Secure Computation in Action: Real-World Deployments in Finance
summary
JP Morgan runs Prime Match and Atlas‑X in production: Prime Match performs encrypted inventory matching, Atlas‑X adds differential privacy to hide concentrated clients while keeping P&L loss around $1 per asset per day using 0.3 epsilon. Atlas‑X has produced billions in inventory across three regions for two years. Combining privacy and MPC yields a dark‑pool design that uses only commitments, inserts fake orders, achieves optimal matching, and reaches about 850 orders per second.
AI-written summary · about
chapters (9)
- 0:00Secure computation in finance
- 1:28Research to real-world at JPMorgan
- 3:58Prime Match in trading
- 9:50Differential privacy for clients
- 11:18Atlas-X deployment challenges
- 18:50Privacy versus adoption
- 19:35Dark pools and HFT constraints
- 22:02Indifferential privacy for dark pools
- 25:02Q&A
related talks
29mUpending Stock Market Structure Using Secure Computation
25mPractical Accountability of Secret Processes
37mPETs of the world unite! Conquer the middle ground in the performance vs...
40mConfidential Computing and Privacy-Enhancing Technologies - The Landscape
17mThe Privacy Paradox: Merging Confidentiality & Compliance in Finance
39mData Privacy Engineering to power Responsible Enterprise AI
45mConfidential Computing: Protecting Customer Data in the Cloud
24mSafeguarding Sensitive Data Access At Scale with Privacy-Enhancing Technologies