Ready for some fun? I call it, “Frolicking with the Data,” and tonight’s episode is sponsored by Altos Research, a local company providing real estate data analysis tools. [Full disclosure: I'm friends with the founders Mike Simonsen and Jason Buberel.] They noticed a missing niche in the market (”gaping hole” is more like it) for high-quality, sensible, insightful real estate data analysis (ie completely unlike what you get in most newspapers) and charged in head-first.

Tonight we’ll look at Palo Alto, the home of Stanford University and numerous startups.

We start with a very simple chart of the median list price.

A bit small, isn’t it? No worries — we adjust one of the parameters and — Voila! — we get a bigger version:

Much better! (And yes, that number is correct…the median list price of homes in Palo Alto is indeed currently just under $1.8M; half are above that, and the other half below. The lucky one sitting right at the median is on Fulton Street, near to downtown).

You’ll notice, however, that these numbers only go back one month. We tweak another parameter, and once again, like magic, a new chart appears:

Much better! Now we’re going back a full year.

Notice how volatile the chart is? A favorite tactic of statistically-challenged real estate reporters is to pluck a favorable portion of a chart — preferably one with a steep climb or drop — and make an entire story out of it. “From January to February of 2006, prices in Palo Alto plunged by over 20% from nearly $2M to just over $1.5M. Surely, the apocalypse is upon us.”

The only way to fight bad data analysis is with good data analysis. In this case, that means quite simply looking at the bigger picture — ie the whole year instead of just a month. Also, we can smooth out those variations by taking a longer-term “rolling average” which gives a better high-level perspective, like so:

Now the “Jan-Feb” apocalypse story doesn’t look so compelling, does it? Instead, we have a “First half of the year apocalypse” story, followed by a second half “Hold on to your hats” story as prices recover.

What’s behind those movements? Let’s go down one more level of detail and look at the quartiles.

First Quartile

Second Quartile

Third Quartile

Fourth Quartile

The second, third, and fourth quartiles show the same general trend as the overall chart above — a steady decline in the first half of the year, followed by a steady increase. The first quartile — the lowest 25% of the homes — showed less of a decline but a similar pattern. What does this tell us? The decline was felt in all sectors of the market.

Altos Research offers a lot more metrics than just the median, and we’ll explore those later on.

How could they improve their product? I see three things:

1. Offer analysis on “sold” as well as “active” prices
2. Provide numbers going back further than just one year
3. Do geographic analysis more fine-grained than zip codes — down to the neighborhood level, ideally

I understand these capabilities may be coming soon!

• Zaccuracy Update
• Arms dealers, mercenaries, and keys to the castle
• Sunday Morning Data Geekage