Upload 52 files

app.py

@@ -38,6 +38,7 @@ from kotak_neo import (
     KotakNeoSessionRequired,
     kotak_neo_manager,
 )
+from scraper import get_stock_info
 
 
 app = FastAPI(title="NIFTY 50 Forecaster Backend")
@@ -667,3 +668,12 @@ def data_refresh_market_close(
     session_date: date | None = Query(default=None, description="Optional YYYY-MM-DD session date in IST."),
 ) -> dict:
     return refresh_market_close_data(session_date=session_date)
+
+
+@app.get("/api/info/{ticker}")
+def stock_info(ticker: str) -> dict:
+    data = get_stock_info(ticker)
+    if "error" in data:
+        raise HTTPException(status_code=404, detail=data["error"])
+    return data
+

models/nifty_opening_mfe_regressor/outputs/latest_prediction.csv

@@ -1,2 +1,2 @@
 input_date,first5_start,first5_end,first5_close,predicted_up_points,predicted_down_points
-2026-06-09,2026-06-09 09:15:00,2026-06-09 09:19:00,23234.849609375,63.15162391627527,71.5788486914341
+2026-06-11,2026-06-11 09:15:00,2026-06-11 09:19:00,23112.650390625,97.5434972440442,76.48054546871592

models/nifty_opening_mfe_regressor/outputs/mfe_live_history.csv

@@ -0,0 +1,2 @@
+input_date,first5_start,first5_end,first5_close,predicted_up_points,predicted_down_points
+2026-06-11,2026-06-11 09:15:00,2026-06-11 09:19:00,23112.650390625,97.5434972440442,76.48054546871592

models/tomorrow_latest_prediction.csv

@@ -1,2 +1,2 @@
 input_date,target_date,prediction,prob_up,confidence,threshold,model_name,source_model,validation_accuracy,test_accuracy,source
-2026-06-11,2026-06-12,DOWN,0.46778826961945674,0.5322117303805433,0.534,nifty_tomorrow_direction_model,locked_multiwindow_nifty50_ensemble_v2,0.5886524822695035,0.7070707070707071,live
+2026-06-12,2026-06-15,UP,0.5568798000516237,0.5568798000516237,0.534,nifty_tomorrow_direction_model,locked_multiwindow_nifty50_ensemble_v2,0.5886524822695035,0.7070707070707071,live

models/tomorrow_prediction_history.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:89f100aced6b7ff9d8d0c5cd0ce10a00e5a402bbf935949fac103b8a33d0dbca
-size 8273
+oid sha256:ba49d81e54af07af6cd1b7defee306558ab4377dc09888b1d8452446aae13f4a
+size 8300

models/yahoo_history_cache.sqlite3

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:75c3f55a2adee670c10dbadd1c998d3ce773ed067c8e196ed2c7539ca6f8eb87
-size 397312
+oid sha256:8198956ffb6a3aca16eabd5f6fc4aa188ba816d324a7f9987caf631d1b0eea14
+size 421888

nifty_backend/__pycache__/runtime.cpython-311.pyc

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eac2a9d465d264f6ebb982a4f2b7fbd61ef4db81c5a3549543b07cb103b8d680
-size 138158
+oid sha256:83ba07597a5f92396b6297f2fe401f732a2a617e82c6fd3311f2d3e5537475e5
+size 143804

nifty_backend/runtime.py

@@ -60,6 +60,8 @@ MFE_OUTPUT_DIR = MODEL_DIR / "nifty_opening_mfe_regressor" / "outputs"
 MFE_SUMMARY_PATH = MFE_OUTPUT_DIR / "summary.json"
 MFE_LATEST_PATH = MFE_OUTPUT_DIR / "latest_prediction.csv"
 MFE_TEST_PREDICTIONS_PATH = MFE_OUTPUT_DIR / "test_predictions.csv"
+MFE_MODEL_PATH = MFE_OUTPUT_DIR / "nifty_opening_mfe_regressor.joblib"
+MFE_LIVE_HISTORY_PATH = MFE_OUTPUT_DIR / "mfe_live_history.csv"
 TPLUS1_MODEL_PATH = MODEL_DIR / "nifty_1420_tplus1_logistic_model.joblib"
 TPLUS1_LATEST_PATH = MODEL_DIR / "tplus1_latest_prediction.csv"
 TPLUS1_SUMMARY_PATH = MODEL_DIR / "tplus1_summary.json"
@@ -945,6 +947,80 @@ def latest_tplus1_prediction() -> dict[str, Any]:
     }
 
