsAsync

RuntimeWarning fires every cycle: risk_manager.py:731 and main.py:571 call async versions of IBKR methods without await. Cosmetic but noisy. Fix: replace both with synchronous equivalents ib.accountSummary() and ib.qualifyContracts().

risk_manager.py:731main.py:571

Bug

Phase 9 — Roadmap

Thematic expansion & commercialization

Upcoming

✓ Shipped

✓Energy + oil geopolitics

✓Defense + military

✓AI + semiconductor wars

✓Safe haven flows

✓Emerging market shocks

✓POTUS / US policy

✓Volatility arbitrage engine

✓Self-improving AI loop

✓Comprehensive logging

✓711-event geopolitical DB

✓Historical analog engine (15% signal)

✓24/7 intelligence cycle

✓RSS → corroboration pipeline

✓Crisis monitor (Iran/Hormuz)

✓Capital allocation engine (BIL parking)

✓Mac launcher app (BotLauncher.app)

✓Embedded WKWebView dashboard

✓Settings panel + first-run wizard

✓Multi-AI benchmarking arena

✓IP protection documented (4 claims)

▶ In progress

●Trading floor (multi-pane live vs paper)

●AI deliberation arena (95% consensus)

●Market expansion lab (autonomous engines)

●Provisional patent application

●GitHub Pages site launch

●GDELT DB expansion (target 2,500+)

●Onboarding wizard (non-technical users)

○ Planned

○Live risk distribution curve

○Biopharma engine (FDA events)

○Rare earth / critical minerals engine

○Space / nuclear / quantum engine

○Weekly PDF/CSV performance report

○Friends + family fund

○Prop firm / fund licensing

○Go live (paper → real capital)

90-day commercialization track — Day 3 of 90 Target: Sharpe ≥ 1.5 · Max DD < 10% · 30 consecutive days

⚖️ Intellectual Property

Novel IP — established March 23, 2026

Invention date documented: March 23, 2026. Four novel systems identified and timestamped below. Patent attorney consultation in progress.

IP-01

AI Deliberation Trading System

Multi-AI hypothesis/critique loop with consensus threshold (95%+) as trade execution gate. AIs iterate rounds until agreement — not voting, deliberation.

IP-02

Multi-Pane Paper/Live Competition Framework

Parallel strategy configurations competing against identical real-time market data, with automated promotion criteria from paper to live.

IP-03

Autonomous Market Expansion Lab

AIs autonomously identify, propose, backtest, and promote new trading engines via deliberation-gated pipeline before any engine goes live.

IP-04

Historical Analog Engine as Live Signal

Vector similarity against calibrated 711-event historical database as weighted real-time signal component (15% weight, tunable by optimizer).

Next steps: Provisional patent application · NDA for any external sharing · Attorney consultation via patent-attorney-questionnaire.html

Geopolitical Trading BotBuild History

Geopolitical Trading Bot
Build History