Sensex Backtesting for Free in Python (^BSESN Walkthrough)

#nse #stocks #india #trading

Sensex Backtesting for Free in Python (^BSESN Walkthrough)

By Shakti Tiwari — Nifty Option Trader, Research Analyst & XGBoost Expert

The BSE Sensex is India's oldest benchmark, and you can backtest strategies on it for free — no Bloomberg terminal, no paid data vendor. Yahoo Finance carries the index as ^BSESN, and with yfinance plus pandas you can build an honest backtest in under 60 lines. This walkthrough is code-first and India-specific, and it flags the traps that make Sensex backtests lie.

Step 1: Pull Free Sensex Data with the Right Ticker

pip install yfinance pandas numpy

The correct symbol for the S&P BSE Sensex on Yahoo is ^BSESN (not SENSEX). Get it wrong and you'll download nothing or the wrong series.

import yfinance as yf
import pandas as pd
import numpy as np

df = yf.download("^BSESN", start="2010-01-01", end="2024-12-31", auto_adjust=True)
df = df.dropna()
print(df.head(), df.tail(), sep="\n\n")
print(f"Total trading days: {len(df)}")

Sensex history on Yahoo goes back well over a decade — enough to cover multiple market regimes (2013 taper tantrum, 2020 COVID crash, 2021 bull run), which is exactly what you want a strategy to survive.

Step 2: Define a Strategy You Can Verify by Hand

We'll use RSI mean-reversion — buy when the index is oversold, exit when it recovers. Simple enough to sanity-check, and it behaves differently from a trend strategy, so it's a good second tool in your kit.

def rsi(series, period=14):
    delta = series.diff()
    gain = delta.clip(lower=0).rolling(period).mean()
    loss = (-delta.clip(upper=0)).rolling(period).mean()
    rs = gain / (loss + 1e-9)
    return 100 - (100 / (1 + rs))

df["rsi"] = rsi(df["Close"], 14)
df["signal"] = np.where(df["rsi"] < 30, 1, np.where(df["rsi"] > 55, 0, np.nan))
df["signal"] = df["signal"].ffill().fillna(0)

The ffill holds the position between an entry (RSI < 30) and an exit (RSI > 55) — a common, realistic way to model a mean-reversion trade.

Step 3: Prevent Lookahead — Shift the Position

This is where most free backtests go wrong. You act on a signal only on the next bar, never the same bar it was generated.

df["position"] = df["signal"].shift(1)
df["mkt_ret"] = df["Close"].pct_change()
df["strat_ret"] = df["position"] * df["mkt_ret"]
df = df.dropna()

If you skip the .shift(1), your RSI strategy will look like it perfectly bought every bottom — because it's cheating with future data.

Step 4: Subtract Real Costs

Every entry and exit costs money: brokerage, STT, exchange fees, and slippage. Charge them on each position change.

COST = 0.0005   # ~5 bps round-trip proxy; set to your Zerodha/Angel reality
df["trade"] = df["position"].diff().abs()
df["strat_net"] = df["strat_ret"] - df["trade"] * COST

Step 5: Equity Curve and Honest Metrics

df["equity"] = (1 + df["strat_net"]).cumprod()
df["buyhold"] = (1 + df["mkt_ret"]).cumprod()

def metrics(r, periods=252):
    r = r.dropna()
    cagr = (1 + r).prod() ** (periods / len(r)) - 1
    sharpe = np.sqrt(periods) * r.mean() / (r.std() + 1e-9)
    eq = (1 + r).cumprod()
    maxdd = (eq / eq.cummax() - 1).min()
    return {"CAGR": round(cagr,4), "Sharpe": round(sharpe,2), "MaxDD": round(maxdd,4)}

print("Sensex RSI strategy:", metrics(df["strat_net"]))
print("Sensex buy & hold:  ", metrics(df["mkt_ret"]))

I won't quote a specific CAGR — your window and cost assumptions change it, and a fixed number would be a fabricated stat. Run the code and read your output.

Sensex vs Nifty: Practical Differences

The Sensex has 30 stocks; the Nifty 50 has 50. Sensex is more concentrated in large-cap heavyweights.
Yahoo's ^BSESN and ^NSEI are highly correlated — a strategy that works on one usually behaves similarly on the other. Test both to check robustness.
There is no liquid Sensex options market comparable to Nifty. Sensex derivatives exist but Nifty F&O is far deeper, so most Indian derivative traders execute on Nifty even if they research on Sensex.

Where Free Data Ends

Yahoo ^BSESN is end-of-day and can have occasional missing sessions. Always dropna() and check date continuity.
Intraday Sensex history on Yahoo is very limited. For minute-level BSE data use Angel One SmartAPI (free, generous history) or Zerodha Kite Connect (paid). Both give exchange-grade candles.

Pitfalls Checklist

Wrong ticker (SENSEX instead of ^BSESN) → empty download.
Missing .shift(1) → lookahead, inflated results.
Zero-cost assumption → mean-reversion strategies especially over-trade and die once costs are real.
Testing a single date range → validate across regimes or use walk-forward analysis.

Takeaway

Backtest the Sensex for free using ^BSESN with yfinance + pandas.
Get the ticker right, dropna(), and always .shift(1) your position.
Net out realistic Indian costs; RSI mean-reversion is cost-sensitive.
Judge with Sharpe and max drawdown, never final return alone.
For intraday BSE data beyond Yahoo's limits, use Angel SmartAPI or Zerodha Kite.

Related: My book Option Trading with AI: XGBoost, Transformers & Quantized Models for the Retail Nifty Trader shows retail traders how to research and validate strategies with free tools.

Disclaimer: This article is for education only and is not financial, investment, or trading advice. SEBI-registered research rules apply — verify everything before acting.

🔗 Get the book on Amazon: https://www.amazon.in/dp/B0H9ZNTBPK

📢 Daily Nifty analysis on Telegram: https://t.me/shaktitrade

📧 Email: shaktitiwari715@gmail.com