 
+def load_mfe_summary() -> dict[str, Any]:
+    if MFE_SUMMARY_PATH.exists():
+        return json.loads(MFE_SUMMARY_PATH.read_text(encoding="utf-8"))
+    return {}
+
+
+def latest_mfe_prediction() -> dict[str, Any]:
+    if MFE_LATEST_PATH.exists():
+        row = pd.read_csv(MFE_LATEST_PATH).iloc[-1].to_dict()
+        return {k: (None if pd.isna(v) else v) for k, v in row.items()}
+    summary = load_mfe_summary()
+    return {
+        "input_date": summary.get("latest_input_date"),
+        "first5_start": summary.get("latest_first5_start"),
+        "first5_end": summary.get("latest_first5_end"),
+        "predicted_up_points": summary.get("latest_predicted_up_points"),
+        "predicted_down_points": summary.get("latest_predicted_down_points"),
+    }
+
+
+def refresh_mfe_prediction(session_date: date | None = None) -> dict[str, Any]:
+    if not MFE_MODEL_PATH.exists():
+        return {}
+    payload = joblib.load(MFE_MODEL_PATH)
+    up_model = payload["up_model"]
+    down_model = payload["down_model"]
+    up_features = payload["up_features"]
+    down_features = payload["down_features"]
+    up_calib = payload["up_calibration"]
+    down_calib = payload["down_calibration"]
+
+    dataset = pd.read_parquet(OPENING_DATASET_PATH)
+    dataset["_session_date"] = pd.to_datetime(dataset["date"], errors="coerce").dt.normalize()
+    if session_date is not None:
+        latest_df = dataset[dataset["_session_date"].dt.date == session_date].tail(1)
+    else:
+        latest_df = dataset.tail(1)
+    
+    if latest_df.empty:
+        return {}
+    
+    raw_up = float(np.clip(up_model.predict(latest_df[up_features])[0], 0.0, None))
+    pred_up = float(np.clip((raw_up * up_calib["scale"]) + up_calib["offset"], 0.0, None))
+    
+    raw_down = float(np.clip(down_model.predict(latest_df[down_features])[0], 0.0, None))
+    pred_down = float(np.clip((raw_down * down_calib["scale"]) + down_calib["offset"], 0.0, None))
+    
+    out = {
+        "input_date": latest_df["_session_date"].dt.date.iloc[0].isoformat(),
+        "first5_start": str(latest_df["first5_start"].iloc[0]),
+        "first5_end": str(latest_df["first5_end"].iloc[0]),
+        "first5_close": float(latest_df["first5_close"].iloc[0]),
+        "predicted_up_points": pred_up,
+        "predicted_down_points": pred_down,
+    }
+    pd.DataFrame([out]).to_csv(MFE_LATEST_PATH, index=False)
+    
+    # Append to live history
+    live_df = pd.DataFrame([out])
+    if MFE_LIVE_HISTORY_PATH.exists():
+        try:
+            existing = pd.read_csv(MFE_LIVE_HISTORY_PATH)
+            # Avoid duplicates if refreshed multiple times in the same session
+            existing = existing[existing["input_date"] != out["input_date"]]
+            pd.concat([existing, live_df], ignore_index=True).to_csv(MFE_LIVE_HISTORY_PATH, index=False)
+        except Exception:
+            live_df.to_csv(MFE_LIVE_HISTORY_PATH, index=False)
+    else:
+        live_df.to_csv(MFE_LIVE_HISTORY_PATH, index=False)
+        
+    clear_dashboard_payload_cache()
+    return out
+
+
 def _minute_frame_for_tplus1() -> pd.DataFrame:
     minute = pd.read_parquet(NIFTY_1M_PATH)
     minute = minute.copy()
@@ -1267,6 +1343,9 @@ def dashboard_payload() -> dict[str, Any]:
         _file_cache_key(NIFTY_1M_PATH),
         _file_cache_key(LIVE_ACCURACY_PATH),
         _file_cache_key(TOMORROW_PREDICTION_HISTORY_PATH),
+        _file_cache_key(MFE_SUMMARY_PATH),
+        _file_cache_key(MFE_LATEST_PATH),
+        _file_cache_key(MFE_MODEL_PATH),
     )
     with _dashboard_payload_lock:
         return copy.deepcopy(_dashboard_payload_cached(key))
@@ -1442,6 +1521,8 @@ def _dashboard_payload_cached(key: tuple[tuple[str, int | None, int | None], ...
     tomorrow_test = load_tomorrow_test_predictions()
     tomorrow_history = _load_prediction_history(TOMORROW_PREDICTION_HISTORY_PATH)
     tplus1_test = load_tplus1_test_predictions()
+    mfe_summary = load_mfe_summary()
+    mfe_latest = latest_mfe_prediction()
     daily = pd.read_parquet(NIFTY_1D_PATH)
     daily["date"] = pd.to_datetime(daily["date"], errors="coerce")
     daily = daily.sort_values("date").tail(180)
@@ -1521,6 +1602,10 @@ def _dashboard_payload_cached(key: tuple[tuple[str, int | None, int | None], ...
                 "latest": tplus1_latest,
                 "summary": tplus1_summary,
             },
+            "mfe": {
+                "latest": mfe_latest,
+                "summary": mfe_summary,
+            },
         },
         "metrics": metrics,
         "models": model_metrics,
@@ -1551,6 +1636,10 @@ def refresh_first5_prediction(session_date: date | None = None, minutes: pd.Data
     merged = merged.drop_duplicates(subset=["date"], keep="last").sort_values("date").reset_index(drop=True)
     merged.to_parquet(OPENING_DATASET_PATH, index=False, compression="zstd")
     prediction = predict_row(row)
+    try:
+        refresh_mfe_prediction(session_date=session_date)
+    except Exception as exc:
+        print(f"MFE refresh failed: {exc}", flush=True)
     return prediction
 
