deephedging_experiments

Deep Hedging Experiments

Multi-Agent Environment

• bsagent: Greek Hedging Agent (Single agent environment)

  1. bsm: without market impact
  2. bsm_mi: with market impact

• mean_var: RL Hedging Agent trained with mean-var objective function (market impact)

  1. sa: single agent environment (trained agent RL1)
  2. ma1: greek hedging vs. RL1
  3. ma2: greek hedging vs. RL hedging (trained agent RL2)
  4. ma3: RL1 vs. RL2
  5. ma4: greek hedging vs. RL1 vs. RL2

• var: RL Hedging Agent trained with variance objective function (market impact)

    tests same as mean_var