Trade App

A crypto watchlist app built on top of honest trading research.

Role: Solo — design, mobile & research
Timeline: Aug 2026 – present
Stack: FlutterDartPython
Links: Code

Problem

Most trading “strategies” are tuned and tested on the same data, so they look better than they really are.

My role

Designed and built the Flutter app, and ran the research end to end — protocol, data, backtests and tests.

Result

One setup passed every out-of-sample test: worst drawdown about −10% while buy-and-hold lost 64–88%.

Overview

I wanted to know which crypto trade setups actually work — not just which ones look good on a chart. So I wrote the rules down before downloading any data, logged every backtest, and only then started building an app around the setup that survived.

What I built

A Flutter app with live prices from the CoinGecko API, search across the top 100 coins, and a watchlist saved on the device.
Design tokens from Figma (40 colors, spacing and radius) and a five-tab navigation shell.
A research project with 20 pre-registered backtests on BTC, ETH and SOL, based on 22 papers.
55 automated tests, including lookahead checks for every setup and timeframe.

Results

−9.8%

worst drawdown, vs. −85% for buy-and-hold

1.39

Sharpe ratio on the 2020–2025 practice set

tests, including lookahead checks

Screenshots

What I learned

A good-looking backtest means little until it survives data it has never seen. Writing the rules down first kept me honest — and the same discipline now shapes how I build the app.

Next projectQuizzAI