I’ve built a chatbot that uses Vedic astrology (planetary positions, nakshatras, dashas, etc.) to predict S&P500 price movements. It generates buy/sell signals based on astrological patterns, but I’m stuck on how to rigorously test its historical accuracy and automate it for real-time trading. Here’s where I need help:

Mixing Astrology and Market Data

Astrological timings (like planetary transits) often rely on lunar calendars or timezone-specific calculations, but the S&P500 uses Gregorian dates and NY time. How do I programmatically align these timelines without introducing errors?

Turning Astrology Rules into Numbers

My model uses qualitative rules like “Saturn retrograde in Aquarius = bearish signal.” How can I systematically convert these into numerical signals (e.g., -1, 0, +1) for backtesting?

Automating Live Predictions

What’s the best way to pull upcoming astrological events (e.g., via an ephemeris API) and pair them with live market data?

Measuring Success

Beyond standard metrics like Sharpe ratio, how do I validate if the astrology logic actually works? For example, do signals during Mercury retrograde perform worse?