Update app.py

app.py

@@ -8,6 +8,7 @@ st.set_page_config(page_title="Chronos-Bolt Zero-Shot Forecast", layout="centere
 st.title("Chronos-Bolt Zero-Shot Forecast")
 st.caption("Zero-shot probabilistic forecasting (q10/q50/q90) using amazon/chronos-bolt-* models.")
 
+# -------------------- Indicator helpers (no pandas-ta needed) --------------------
 def ema(series, length=20):
     s = pd.Series(series).astype("float64")
     return s.ewm(span=length, adjust=False).mean()
@@ -65,7 +66,7 @@ def parse_pasted_series(txt: str):
     toks = re.split(r"[,\s]+", txt.strip())
     vals = []
     for t in toks:
-        if not t: 
+        if not t:
             continue
         try:
             vals.append(float(t))
@@ -158,8 +159,9 @@ if series is not None and series.size > 5:
             )
 else:
     st.info("Load a ticker, paste values, or upload a CSV to begin.")
-    # ================================
-# Train with RSI / EMA / Stochastic (AutoGluon)
+
+# ================================
+# Train with RSI / EMA / Stochastic (AutoGluon) — no pandas-ta
 # ================================
 with st.expander("Train with Indicators (RSI, EMA, Stochastic)"):
     st.write("Fine-tune Chronos-Bolt on one ticker using indicator covariates (past-only).")
@@ -176,7 +178,7 @@ with st.expander("Train with Indicators (RSI, EMA, Stochastic)"):
 
     if run_ft:
         with st.spinner("Downloading & computing indicators…"):
-            import yfinance as yf, pandas_ta as ta
+            import yfinance as yf
             from autogluon.timeseries import TimeSeriesPredictor, TimeSeriesDataFrame
 
             # 1) Load OHLC so we can compute Stochastic (needs High/Low/Close)
@@ -194,14 +196,10 @@ with st.expander("Train with Indicators (RSI, EMA, Stochastic)"):
             # Keep only needed cols and drop NaNs
             df = df[["Close", "High", "Low"]].dropna().copy()
 
-            # 2) Indicators
-            # RSI(14), EMA(20), Stochastic %K/%D (14,3,3)
-            df["rsi14"] = ta.rsi(df["Close"], length=14)
-            df["ema20"] = ta.ema(df["Close"], length=20)
-            stoch = ta.stoch(df["High"], df["Low"], df["Close"], k=14, d=3, smooth_k=3)
-            # pandas_ta returns columns like 'STOCHk_14_3_3' / 'STOCHd_14_3_3'
-            df["stoch_k"] = stoch.iloc[:, 0]
-            df["stoch_d"] = stoch.iloc[:, 1]
+            # 2) Indicators (helpers above)
+            df["rsi14"] = rsi(df["Close"], 14)
+            df["ema20"] = ema(df["Close"], 20)
+            df["stoch_k"], df["stoch_d"] = stochastic_kd(df["High"], df["Low"], df["Close"], 14, 3, 3)
 
             df = df.dropna().astype("float32")
             if df.shape[0] < 200:
@@ -246,7 +244,7 @@ with st.expander("Train with Indicators (RSI, EMA, Stochastic)"):
         med = ypred["0.5"].to_numpy()
         hi = ypred["0.9"].to_numpy()
 
-        import numpy as np, matplotlib.pyplot as plt
+        import matplotlib.pyplot as plt
         hx = np.arange(len(yhist))
         fx = np.arange(len(yhist), len(yhist) + len(med))
 
@@ -264,4 +262,3 @@ with st.expander("Train with Indicators (RSI, EMA, Stochastic)"):
             file_name=f"{item}_chronos_finetuned.csv",
             mime="text/csv",
         )
-