Moving Averages with Pandas

From Price Data to Market Insights

Raw price data is just a sequence of numbers. To make sense of it, we create indicators. One of the most fundamental and widely used indicators is the moving average. Its purpose is to smooth out price fluctuations and help identify the underlying trend of the market. Think of it as reducing the "noise" of candle-to-candle volatility to see the bigger picture.

We will focus on two main types:

1. Simple Moving Average (SMA): The average price over a specific number of periods. Every data point in the window is weighted equally.
2. Exponential Moving Average (EMA): A type of moving average that gives more weight to recent prices, making it more responsive to new information and reducing lag.

For a trader, the choice between SMA and EMA often comes down to a trade-off between smoothness (SMA) and responsiveness (EMA).

Calculating a Simple Moving Average (SMA)

The SMA is the most straightforward type of moving average. To calculate a 10-period SMA, you simply sum the closing prices of the last 10 candles and divide by 10.

In pandas, this is achieved with the .rolling() method. This method creates a "rolling window" of a specified size that moves over your data, allowing you to perform a calculation on the data within that window at each step.

To see this in action, let's watch a video that demonstrates the process.

As you saw, the syntax is clean and intuitive. Here is the complete pattern applied to a familiar OHLCV DataFrame:

import pandas as pd
import yfinance as yf # We'll use yfinance to get some real data

# Let's fetch some recent hourly data for EUR/USD
# Note: You may need to run 'pip install yfinance' in your terminal
data = yf.download('EURUSD=X', period='10d', interval='1h')

# --- SMA Calculation ---
# We want to calculate a 20-period SMA on the 'Close' price.
# 1. Select the 'Close' price column: data['Close']
# 2. Create a rolling window of size 20: .rolling(window=20)
# 3. Calculate the mean of each window: .mean()
sma_20 = data['Close'].rolling(window=20).mean()

# Now, let's add this back to our DataFrame as a new column
data['SMA20'] = sma_20

# Display the last few rows to see our new column
# Notice the first 19 rows of 'SMA20' will be NaN (Not a Number)
# because you need 20 periods of data to calculate the first SMA.
print(data.tail())

Calculating an Exponential Moving Average (EMA)

The EMA is often preferred in trading because it reacts more quickly to recent price changes. The calculation is more complex, but the key idea is that the most recent price gets the most weight, and the weight of older prices decays exponentially.

Fortunately, pandas handles this complexity for us with the .ewm() method, which stands for "exponentially weighted moving." Instead of a window, its key parameter is span, which is conceptually similar to the window size of an SMA.

This article provides a good textual explanation of the EMA and its implementation.

The adjust=False parameter mentioned in the article is often used to match the calculation method used by many trading platforms.

Let's add a 50-period EMA to our DataFrame:

# --- EMA Calculation ---
# To calculate a 50-period EMA on the 'Close' price.
# 1. Select the 'Close' price column: data['Close']
# 2. Apply the exponentially weighted function with a span of 50: .ewm(span=50, adjust=False)
# 3. Calculate the mean: .mean()
ema_50 = data['Close'].ewm(span=50, adjust=False).mean()

# Add it to our DataFrame
data['EMA50'] = ema_50

# Display the tail again to see both MAs
print(data.tail())

data['SMA200'] = data['Close'].rolling(window=200).mean()

Practical Application: Building a MACD Indicator

To solidify your understanding, let's build a real technical indicator from scratch using the skills you just learned. The Moving Average Convergence Divergence (MACD) is a popular momentum indicator built entirely from EMAs.

It consists of three parts:

1. MACD Line: The 12-period EMA minus the 26-period EMA.
2. Signal Line: A 9-period EMA of the MACD Line itself.
3. Histogram: The MACD Line minus the Signal Line (we'll cover this when we get to visualization).

This is a perfect exercise, as it requires you to calculate multiple EMAs and then perform arithmetic between them. The following video walks through exactly how to do this.

Let's implement this on our own data DataFrame.

# --- MACD Calculation ---

# 1. Calculate the 12-period and 26-period EMAs
ema_12 = data['Close'].ewm(span=12, adjust=False).mean()
ema_26 = data['Close'].ewm(span=26, adjust=False).mean()

# 2. Calculate the MACD Line
data['MACD'] = ema_12 - ema_26

# 3. Calculate the Signal Line (a 9-period EMA of the MACD line)
data['Signal_Line'] = data['MACD'].ewm(span=9, adjust=False).mean()

# Display the final DataFrame with all our calculated indicators
print("DataFrame with SMA, EMA, and MACD components:")
print(data[['Close', 'SMA20', 'EMA50', 'MACD', 'Signal_Line']].tail(10))

By following these steps, you've gone from raw price data to a DataFrame enriched with multiple, standard technical indicators. This is a massive step forward in your ability to programmatically analyze market data.

Conclusion

In this lesson, you've bridged the gap from data preparation to data analysis. You now have the foundational skill to calculate any rolling or expanding indicator, which is a cornerstone of building automated trading strategies.

Key Takeaways:

• A moving average smooths price data to help identify trends.
• Use .rolling(window).mean() to calculate a Simple Moving Average (SMA).
• Use .ewm(span=N).mean() to calculate an Exponential Moving Average (EMA).
• EMAs are generally more responsive to recent price changes than SMAs.
• You can combine these simple calculations to build more complex indicators like the MACD.

Preview for the Next Lesson:
Looking at a table of numbers is useful, but a chart is worth a thousand data points. In our next lesson, "Create basic price visualizations using a plotting library," we will bring our analysis to life. You will learn how to plot the price data and overlay the moving averages and MACD lines you calculated today, creating the exact kind of charts you're used to seeing on your trading platform.