Brian R. Demsey
Modernizing market data
begins with trusting it.
Decades of getting financial numbers right, a habit of building with AI, and one conviction: data people cannot trust is the problem worth solving.
Brian R. Demsey
Decades of getting financial numbers right, a habit of building with AI, and one conviction: data people cannot trust is the problem worth solving.
To the team at Massive:
I'm writing about helping you insure data can be trusted by customers.
A word on who I am, because the combination is unusual. I've spent decades in actuarial science and enterprise systems, a career whose whole premise is that a number must be able to defend how it was derived. More recently I've built and shipped AI systems whose entire job is verification: querying many models at once, measuring where they disagree, and scoring how far each answer can be trusted.
I've looked closely at Massive, and the instinct it was built on is right: market data should be fast, clean, and made for developers, because the incumbents' technology is old and data hard to trust. The frontier now isn't more data or more speed. It's trust, being able to prove the data is right.
That's the problem I want to help with, and it's the one your customers feel most: when Revolut shows prices in its trading terminal, or The Motley Fool shows equity data to millions of readers, they are betting their own product on your numbers being correct.
The method isn't exotic, and you already use a version of it: your 8-K tagging cites a supporting quote, verifies it against the source word for word, scores it, and holds back anything below a threshold.
I'd extend that same discipline to the price and corporate-actions data customers price against: cross-checking values across independent sources, reconciling aggregates against the trades beneath them, validating adjustments across full history, scoring confidence, and catching anomalies before a customer does.
I'll be honest about fit. I won't out-engineer your Rust team, and I don't arrive with a market-data symbology pedigree. What I bring is the ability to put the pieces together and hold a correctness bar, and decades of clients who never had reason to complain. That last part isn't a boast; it's the point. Trust is earned by making the complaints stop.
I've prepared a detailed analysis for one reader in particular, Mr. Pike. I'd welcome a conversation.
Brian R. Demsey
· bdemsey@demsey.com · 949 291 1422 · massive.demsey.com
To continue to my résumé and the rest, enter your email.
For years I've paddled six-person outrigger canoes, and it taught me something about where a person belongs on a crew. In an OC6 you don't pick your own seat. The coach places you, and where you sit depends on the water that day.
Seat one sets the cadence. Everyone behind follows your timing, so you read the conditions and set a stroke the whole boat can hold: steady in flat water, quicker when the swell asks for it. Seat three is the power seat, the engine room; the job there is to pull hard and move the boat. Seat six steers, reading the water ahead, holding the line, keeping the crew on course.
I've been put in all three, and I learned to trust the coach's read over my own preference. I'd say the same to you about where I fit. I can set the cadence: the standard and the rhythm a data-quality effort runs on. I can be the power: the one who does the hard, hands-on work of getting the data right. Or I can steer: reading what's ahead and holding the course on data trust. Tell me where the conditions call for me. You know the water better than I.
This message is written for one reader: Quinton Pike. Verify with a phone number to continue.