Case-Shiller vs. MLS Data in Seattle: Two Different Thermometers

A few times a year, national coverage says one thing about Seattle home prices while the local MLS numbers say another. Readers assume someone’s lying. Usually nobody is — the two sources are different instruments measuring different things on different clocks. Knowing which thermometer you’re reading is most of the skill.

Instrument one: the MLS median

The number local reports lead with is the median price of homes that closed in an area in a month. It’s fast (published days after month-end), granular (county, city, sometimes neighborhood), and split by property type. Its weakness is the one that haunts this whole series: it measures what sold, not what homes are worth. A month heavy in townhome closings or light on luxury sales moves the median while no individual home changes value — the mix-shift problem in full.

Instrument two: the repeat-sales index

Case-Shiller (and other repeat-sales indexes built on the same idea) attacks mix head-on. Instead of pooling whatever sold this month, it finds pairs: the same individual house selling twice. If a specific home sold years ago and sold again last month, the price change between those two transactions is a clean, apples-to-apples observation of appreciation — same house, same lot, same street. Aggregate thousands of such pairs and you get an index of how existing homes are actually repricing, with composition largely neutralized. Sale pairs that look suspicious — implausibly large swings, very short holding periods that smell like flips — get down-weighted or filtered.

That design buys accuracy at a price:

• Lag. The index is published roughly two months after the period it covers, and each release is a rolling three-month average. By the time it confirms a turn, the local MLS data flagged it a quarter earlier — and pending sales earlier still.
• No zoom. The “Seattle” index covers the broad multi-county metro area as a single number. It cannot tell you about Ballard vs. Bellevue, and it isn’t designed to.
• Single-family only, resales only. Condos sit in a separate index where one exists at all, and a brand-new home can’t be in a repeat-sales pair until it has sold twice.
• The remodel blind spot. Repeat-sales assumes the house is the same house both times. A home gutted and rebuilt between sales injects renovation value into the pair as if it were market appreciation. Filters catch the extremes, but quiet quality drift — a metro full of kitchen remodels — leaks into the index as price growth.

Side by side

	MLS median	Repeat-sales index
Measures	Price of what sold this month	Repricing of the same homes over time
Speed	Days after month-end	~2-month lag, 3-month smoothing
Geography	County / city / neighborhood	Metro-wide only
Mix shifts	Fully exposed	Largely immune
Blind spots	Composition, small samples	Remodels, new construction, condos

How to actually use the pair

Use the MLS median to ask “what’s happening?” — it’s the timely, local gauge, best read year-over-year alongside the rest of the dashboard.

Use the repeat-sales index to ask “is it real?” — when the median lurches, the index a couple of months later tells you whether homes repriced or the mix just shifted. If the median jumps and the index later shrugs, it was mix.

Use neither to price a specific home. Both are area aggregates. Your house gets priced by comps — that’s a CMA’s job, not an index’s.

And when both thermometers disagree with the number you really care about — what your home would fetch — remember the order of authority: comps first, median second, index for confirmation.

One more dataset worth demanding in the open: agent fees. Manaky Homes is a free marketplace where Greater Seattle agents publish theirs side by side — add yourself to the waitlist and see the real numbers at launch.