update changes

task1_grader.py

@@ -1,10 +1,20 @@
-def grade_task1(agent_action, observation):
+def grade_task1(action, observation):
     """
     Task 1: Fetch Market Data
-    Returns score strictly between 0 and 1
+    Returns score based on whether price was retrieved
     """
-    # Simple scoring logic
-    score = 0.75
+    score = 0.75  # Base score
+    
+    # Check if observation has a price
+    if observation and observation.get('price', 0) > 0:
+        score = 0.95
+    else:
+        score = 0.55
+    
+    # Ensure score is never 0.0 or 1.0
+    if score <= 0.0:
+        score = 0.01
+    if score >= 1.0:
+        score = 0.99
     
-    # You can add actual logic later, but this passes validation
     return score

task2_grader.py

@@ -1,7 +1,25 @@
-def grade_task2(agent_action, observation):
+def grade_task2(action, observation):
     """
     Task 2: News Sentiment Analysis
-    Returns score strictly between 0 and 1
+    Returns score based on explanation quality
     """
     score = 0.75
+    
+    # Check if agent provided explanation
+    if action and action.get('explanation'):
+        explanation = action.get('explanation', '')
+        if len(explanation) > 20:
+            score = 0.95
+        elif len(explanation) > 10:
+            score = 0.85
+        else:
+            score = 0.65
+    else:
+        score = 0.55
+    
+    if score <= 0.0:
+        score = 0.01
+    if score >= 1.0:
+        score = 0.99
+    
     return score

task3_grader.py

@@ -1,7 +1,25 @@
-def grade_task3(agent_action, observation):
+def grade_task3(action, observation):
     """
     Task 3: Backtest Strategy
-    Returns score strictly between 0 and 1
+    Returns score based on backtest results
     """
     score = 0.75
+    
+    # Check if backtest results exist
+    if observation and observation.get('backtest_results'):
+        results = observation.get('backtest_results', {})
+        if results.get('sharpe_ratio', 0) > 1.0:
+            score = 0.95
+        elif results.get('sharpe_ratio', 0) > 0.5:
+            score = 0.85
+        else:
+            score = 0.65
+    else:
+        score = 0.55
+    
+    if score <= 0.0:
+        score = 0.01
+    if score >= 1.0:
+        score = 0.99
+    
     return score

server/environment.py

@@ -27,6 +27,7 @@ class TradingEnvironment:
         self.shares = 0
         self.total_steps = len(self.prices)
         self.tasks_completed = []
+        self.task_scores = {}  # Track scores for each task
         return self._get_observation()
     
     def step(self, action: AgentAction):
@@ -34,6 +35,14 @@ class TradingEnvironment:
         self.idx = min(self.idx + 1, self.total_steps - 1)
         price = self.prices[self.idx]
         
+        # Track which task is being attempted
+        if action.type == "GET_PRICE":
+            self._complete_task("task1", 0.85)
+        elif action.type == "GET_NEWS" or (action.explanation and len(action.explanation) > 5):
+            self._complete_task("task2", 0.85)
+        elif action.type == "BACKTEST":
+            self._complete_task("task3", 0.85)
+        
         if action.type == "BUY" and action.amount:
             cost = price * action.amount
             if cost <= self.cash:
@@ -48,6 +57,12 @@ class TradingEnvironment:
         
         return self._get_observation()
     
+    def _complete_task(self, task_id: str, score: float):
+        """Mark a task as completed with a score"""
+        if task_id not in self.tasks_completed:
+            self.tasks_completed.append(task_id)
+            self.task_scores[task_id] = score
+    
     def _get_observation(self):
         price = self.prices[self.idx]
         news_idx = self.idx % len(self.news)
@@ -74,5 +89,10 @@ class TradingEnvironment:
             "current_step": self.idx,
             "total_steps": self.total_steps,
             "observation": self._get_observation().dict(),
-            "tasks_completed": self.tasks_completed
+            "tasks_completed": self.tasks_completed,
+            "task_scores": self.task_scores
         }
+    
+    def get_task_score(self, task_id: str) -> float:
+        """Return score for a specific task (for grader integration)"""
+        return self.task_scores.get(task_id, 0.75)