Optimize Stop-Loss Policies with Dynamic Programming at a Quant Fund
Overview
What this challenge is about.
You receive five years of daily PnL series for 12 momentum strategies plus a small set of state features (rolling vol, drawdown, regime indicator). Calibrate a discrete Markov model over states, define a stop-loss action set (hold, scale down, exit), and solve with backward induction. Run a back-test comparing the DP policy to fixed stops on Sharpe ratio, maximum drawdown, and tail-loss frequency. Deliverable is the calibrated model, the solver, the back-test report, and a risk-committee memo with explicit limits.
The Brief
What you'll do, and what you'll demonstrate.
Beat the current fixed-percentage stop-loss policy on Sharpe and max drawdown using a state-dependent DP policy in back-test.
Earning criteria — what you'll demonstrate
- Calibrate a discrete Markov model from financial time series
- Implement backward induction for finite-horizon decision problems
- Run an honest back-test (out-of-sample, no peek-ahead)
- Communicate model risk to a risk-committee audience
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Applied AI Scientist
DP-based risk-policy design with honest back-testing is the kind of quantitative work applied AI scientists do at funds and fintechs.
This challenge sharpens
- dynamic-programming
- back-testing
- risk-modeling
Data Scientist
State-model calibration on financial time series is a foundational data-scientist skill on quant teams.
This challenge sharpens
- state-modeling
- back-testing
- python
ML Researcher
Finite-horizon DP is the on-ramp to deeper RL research; this challenge proves the student can ship the simpler primitive.
This challenge sharpens
- dynamic-programming
- backward-induction
- state-modeling