ECSTA.SI / Project 003

Auction
Intelligence.

An auction data system that turns a flood of lots into rankings, watchlists and decisions before scrolling becomes a medical condition.

TypeInternal product
ServicesData / Automation / UI
StatusWorking prototype
ProphecyNot included
LOTS
WITH SIGNAL.

The problem

Auction listings are built for browsing, not analysis. Finding the strongest items means repeatedly checking prices, categories, descriptions and changes across a catalogue large enough to make manual scrolling feel punitive.

Constraint

The source data is inconsistent. Descriptions are optimistic, categories are creative and photographs occasionally appear to have been taken during an earthquake.

The idea

Treat the auction as a dataset. Pull the listings, clean the information and rank the lots using useful signals so attention goes to the strongest opportunities first.

  • Keep source information visible.
  • Separate automated ranking from human judgement.
  • Surface unusual movement and high-value lots quickly.

The build

The system combines scraping, structured storage and a compact interface for top values, unusual movement and manually selected opportunities. Recommendations remain explainable so they can be challenged rather than worshipped.

Still imperfect

Ranking depends on the quality of source data and historical comparisons. It is a decision aid, not a crystal ball with an API endpoint.

The outcome

The workflow removes a large amount of repetitive scanning and creates a repeatable weekly view of the auction. Human attention can now be spent on judgement instead of finding Lot 618 for the fourth time.

See current status