SignalPilot

SignalPilot

SignalPilot is an AI-assisted crypto decision system designed to help everyday traders make smarter, safer capital decisions without becoming full-time market analysts.

It scans the market, identifies credible opportunities, compares them against current holdings, explains the risks, and tracks what happened after each recommendation.

SignalPilot does not trade automatically. The human remains in control.

Inspiration

Crypto traders have access to more data than ever, but more information has not necessarily produced better decisions.

Charts, alerts, influencers, exchange dashboards, social media, and market commentary all compete for attention. Traders are often left piecing together fragmented signals and making capital decisions based on noise, familiarity, or fear of missing out.

Most crypto tools ask: Which coin might rise next?

SignalPilot asks a harder and more useful set of questions:

• Why is this opportunity credible?
• What return is expected?
• What could go wrong?
• What would invalidate the thesis?
• Is this better than holding the current portfolio?
• What happened after the recommendation?

SignalPilot turns those questions into a structured, evidence-based decision process.

What SignalPilot Does

SignalPilot converts a noisy crypto market into a clear capital decision.

It scans a broad universe of cryptocurrencies, filters weak or incomplete opportunities, and researches the strongest candidates in greater depth.

Each opportunity is assessed for:

• Expected return
• Probability
• Downside risk
• Confidence
• Portfolio fit
• Invalidation The strongest candidates are then compared against one another and against the trader’s existing holdings.

SignalPilot can recommend:

• HOLD
• WATCH
• REDUCE
• ROTATE
• EXIT
• Take no action

Every recommendation is recorded before the outcome is known. The original thesis, entry price, expected return, invalidation level, repeat recommendations, checkpoints, and later results are preserved.

This creates a permanent decision record and prevents the system from rewriting history after the market moves. The result is not another stream of crypto signals. It is a disciplined decision system that helps traders understand:

What should I do, why should I do it, what is the risk, and was the decision ultimately correct?

How We Built It

SignalPilot was built through an ongoing development loop using OpenAI Codex and ChatGPT with GPT-5.6.

GPT-5.6 helped define and challenge:

• Product logic
• Capital framework
• Risk rules
• Invalidation requirements
• User experience

Codex worked directly with the Apps Script codebase to:

• Inspect the system
• Trace data flow
• Repair broken workflows
• Strengthen the recommendation lifecycle
• Validate workbook outputs
• Prepare the technical submission

This allowed the project to move from an experimental spreadsheet system into a working, testable decision platform.

Challenges

The hardest problem was not generating more signals. It was creating a trustworthy chain between research, recommendation, capital action, and later outcome.

SignalPilot had to prevent:

• Incomplete research becoming actionable
• Recommendations being overwritten
• Missing data creating false confidence
• Current holdings being ignored
• Rotations being suggested without a clear source of capital
• Provider failures stopping the workflow
• The system appearing to have authority to trade

These challenges were addressed by separating research, ranking, portfolio comparison, recommendation tracking, validation, and audit into distinct layers.

Accomplishments

SignalPilot currently supports:

• Broad market scanning
• Deep historical research
• Ranked opportunity selection
• Expected return and drawdown analysis
• Portfolio-aware recommendations
• Explicit invalidation rules
• Recommendation lifecycle tracking
• Human approval before every capital action
• Permanent decision and outcome records

During its initial live operating period, SignalPilot supported approximately 20% portfolio growth.

More importantly, it created a repeatable way to answer: What did we know, what did we decide, and what happened afterward?

What Is Next

The next stage is to turn SignalPilot into a simple consumer APP. For now, I am focused on building and validating the decision system that will power it.

The immediate priorities are:

• Improve the quality and consistency of recommendations
• Strengthen how SignalPilot evaluates risk, confidence, and invalidation
• Make portfolio comparisons and rotation decisions more accurate
• Continue testing recommendations against real outcomes
• Refine the system’s learning and accountability loops
• Design a cleaner consumer experience around the Dashboard and Candidate Watchlist

Right now, the focus is on building and validating the decision engine first. Once the system is reliable enough to carry its own weight, the next step is to bring that intelligence into an accessible APP.

Built With

• OpenAI Codex
• ChatGPT with GPT-5.6
• Google Sheets
• Google Apps Script
• JavaScript
• GitHub
• External cryptocurrency market-data sources

Built With

• api
• apps
• binance
• chatgpt
• codex
• coingecko
• data
• drive
• financial
• google
• javascript
• learning
• machine
• openai
• script
• sheets
• technology