We’re A Predictive Analytics Company, So We Made A Public Project Out Of Showing Off.
Every month we forecast something that hasn’t happened yet, then publish how close we got. Watch for as long as you like. Whenever you’re ready, we’ll take a look at what’s sitting in your data.
The forecast covers where prices are heading in a corner of the US economy, thirty days before the government publishes the real figure. Every prediction is timestamped and locked before the real numbers are published. Nine of the last ten got the direction right, and the whole record is public, (including the one we missed).
We’re building this in public so you can trust us with what’s private.
Hiring an analytics firm normally means taking their word for it. You read a case study about a company you’ve never met, in some industry you’ve never heard about, with numbers you have no way to check.
Then you hand over your data and hope for the best. We’d prefer that you never had to do that, so we built something you can watch instead.
The public forecast
Data everybody already has
Federal series any analyst in the country can download this afternoon. Thousands of people have stared at exactly these numbers. Anything the system finds here, it found with no advantage in the data at all, which is what makes the score worth reading.
Your business
Data nobody has ever touched
Your customer file. Your transaction history. Your claims, your policy book, the CRM that’s been filling up for ten years and getting queried for reports. Nobody outside your building has ever modeled any for it, and nobody else is in a position to do that either.
Most companies are making decisions with a model that somebody built a year or two ago. A model costs a quarter of somebody’s year to build, so it makes sense that you build it once and then live with it. Meanwhile your customers changed, your pricing changed, your competitors changed, and the model keeps answering as though none of that happened.
That’s the constraint we take off the table. Rebuilding stops being a project that you need to plan a year around, so the model gets updated when your data changes rather than when somebody can free up a quarter to do it. Your model stays current with your business and it’s the only kind you should use to make decisions. The forecast that we’re running here is that same thing running on a schedule, in public, where you can watch it work.
That’s how we work. We build a system that creates its own models with data that your company already owns, and we find risk and opportunity that your reports were never asked to show. Sometimes that’s a pricing band that’s upside down. Sometimes it’s the slice of a customer list worth more than all the rest of it combined. Neither one shows up in a report, because a report answers the question it was built to answer, and it was built a while ago.
So the whole thing comes down to this: Watch us commit to a number every month and check us against a result that we couldn’t possibly control. Do that for as long as you need to, and then reach out to us when you want to.
What we predict
The Producer Price Index, thirty days before it lands.
Every month the Bureau of Labor Statistics publishes the Producer Price Index. It tracks what American producers are charging for what they sell, which makes it the earliest official reading on where prices are heading, well ahead of anything that turns up on a shelf or a renewal notice. We forecast three of its components.
Business insurance premiums
What commercial carriers are charging for coverage. It moves ahead of the renewal cycle, so a month of warning is worth real money to anybody buying or pricing it.
Home construction
What it costs to build. Housing supply, lending, and materials procurement all key off this number, and all of them plan months out.
We picked this index because nobody can nudge it. It’s an official number, it lands on a schedule that somebody else controls, and a prediction is either right or wrong in front of everyone. Being early is the whole value of a forecast, and thirty days is enough lead time to do something with it.
Grocery prices
What food costs on its way to the shelf. Of the three it reaches consumers most directly, which makes it the plainest test of whether the system sees a turn coming.
Thirty years of doing this by hand, before we taught software to do it.
Our founder was a VP of Analytics at American Express. He’s spent three decades finding patterns in data for institutions where being wrong is expensive, and he built the practice that did it for AARP, UBS, and Bank of America.
The monthly forecast is what that work became once it was encoded into software that runs on its own. We publish the score because a track record you can audit is the only claim in this field that you can trust.
Get the next one
Grade us for a few months before you ever talk to us.
You get our predictions delivered straight to your inbox before the numbers exist, so you can decide for yourself whether we’re any good at this. It costs an email address and about four minutes a month.