Aggregation & Arbitrage Pipeline

Live data pulled from public Craigslist metros across the US, pushed through a 6-stage engine that turns scattered, inconsistently-formatted listings into acquisition-ready deals. This is the ops brain behind the storefront.

Raw listings ingested
—
4 sources · pulled just now
Unique machines (deduped)
—
—
Arbitrage opportunities
—
Underpriced vs market median
Est. gross margin
—
If we buy every opp & resell at median
1

Ingestion

Raw feeds from 4 public marketplaces, pulled live

Production would also pull from MachineryTrader, IronPlanet, Ritchie Bros, EquipmentTrader, Facebook Marketplace, dealer RSS feeds, and auction result APIs — mix of partnerships, rate-limited crawling, and user submissions to stay within ToS.

2

Normalization

Parse messy titles into structured fields · click a raw title to see parser output

Raw titles (as ingested)

Parser output

Regex + keyword classifiers extract year, make, model, type, hours. Production would use an LLM fallback for edge cases & a trained model for equipment-type classification.

3

Deduplication

Same machine cross-posted across metros — grouped into a single deal
4

Pricing Intelligence

Compare each listing to market median for its (type, year bucket) — flag arbitrage opportunities
ScoreListingSourceAskMarket medianΔ vs medianOpportunity
5

Deal Capture

Drag cards across stages · simulates ops team progressing each arbitrage opp
6

Storefront Output

Acquired inventory, listed on HeavyYard.com at market price → margin captured

→ See public-facing storefront · these same units would go live with pro photography, inspection report, and financing quote.