 

scraper.py

@@ -0,0 +1,97 @@
+import requests
+from bs4 import BeautifulSoup
+
+def get_stock_info(ticker):
+    url = f"https://www.screener.in/company/{ticker.upper()}/consolidated/"
+    headers = {
+        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/115.0.0.0 Safari/537.36"
+    }
+    
+    response = requests.get(url, headers=headers)
+    if response.status_code == 404:
+        url = f"https://www.screener.in/company/{ticker.upper()}/"
+        response = requests.get(url, headers=headers)
+    
+    if response.status_code != 200:
+        return {"error": f"Failed to fetch data for {ticker.upper()}. Status {response.status_code}"}
+    
+    soup = BeautifulSoup(response.text, 'html.parser')
+    
+    data = {
+        "ticker": ticker.upper(),
+        "key_metrics": {},
+        "pros": [],
+        "cons": [],
+        "growth": [],
+        "history": {}
+    }
+    
+    # 1. Key Metrics
+    top_ratios = soup.find('ul', id='top-ratios')
+    if top_ratios:
+        for li in top_ratios.find_all('li'):
+            n_span = li.find('span', class_='name')
+            v_span = li.find('span', class_='value')
+            if n_span and v_span:
+                name = n_span.text.strip().replace('₹', 'Rs.')
+                val = ' '.join(v_span.text.split()).replace('₹', 'Rs.')
+                data["key_metrics"][name] = val
+                
+    # 2. Pros/Cons
+    analysis = soup.find('section', id='analysis')
+    if analysis:
+        pros = analysis.find('div', class_='pros')
+        if pros:
+            data["pros"] = [li.text.strip() for li in pros.find_all('li')]
+        cons = analysis.find('div', class_='cons')
+        if cons:
+            data["cons"] = [li.text.strip() for li in cons.find_all('li')]
+            
+    # 3. Growth Metrics
+    ranges = soup.find_all('table', class_='ranges-table')
+    for table in ranges:
+        th = table.find('th')
+        if not th: continue
+        metric = th.text.strip()
+        for tr in table.find_all('tr')[1:]:
+            tds = tr.find_all('td')
+            if len(tds) == 2:
+                data["growth"].append({
+                    "Metric": metric,
+                    "Period": tds[0].text.strip(),
+                    "Value": tds[1].text.strip()
+                })
+                
+    # 4. Tables
+    sections = {
+        'quarters': 'Quarterly Results',
+        'profit-loss': 'Profit & Loss',
+        'balance-sheet': 'Balance Sheet',
+        'cash-flow': 'Cash Flows',
+        'ratios': 'Financial Ratios',
+        'shareholding': 'Shareholding Pattern'
+    }
+    for sec_id, sec_name in sections.items():
+        sec = soup.find('section', id=sec_id)
+        if not sec: continue
+        tbl = sec.find('table')
+        if not tbl: continue
+        thead = tbl.find('thead')
+        headers = [th.text.strip().replace('₹', 'Rs.') for th in thead.find_all('th')] if thead else []
+        rows = []
+        for tr in tbl.find('tbody').find_all('tr'):
+            cols = [td.text.strip().replace('₹', 'Rs.') for td in tr.find_all('td')]
+            rname = tr.find('td', class_='text')
+            if rname:
+                cols[0] = rname.text.replace('+', '').strip()
+            if cols:
+                rows.append(cols)
+        if headers and len(headers) == len(rows[0]) - 1:
+            headers.insert(0, 'Metric')
+        data["history"][sec_id] = {
+            "title": sec_name,
+            "headers": headers,
+            "rows": rows
+        }
+        
+    return data