Cortana North Star
Cortana is the engine. The market is the field. The only exit I want is the one into this.
1. The Vision
After the Asymbl acquisition closes, I am not looking for another job. I am building the operation that replaces it. My livelihood will come from trading the markets with a system I own, understand, and continuously improve. The working name for that system is Cortana.
Cortana is not a single strategy. She is an engine designed to run multiple trading layers in parallel, each with a distinct risk profile and time horizon, all feeding into a unified signal infrastructure. Short-term cash generation funds the operation. Swing trades compound capital. Trend trades position me for the macro cycles that produce generational wealth.
The long-term goal, once the engine is proven in live markets over multiple regimes, is to commercialize Cortana as a signal data service. Comparable references in the space are NautilusTrader and zTradez. Subscribers plug into the signals for a fee. I never advise anyone what to do with them.
2. The Three Layers
Cortana operates as a hub-and-spoke architecture. The hub is the signal engine. The spokes are the trading layers below, each a self-contained strategy that reports back into the hub. This structure is non-negotiable because it is also the foundation for the future signal service.
Layer 1 | 0DTE Cash Generation
Purpose. Daily cash flow. This is the quick slant or out route of the operation. It is designed to generate consistent base-hit income off 0DTE SPY signals, not to swing for fences.
Target. 10% wins, 80% win rate, with larger upside captured when price action confirms the thesis intraday. Position sizing and daily loss limits enforce the mandate that this layer generates cash, it does not gamble.
Status. Beta MK2 is built and currently in paper trading. IBKR Pro + IB Gateway + Python + ib_async. ML signal layer using XGBoost (LightGBM as backup) for tabular financial data. Walk-forward validation and purged cross-validation are in place to prevent leakage.
Layer 2 | Swing Trades
Purpose. Compound capital on 2 to 5 day swing trajectories. This is the layer where technical setups and short-term catalysts drive positioning in individual equities.
Status. John Neal is building the beta. I will take it from beta to production by integrating it into the Cortana signal engine, applying the same validation discipline used on Layer 1, and wiring execution through the same IBKR infrastructure.
Layer 3 | Trend Trades
Purpose. Longer holds based on congressional disclosures, insider filings, and macroeconomic regime signals. The objective is to be correctly positioned through market cycles and benefit from multi-month or multi-quarter trends.
Status. Not built. This is a deliberate sequencing decision. Layer 1 must be proven live before capital or development time flows to Layer 3. This layer is where the Munger 200-week MA framework and economic moat thinking converge with the technical and signal-driven methodology from the earlier layers.
3. Architecture | Hub and Spoke
Every component of Cortana is built to plug into a central signal hub. This is a design decision made now, not later, because retrofitting a modular architecture after the fact is significantly harder than enforcing it from day one.
The rules of the architecture:
- Each trading layer is a spoke. It consumes market data, produces signals, and reports positions and performance back to the hub.
- The hub owns signal distribution, risk aggregation across layers, and the future API surface that external subscribers will eventually consume.
- No spoke may directly depend on another spoke. All coordination happens through the hub. This keeps layers independently testable and independently commercializable.
- Execution is centralized through IBKR. Data, logging, and performance attribution are centralized through the hub.
4. Commercialization | The Signal Data Service
Once Cortana has a multi-regime live track record, the engine becomes a product. Subscribers pay a fee to plug into the signals the hub generates. Reference models are NautilusTrader and zTradez. The minutia of pricing, tiers, and go-to-market is not the focus today. The point is that every architectural decision made today is made with this end state in mind.
The Regulatory Constraint
I never register with FINRA or the SEC. Full stop.
This is a hard constraint, not a preference. It is the single most important input to the business model, and every product decision flows from it.
What this means for the business model
Cortana is a data and software product. Not an advisor. Not a broker. Not a fund. This distinction is everything. The business sells access to signals and tooling. It does not manage client money, it does not personalize recommendations, and it does not tell anyone what to trade. Subscribers make their own decisions.
Practical implications:
- The product is positioned as information, research, or software, consistent with how NautilusTrader and similar vendors operate.
- No managed accounts, no pooled investment vehicles, no performance-based fees, no individualized advice.
- Marketing language avoids prediction claims and recommendation language. The signals are signals, not instructions.
- Before the service goes live, a securities attorney reviews the structure, the terms of service, and the marketing language. The line between publisher and advisor is real, and staying on the right side of it is not optional.
This constraint is not a limitation. It is the shape of the business. Building with it in mind from day one is what makes it viable.
5. Operating Principles
- Paper first, capital second. No layer trades real money until it has demonstrated statistical edge on out-of-sample data and survived paper trading across multiple market regimes.
- Validate ruthlessly. Walk-forward validation and purged cross-validation per Lopez de Prado. No in-sample backtests presented as evidence of edge.
- Survive first, compound second. Position sizing, daily loss limits, and drawdown controls sit above every strategy decision. A 55% edge means nothing if leverage takes me out in a drawdown.
- Own the infrastructure. IBKR Pro + IB Gateway + Python + ib_async is the foundation. Mac-based development. Claude Code for implementation velocity.
- Tax structure matters. Trader Tax Status classification with a specialist CPA. Section 475(f) mark-to-market election evaluated each year. SPX vs SPY tax treatment factored into strategy selection.
- Every line of code is built for the hub. If a component cannot plug into the signal engine cleanly, it is not built correctly. No exceptions.
6. Sequencing
Near term | Now through the Asymbl exit
- Take Layer 1 from paper to live capital on a small sleeve once validation metrics hold.
- Integrate John’s swing trade beta into the Cortana hub. Apply the same validation discipline used on Layer 1.
- Finalize Trader Tax Status filing with CPA. Evaluate Section 475(f) election timing.
Medium term | Post-acquisition
- Full-time operation. Trading is the job.
- Layers 1 and 2 running live with attributed performance tracked at the hub level.
- Layer 3 design and build begins. Congressional disclosure data pipelines, insider filing ingestion, macro regime detection.
Long term | The signal service
- Multi-regime live track record established across all three layers.
- Legal review completed. Product positioned as data and software, never as advisory.
- Subscriber API launched. Hub-and-spoke architecture delivers commercial signal product without structural rework.
7. The One Thing
If I forget everything else in this document, the one thing that matters is this: the only way this works is if every decision I make is made in service of the engine I am building, not the trade I am taking.
Cortana is the asset. Individual trades are fuel. Confusing the two is how this fails.