Good spot — the “missing” 2004–2006 portion usually happens when there aren’t enough recorded condo sale transactions with usable price/size fields in our records for those earlier years to compute a reliable average $psf by region.
In this chart, the first plotted points only start from 2007 (then 2008, 2009, 2010). So it’s not that prices were zero in 2004–2006 — it’s that the dataset doesn’t have sufficient consistent transactions for those years to generate the regional trend line.
If you want, I can try a couple of workarounds to surface earlier years, like:
- switching the metric (e.g., median instead of average if supported)
- using a broader grouping or different property scope (e.g., all private residential instead of only condos)
- pulling raw transactions for 2004–2006 and summarising them
Tell me which you’d prefer: keep it condo only, or broaden to all private homes?