Fix: Track actual data source (SEC vs Yahoo) for fundamentals metrics

Backend: _extract_metrics_from_raw_data() now tracks which source
(sec_edgar or yahoo_finance) each metric came from via new data_source field.

Frontend: inferDataSource() uses explicit data_source if available,
with fallback to form-based inference for backward compatibility.

This ensures SEC EDGAR data displays correctly even when form field is missing.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

frontend/src/components/MCPDataPanel.tsx

@@ -97,10 +97,14 @@ function formatValue(value: string | number): string {
 }
 
 // Infer data source from category and metric
-function inferDataSource(category: string, metric: string, form?: string): string {
+function inferDataSource(category: string, metric: string, form?: string, dataSource?: string): string {
   const lowerMetric = metric.toLowerCase()
 
   if (category === 'fundamentals') {
+    // Use explicit data_source if provided, otherwise fall back to form-based inference
+    if (dataSource === 'sec_edgar') return 'SEC EDGAR'
+    if (dataSource === 'yahoo_finance') return 'Yahoo Finance'
+    // Legacy fallback: infer from form field
     return form ? 'SEC EDGAR' : 'Yahoo Finance'
   }
   if (category === 'valuation') return 'Yahoo Finance'
@@ -161,6 +165,7 @@ export function MCPDataPanel({ metrics, rawData, companyName, ticker, exchange,
       fiscalPeriod?: string | null
       endDate?: string
       form?: string
+      dataSource?: string
     }>> = {
       fundamentals: [],
       valuation: [],
@@ -180,7 +185,8 @@ export function MCPDataPanel({ metrics, rawData, companyName, ticker, exchange,
           value: m.value,
           fiscalPeriod,
           endDate: m.end_date,
-          form: m.form
+          form: m.form,
+          dataSource: m.data_source
         })
       }
     }
@@ -207,7 +213,7 @@ export function MCPDataPanel({ metrics, rawData, companyName, ticker, exchange,
           value: formatValue(m.value),
           dataType: inferDataType(m.form, m.metric),
           asOf: m.endDate || '-',
-          source: inferDataSource(cat, m.metric, m.form),
+          source: inferDataSource(cat, m.metric, m.form, m.dataSource),
           category: cat.charAt(0).toUpperCase() + cat.slice(1)
         })
       }

frontend/src/lib/api.ts

@@ -34,6 +34,7 @@ export interface MetricEntry {
   end_date?: string       // Fiscal period end date, e.g., "2023-09-30"
   fiscal_year?: number    // Fiscal year number, e.g., 2023
   form?: string           // SEC form type: "10-K" (annual) or "10-Q" (quarterly)
+  data_source?: string    // Actual data source: "sec_edgar" or "yahoo_finance"
 }
 
 // MCP status for each server (partial = some data but with errors)

src/services/workflow_store.py

@@ -130,31 +130,46 @@ def _extract_metrics_from_raw_data(raw_data: dict) -> list:
 
     for metric_name in fin_metrics:
         # Prefer SEC EDGAR data, fall back to Yahoo Finance
-        metric_data = sec_data.get(metric_name) or yf_fund.get(metric_name)
-        if metric_data is not None:
-            entry = {
-                "timestamp": timestamp,
-                "source": "fundamentals",
-                "metric": metric_name,
-            }
-            if isinstance(metric_data, dict):
-                entry["value"] = metric_data.get("value")
-                if metric_data.get("end_date"):
-                    entry["end_date"] = metric_data["end_date"]
-                if metric_data.get("fiscal_year"):
-                    entry["fiscal_year"] = metric_data["fiscal_year"]
-                if metric_data.get("form"):
-                    entry["form"] = metric_data["form"]
-            else:
-                entry["value"] = metric_data
+        # Track which source the data actually came from
+        sec_metric = sec_data.get(metric_name)
+        yf_metric = yf_fund.get(metric_name)
+
+        if sec_metric is not None:
+            metric_data = sec_metric
+            actual_source = "sec_edgar"
+        elif yf_metric is not None:
+            metric_data = yf_metric
+            actual_source = "yahoo_finance"
+        else:
+            continue
 
