[rbres]

It's not zero sum if you have overall greater precision. We believe we do and have proven it across dozens of case studies.

As we publish more, we hope to let data talk. If I were in your shoes, I'd be similarly skeptical. Most companies in the space overpromise and underdeliver.

[amcnett]

I would argue it is zero sum ("you've got some fraud" vs "you have no fraud") if there is any great volume of transactions and what's being sold is valuable, and/or marketable at a black market price below retail, and/or offers fraudsters liquidity options. I am skeptical because an untrained system is going to make mistakes out of the gate, even if it's trained on oodles of transactions from other businesses. Anti-fraud isn't a one-size-fits-all game.

[rbres]

We've been processing orders for a year and a half in stealth with many clients. So we have a lot of data.

The key: everyone is focused on large data sets (breadth of data). We have some of that, but not nearly as much as large processors. We have, however, much more depth (sometimes 10X-20X as much). This allows us to achieve high accuracy in short amounts of time. Often times we'll lose money in the beginning to ensure high approval rates and, in essence, pay for learning data.