An applied, hands-on course on the ML that runs modern financial services — credit scoring, fraud detection, market risk and Value-at-Risk, financial time-series, and anti-money-laundering. Real public datasets, real in-browser math, two case studies on how a fintech actually does it, and a capstone where you build and run a systematic strategy.
Default rate by credit-score band — everything left of the cutoff is declined.
A real trade-off from the course. Drag the cutoff.
In finance the model is the easy half. The decision is the hard half.
Approve the loan or decline it. Block the card or let it through. File the suspicious-activity report or don't. This course teaches the models — and the costs, thresholds, and trade-offs that turn a probability into a decision you can defend.
Build a probability-of-default scorecard with weight-of-evidence features and calibration, then a fraud detector under extreme class imbalance with cost-sensitive thresholds.
Value-at-Risk and Expected Shortfall, portfolio optimization and the efficient frontier, algorithmic trading and honest backtesting, and options pricing with the Greeks.
Anti-money-laundering on a transaction graph, the false-positive problem, and two case studies: thin-file lending at a neobank and real-time scoring at a payments company.
Every unit is live, with 133 interactive lessons across 12 units — each built on a real public dataset with real in-browser math, and a capstone that ties them all together.
The course assumes you're comfortable with logistic regression, classification metrics, and the idea of training and evaluation. If those are new, start with Finguard ML — it leads straight into this.
Who it's for
No account, no install. Progress saves automatically in your browser, separate from your other courses.