The Foundations of Risk: Stock Risk Analysis with Descriptive Statistics
Before you start building complex machine learning models, you need to understand the data you’re actually working with. In Chapter 7 of Data Analytics for Finance Using Python, we go back to the basics: Descriptive Statistics.
Think of descriptive statistics as the “vibe check” for your data. It tells you if your stock is consistent, volatile, or just plain weird.
The Big Three: Mean, Median, and Mode
The authors looked at MRF stock prices over 250 days and found some interesting things:
- Mean (Average): 100,703
- Median (Middle): 100,773
- Mode (Most Frequent): 108,500
When the mode is higher than the mean and median, it’s a good sign for investors. It means the stock hits those higher price points more often than the average would suggest.
Measuring the Mess: Standard Deviation and Range
This is where risk assessment really happens.
- Range: 49,700 (The gap between the highest and lowest price).
- Standard Deviation: 11,169.
Here’s the kicker: because the standard deviation is relatively small compared to the range, the stock is actually pretty consistent. It doesn’t just jump around randomly; it follows a somewhat predictable spread.
The “Shape” of the Risk: Skewness and Kurtosis
This is some high-level stuff that’s actually super useful:
- Skewness (0.24): This is positive skewness. In plain English? The stock might give small losses in the short term, but it has a tendency for good returns in the long term. The book suggests a “Hold” strategy here.
- Kurtosis (-0.50): Negative kurtosis means the “tails” of the distribution are thin. You’re less likely to see extreme, bank-breaking outliers. This is great for “boring” investors who hate surprises.
Why it matters
You can’t manage risk if you can’t measure it. By looking at these basic numbers, you can decide if a stock fits your risk tolerance before you ever write a single line of predictive code.
And that’s why it matters. Sometimes the simplest tools are the most revealing.
Next: Stock Prediction with Multiple Regression | Previous: Investment Management with Decision Trees