gapwatch

the problem

Premarket gaps are worth knowing about before the open, and any strategy that acts on them is worth testing against history before trusting it with attention. I wanted both in one tool: a scanner and the backtester that validates what it flags.

architecture

A Python system pulls market data from the Alpaca and Finnhub APIs, scans for premarket gaps, and runs a backtester over historical data to validate strategies. When a live scan matches, it delivers a notification over Signal.

engineering decisions

• → Two data sources, not one

  Alpaca and Finnhub cover for each other's gaps in coverage and rate limits, so the scan doesn't hinge on a single provider being available.

• → A backtester in the loop, not just a scanner

  A scanner tells you what's happening; a backtester tells you whether acting on it ever made sense. Pairing them keeps signals honest against history.

• → Signal for delivery

  Notifications land on a private, end-to-end encrypted channel I already use, rather than another inbox or dashboard to check.

constraints

Bounded by third-party API rate limits and coverage; built for premarket timing where being early matters more than being fast.