Pandas Basics
DataFrames & Series

When you download NIFTY historical data, you get thousands of rows — dates, open, high, low, close, volume. Pandas is the standard Python library for handling exactly this kind of tabular data. It's fast, expressive, and purpose-built for financial time series.

import pandas as pd

# Install: pip install pandas

# Create a DataFrame from NIFTY candle data
data = {
    "date":   ["2024-01-15", "2024-01-16", "2024-01-17", "2024-01-18"],
    "open":   [21450, 21580, 21320, 21410],
    "high":   [21620, 21710, 21490, 21560],
    "low":    [21390, 21500, 21280, 21350],
    "close":  [21580, 21340, 21460, 21510],
    "volume": [142500, 168700, 129300, 155100]
}

df = pd.DataFrame(data)
print(df)

Series — Single Column Data

When you select a single column from a DataFrame, you get a Series — a one-dimensional labeled array. This is what you'll work with most when calculating indicators.

# A Series is just one column
close = df["close"]
print(type(close))   # <class 'pandas.core.series.Series'>
print(close)

# Basic stats on the close Series
print(f"Mean  : {close.mean():.2f}")
print(f"Max   : {close.max()}")
print(f"Min   : {close.min()}")
print(f"Std   : {close.std():.2f}")

Arithmetic on a Series

# Calculate candle body size (vectorised — no loop needed!)
df["body"] = (df["close"] - df["open"]).abs()

# Calculate upper & lower wicks
df["upper_wick"] = df["high"] - df[["open", "close"]].max(axis=1)
df["lower_wick"] = df[["open", "close"]].min(axis=1) - df["low"]

# Candle direction
df["bullish"] = df["close"] > df["open"]

print(df[["date", "body", "upper_wick", "lower_wick", "bullish"]])

Selecting & Slicing Data

Pandas gives you multiple ways to select data. In trading code you'll use these constantly — filtering signals, pulling last N candles, or accessing specific dates.

# Last candle (most recent)
last = df.iloc[-1]
print(f"Last close: {last['close']}")

# Last 3 candles
recent = df.iloc[-3:]
print(recent[["date", "close"]])

# Filter: only bullish candles
bullish_candles = df[df["bullish"] == True]
print(f"Bullish candles: {len(bullish_candles)}")

# Filter: candles with volume above 150,000
high_volume = df[df["volume"] > 150000]
print(high_volume[["date", "volume"]])

DateTime Index — The Key to Market Data

Real market data always has timestamps. Setting the date column as the DataFrame index unlocks time-based slicing, resampling, and alignment — essential for multi-timeframe analysis.

# Build a proper datetime-indexed DataFrame
data_ts = {
    "datetime": pd.date_range("2024-01-15 09:15", periods=6, freq="5min"),
    "open":     [21450, 21480, 21510, 21490, 21530, 21555],
    "high":     [21495, 21520, 21540, 21520, 21570, 21580],
    "low":      [21430, 21465, 21490, 21475, 21520, 21540],
    "close":    [21480, 21510, 21495, 21530, 21555, 21570],
    "volume":   [12400, 9800, 11200, 8700, 13100, 10500]
}

df5 = pd.DataFrame(data_ts)
df5 = df5.set_index("datetime")   # Set datetime as index

print(df5)
print(f"\nIndex type: {type(df5.index)}")

# Access by time range
morning = df5.between_time("09:15", "09:25")
print(morning["close"])

Resampling Timeframes

# Convert 5-min candles to 15-min candles (OHLCV aggregation)
df15 = df5.resample("15min").agg({
    "open":   "first",
    "high":   "max",
    "low":    "min",
    "close":  "last",
    "volume": "sum"
})

print(df15)

Adding Columns — Signal Engine Pattern

In a real trading system, you'll keep adding computed columns to your DataFrame — moving averages, signals, P&L. This is the core DataFrame enrichment pattern.

import pandas as pd

# Simulate 10 NIFTY daily candles
closes = [21450, 21580, 21320, 21410, 21550,
          21620, 21480, 21390, 21510, 21640]

df = pd.DataFrame({"close": closes})

# 1. Simple Moving Average (3-period)
df["sma3"] = df["close"].rolling(3).mean().round(2)

# 2. Percentage change candle-to-candle
df["pct_change"] = df["close"].pct_change() * 100

# 3. Signal: close above SMA = BUY, else SELL
df["signal"] = df.apply(
    lambda row: "BUY" if row["close"] > row["sma3"] else "SELL",
    axis=1
)

print(df.tail(5).to_string())

Essential DataFrame Methods

These methods come up in nearly every algo trading script. Know them by heart.

# Common pre-trade data checks
print(f"Rows loaded    : {df.shape[0]}")
print(f"Columns        : {list(df.columns)}")
print(f"NaN in close   : {df['close'].isnull().sum()}")
print(f"Date range     : {df.index[0]} → {df.index[-1]}")

# Drop NaN rows (important after adding rolling indicators)
df_clean = df.dropna()
print(f"Clean rows     : {len(df_clean)}")

Quiz — Test Your Knowledge

Exercises

Lesson Summary