-            if entry.get("value") is not None:
-                metrics.append(entry)
+        entry = {
+            "timestamp": timestamp,
+            "source": "fundamentals",
+            "metric": metric_name,
+            "data_source": actual_source,  # Track actual data source for frontend
+        }
+        if isinstance(metric_data, dict):
+            entry["value"] = metric_data.get("value")
+            if metric_data.get("end_date"):
+                entry["end_date"] = metric_data["end_date"]
+            if metric_data.get("fiscal_year"):
+                entry["fiscal_year"] = metric_data["fiscal_year"]
+            if metric_data.get("form"):
+                entry["form"] = metric_data["form"]
+        else:
+            entry["value"] = metric_data
+
+        if entry.get("value") is not None:
+            metrics.append(entry)
 
     # Extract valuation metrics from Yahoo Finance
     val_all = multi_source.get("valuation_all", {})
     yf_val = val_all.get("yahoo_finance", {}).get("data", {})
 
+    # Get valuation fetch date if available (point-in-time data)
+    val_fetch_date = yf_val.get("_fetch_date") or yf_val.get("fetch_date")
+
     val_metrics = [
         "market_cap", "enterprise_value", "trailing_pe", "forward_pe",
         "pb_ratio", "ps_ratio", "trailing_peg", "price_to_fcf",
@@ -164,14 +179,27 @@ def _extract_metrics_from_raw_data(raw_data: dict) -> list:
     for metric_name in val_metrics:
         metric_data = yf_val.get(metric_name)
         if metric_data is not None:
-            value = metric_data.get("value") if isinstance(metric_data, dict) else metric_data
-            if value is not None:
-                metrics.append({
-                    "timestamp": timestamp,
-                    "source": "valuation",
-                    "metric": metric_name,
-                    "value": value
-                })
+            entry = {
+                "timestamp": timestamp,
+                "source": "valuation",
+                "metric": metric_name,
+            }
+            if isinstance(metric_data, dict):
+                entry["value"] = metric_data.get("value")
+                # Extract date if available in metric data
+                if metric_data.get("date"):
+                    entry["end_date"] = metric_data["date"]
+                elif metric_data.get("end_date"):
+                    entry["end_date"] = metric_data["end_date"]
+                elif val_fetch_date:
+                    entry["end_date"] = val_fetch_date
+            else:
+                entry["value"] = metric_data
+                if val_fetch_date:
+                    entry["end_date"] = val_fetch_date
+
+            if entry.get("value") is not None:
+                metrics.append(entry)
 
     # Extract volatility metrics
     vol_all = multi_source.get("volatility_all", {})
@@ -193,17 +221,32 @@ def _extract_metrics_from_raw_data(raw_data: dict) -> list:
             metrics.append(entry)
 
     # Beta and volatility from Yahoo Finance
+    # Get volatility fetch date if available
+    vol_fetch_date = yf_vol.get("_fetch_date") or yf_vol.get("fetch_date")
+
     for vol_metric in ["beta", "historical_volatility", "implied_volatility"]:
         metric_data = yf_vol.get(vol_metric)
         if metric_data is not None:
-            value = metric_data.get("value") if isinstance(metric_data, dict) else metric_data
-            if value is not None:
-                metrics.append({
-                    "timestamp": timestamp,
-                    "source": "volatility",
-                    "metric": vol_metric,
-                    "value": value
-                })
+            entry = {
+                "timestamp": timestamp,
+                "source": "volatility",
+                "metric": vol_metric,
+            }
+            if isinstance(metric_data, dict):
+                entry["value"] = metric_data.get("value")
+                if metric_data.get("date"):
+                    entry["end_date"] = metric_data["date"]
+                elif metric_data.get("end_date"):
+                    entry["end_date"] = metric_data["end_date"]
+                elif vol_fetch_date:
+                    entry["end_date"] = vol_fetch_date
+            else:
+                entry["value"] = metric_data
+                if vol_fetch_date:
+                    entry["end_date"] = vol_fetch_date
+
+            if entry.get("value") is not None:
+                metrics.append(entry)
 
     # Extract macro indicators
     macro_all = multi_source.get("macro_all", {})