diff --git a/22_deep_reinforcement_learning/trading_env.py b/22_deep_reinforcement_learning/trading_env.py
index 6d4a05371..e0d0b394e 100644
--- a/22_deep_reinforcement_learning/trading_env.py
+++ b/22_deep_reinforcement_learning/trading_env.py
@@ -176,7 +176,7 @@ def take_step(self, action, market_return):
         trade_costs = abs(n_trades) * self.trading_cost_bps
         time_cost = 0 if n_trades else self.time_cost_bps
         self.costs[self.step] = trade_costs + time_cost
-        reward = start_position * market_return - self.costs[self.step]
+        reward = start_position * market_return - self.costs[max(0, self.step-1)]
         self.strategy_returns[self.step] = reward
 
         if self.step != 0: