What this is, in plain words.
A handbook for the site: every term we use, where the numbers come from, and what the value figures actually mean. Written for me, not for a marketing page — so it's honest about the gaps too.
What it is
the one-paragraph versionThis is a personal tool for the Galway house hunt. It pulls together every Galway sale listing it can reach — Daft, MyHome, the daft sold feed, and a handful of agency sites — into one place, and cross- references each one against the Property Price Register, the Irish state's official record of every residential sale since 2010.
On top of that it runs a measured valuation: for an active listing it asks “what have similar nearby properties actually sold for, and what does a price model think this one is worth?” — and flags the ones that look under-priced, with confidence shown honestly so a thin- evidence listing never reads as a confident bargain.
Galway only. Personal use. No forecasts — just what the data says.
What it gives you
four viewsEvery live listing, list and map side by side, sorted by value, price, beds, or red flags. The sold toggle switches to historic sales — what actually changed hands and how far from asking.
Galway suburb by suburb: active inventory, median asking and €/m², and how the last 12 months of sales closed. Medians over thin samples are suppressed, with the sample size always shown.
Three questions a buyer actually asks — whose market it is, where prices are going, what money buys — answered from the Price Register, Daft's sold feed, the CSO index and live listings. Descriptive, not predictive.
The plumbing, made visible: every data source's last run, whether it succeeded, stalled, or failed, and when each is next due.
How value is determined
Comps, the price model, the residual correction, calibrated bands, the value score, and an honest note on what we tried and rejected.
Open →Data sources
Every source — Daft, MyHome, the agency sites, the PPR, the CSO index — what each provides, how often it refreshes, its lag, and how duplicates collapse into one row.
Open →Glossary
One tight definition for every term the site uses, from AVM and BER to within-10% and the walk-forward backtest.
Open →The honest caveats, up front
read this before trusting a number- Sold data lags. PPR months are roughly 80% complete two weeks after month end and mature by two months; Daft's sold feed matures over 2–3 months — so the latest month or two of any sold series is always provisional.
- Thin evidence is suppressed, not guessed. When too few comparable sales exist, the value score shows nothing rather than a confident-looking number.
- The model is measured, not magic. Its headline accuracy and the coverage of its price bands are tested against real sold outcomes and shown to you — see how value is determined.