Machine Learning
Deep Hedging Framework
Neural hedging policies trained against convex risk measures over simulated markets with transaction costs, accelerated by fused CUDA path kernels, plus a deep BSDE pricer.
Trains feedforward, recurrent, and no-trade-band policies by stochastic gradient descent on Rockafellar-Uryasev CVaR and entropic risk over simulated GBM, Heston, Merton-jump, and local-volatility markets under proportional transaction costs.
Generates paths on the fly with fused CUDA Philox kernels reaching several billion GBM paths per second, with whole-episode graph capture and a noise-regenerative backward that cuts peak training memory by 12.7 times.
At 40 basis points of cost the learned policy lowers expected shortfall by 21 percent against the Black-Scholes delta hedge, and a deep BSDE solver prices a 50-dimensional basket call within 1.3 percent of its closed form.