Inspiration
We wanted to solve OLG's challenge of detecting fraud and risky gambling behavior across online casinos, sportsbooks, and land-based casinos. Our goal was to make player protection smarter, more transparent, and easier for investigators.
What it does
Sentinel X is a cross-channel fraud detection and player protection platform. It analyzes player activity, identifies suspicious patterns, assigns separate AML/fraud and safer-gambling risk scores, and provides explainable alerts.
It also includes investigation tools, network analysis, fairness evaluation, and adversarial testing.
How we built it
We used Python, Streamlit, pandas, NumPy, NetworkX, Plotly, and scikit-learn.
We developed synthetic player datasets, engineered behavioral risk indicators, built an interpretable scoring system, and created an interactive dashboard with investigation workflows.
Challenges we ran into
Our biggest challenges were combining incomplete data from different gaming channels, handling uncertain player identities, reducing false alerts, and making complex risk scores understandable.
Accomplishments that we're proud of
We built a functional prototype analyzing 2,000 synthetic players and 19,214 transactions across three gaming channels.
We're especially proud of our explainable risk assessments, Player 360 investigation workspace, human-review workflow, and Red vs Blue testing environment.
What we learned
We learned that effective fraud detection requires more than identifying unusual transactions. Data quality, behavioral context, explainability, and human oversight are equally important.
What's next for ThreeFlow-Sentinel-X-OLG
We plan to integrate machine learning models, enable real-time monitoring, improve cross-channel identity matching, and expand fairness testing.
Our long-term vision is to make Sentinel X a scalable, responsible player protection and fraud investigation platform.
Built With
- numpy
- python
- scikit-learn
- streamlit
Log in or sign up for Devpost to join the conversation.