[AreaGuy]

The best way to do this is to create a more accurate credit prediction system. Underwriters rely on credit scores because they’re cheap, quick, and easy way to triage risk. Credit scores are riddled with inaccuracies, but because risk is difficult to model precisely to begin with, these companies are still in business because they work decently well even despite the many errors.

If you could create a system using publicly available data that is a statistically and meaningfully superior way of predicting risk, it would be immensely valuable to the asset owners that are buying mortgages, insurance products etc. that rely on credit scores. You’d have to get them to demand it enough to change their underwriting guidelines (a morass of bureaucracy), but once they do, you’d also gain some steady demand and a moat of competitive advantage.

[crxgames]

Some payday lenders already have their own prediction systems in place using scraped public data. The problem that arises is that due to regulation, credit decisions cannot be made based on certain information. I'm sure this varies across sectors of finance though.

[fweespeech]

> If you could create a system using publicly available data

The problem is, making people's financial transaction data public is essentially illegal and represents a competitive advantage.

You'd have to start with convincing major banks / credit issuers to report people to you. With that data, you could build what you suggest. The real hurdle is convincing people to do that and complying with the regulations.

[rahimnathwani]

"Credit scores are riddled with inaccuracies"

Are you mixing up credit scores and credit histories?

"If you could create a system using publicly available data that is a statistically and meaningfully superior way of predicting risk"

That's what underwriters at banks and other lenders try to do, albeit with not just publicly available data.