Algo Strategy: Mean Reversion – Buy the Dip, Sell the Spike

You’ve heard it before:

“Buy low, sell high.” “Stocks always return to their average.”

That’s the philosophy behind mean reversion.

When a stock moves too far from its usual price range, algo traders assume it’ll come back (or "revert") to the mean (average). So they buy the dip or sell the spike.

What is Mean Reversion?

Imagine a stock usually trades between ₹90 and ₹110.

One day, due to panic, it drops to ₹80.

A mean reversion algo will buy at ₹80, expecting it to bounce back toward ₹100.

It assumes the price move was an overreaction — not a new trend.

Common Indicators Used

Indicators to detect mean reversion opportunities:

• Bollinger Bands
  These bands expand and contract based on volatility.
  If price touches the lower band, it may be oversold – a buy signal.
  If it touches the upper band, it may be overbought – a sell signal.
  

import pandas as pd

def calculate_bollinger_bands(data, window=20, num_std=2): """ Calculate Bollinger Bands.

Parameters:
    data    : DataFrame with 'close' prices
    window  : Moving average period
    num_std : Number of standard deviations

Returns:
    DataFrame with columns: MA, Upper Band, Lower Band
"""
data['MA'] = data['close'].rolling(window=window).mean()
data['STD'] = data['close'].rolling(window=window).std()
data['Upper'] = data['MA'] + (data['STD'] * num_std)
data['Lower'] = data['MA'] - (data['STD'] * num_std)
return data[['MA', 'Upper', 'Lower']]

#Example usage: #df = calculate_bollinger_bands(price_data)

• RSI (Relative Strength Index)

RSI < 30 = Oversold (Buy)
RSI > 70 = Overbought (Sell)

def calculate_rsi(data, period=14): """ Calculate Relative Strength Index (RSI).

Parameters:
    data   : DataFrame with 'close' prices
    period : Lookback period (default = 14 days)

Returns:
    Series containing RSI values
"""
delta = data['close'].diff()  # Price changes

#Separate gains and losses
gain = delta.where(delta > 0, 0)
loss = -delta.where(delta < 0, 0)

#Use Exponential Moving Average (smoother than simple average)
avg_gain = gain.ewm(span=period, adjust=False).mean()
avg_loss = loss.ewm(span=period, adjust=False).mean()

#Relative Strength (RS)
rs = avg_gain / avg_loss

#RSI Formula
rsi = 100 - (100 / (1 + rs))
return rsi

• Moving Average Pullbacks

If price falls 5-10% below its moving average, the algo might trigger a buy.

Let’s say:

• Stock's 20-day moving average = ₹100
• Current price = ₹88
• RSI = 25

Your algo says, “This is too cheap!”
It buys the stock and exits when price hits ₹98 or RSI crosses 50.

import pandas as pd

def calculate_moving_average(data, window=20): """ Calculate Simple Moving Average (SMA).

Parameters:
    data   : DataFrame with a 'close' column
    window : Lookback period (default = 20)

Returns:
    Series with SMA values
"""
return data['close'].rolling(window=window).mean()

#Example usage: #price_data['SMA_20'] = calculate_moving_average(price_data, window=20)

Why Mean Reversion Works

• Markets overreact to news, panic, or short-term hype
  
• Prices tend to return to fair value
  
• It works well in sideways or range-bound markets
  

Risks & Precautions

• Trending markets can kill this strategy – the stock may keep falling!
  
• Never assume a dip = reversal. Always use:
  
• Stoploss
  
• Position sizing
  
• Time-based exit (e.g., exit if no reversal in 3 days)
  

Tips to Make It Safer

• Combine multiple indicators (e.g., RSI + Bollinger)
  
• Avoid trading during news events (like RBI policy or earnings)
  
• Limit exposure to a few trades at a time
  

So, momentum = ride the wave
Mean reversion = catch the bounce

Both have their place. Algos help you remove emotion and follow logic.

Next up: A very interesting concept – Pairs Trading, where you don’t care about market direction at all!