kandyb studio
Portfolio · Case study 01

ThesisEdge

Mobile backtesting of stock strategies described in plain language.

iOS · TestFlight Beta 2026 AI · GPT-5-mini 8 languages Offline-first
DCA $500/month into SPY for 20 years CAGR: 9.4% · Max DD: −34% Sharpe 0.68 · 30 years
Get the app: Google Play

About

ThesisEdge lets anyone — beginner or pro — backtest any stock strategy in plain language. Type something like "DCA $500/month into SPY for 20 years, reinvest dividends" and the app simulates it against 30 years of real S&P 500 data in seconds. No spreadsheets, no code, no expensive quant tools.

The heart of the app is an AI translator (GPT-5-mini) that turns natural-language strategies into precise technical parameters. The simulation engine runs locally on the phone across 30+ years of S&P 500 history, accounting for dividends, splits, commissions and capital-gains tax.

Key features

  • Describe strategies in 8 languages (PL / EN / DE / FR / ES / PT / ZH / JP)
  • AI parser validated on 179 test prompts — 100% hit rate
  • Backtest engine cross-validated against vectorbt and backtrader (96% match)
  • 30 years of S&P 500 + 15 ETFs + macro data (VIX, yield curve)
  • Side-by-side strategy comparison and result sharing

Screenshots

Coming soon
Coming soon

Want early access?

iOS TestFlight beta · Android coming next.

Request invite →