🐍 Lesson 5 of 7

Code the statistical models

🕒 About 30 minutes 🐍 pandas + numpy 🧩 Drag-to-order puzzle

Time to run the three statistical models from Lesson 3 as real Python: the moving average, the linear regression trend, and ARIMA. It is only a handful of lines each. Press Run on every block to see the output on NCB's real prices, then build one yourself by dragging code into order.

💡 You do not have to install anything

These examples are made for Google Colab, a free website that runs Python in your browser. There is a full Colab walkthrough in the Week 2 course. Here, the Run buttons show you exactly what each block prints, so you can read the code first and run it for real later.

Step 1: put the prices in a table

pandas gives you a DataFrame, a spreadsheet you steer with code. We load NCB's twenty closing prices and peek at the last few rows.

🐍 load.py

import pandas as pd

# NCB Financial Group (NCBFG) daily closes, Jamaica Stock Exchange
prices = [52.50, 57.15, 58.00, 60.00, 58.99, 58.95, 58.50, 60.00,
          61.00, 60.77, 61.00, 62.99, 62.98, 63.00, 65.00, 66.45,
          70.00, 74.95, 75.35, 74.50]
ncb = pd.DataFrame({"close": prices})
print(ncb.tail())

Output

Statistical model 1: the moving average

The naive forecast (tomorrow equals today) and the moving average are one line each with pandas.

🐍 moving_average.py

# Naive: tomorrow = today (the last close)
naive = ncb["close"].iloc[-1]

# Moving average: the mean of the last 3 closes
ma3 = ncb["close"].tail(3).mean()

print("Naive forecast:      J$", round(naive, 2))
print("Moving-avg forecast: J$", round(ma3, 2))

Output

Naive forecast:      J$ 74.5
Moving-avg forecast: J$ 74.93

Statistical model 2: the linear regression trend

numpy fits a straight trend line in one call with polyfit, then we stretch it out to forecast Tuesday, two trading days after the last close.

🐍 trend.py

import numpy as np

days = np.arange(len(ncb))            # 0, 1, 2, ... 19
slope, start = np.polyfit(days, ncb["close"], 1)

tuesday = start + slope * 21          # day 21 = the next Tuesday
print("Trend rises by J$", round(slope, 2), "per day")
print("Trend forecast for Tuesday: J$", round(tuesday, 2))

Output

Trend rises by J$ 0.98 per day
Trend forecast for Tuesday: J$ 74.39

Statistical model 3: ARIMA

ARIMA is the professional time-series workhorse. The statsmodels library fits it in two lines and forecasts the next few days. The three numbers in order tell it how much recent history and recent error to lean on.

🐍 arima.py

from statsmodels.tsa.arima.model import ARIMA

model = ARIMA(ncb["close"], order=(1, 1, 1)).fit()
forecast = model.forecast(2)          # Monday and Tuesday
print(forecast.round(2))

Output

20    74.62
21    74.71
Name: predicted_mean, dtype: float64

Outputs shown here are pre-generated so you can read along without a Python setup. Notice the three statistical models land close together, near J$74 to J$75, which is a good sign.

Your turn: build the moving average by hand

Drag the six code lines into the right order so the program computes the average of NCB's last four closes and prints it. Watch your indentation: the line inside the loop is pushed in.

Code lines (drag from here)

Your program (drop in order)

Getting real stock data into Colab

The prices above were typed straight into the code, which is perfect for learning. For a bigger project, you have two easy paths:

📄Upload a CSVDownload prices from the JSE site, then read_csv it

→

🔌Use a data libraryTools like yfinance pull prices in one call

Reading a file you uploaded is just pd.read_csv("ncb.csv"). From there, every line above works the same, whether you have twenty days or twenty years.

🧠 Recap

pandas holds prices in a DataFrame you shape with code.
The moving average is one line; the trend is one np.polyfit.
ARIMA from statsmodels forecasts the next days in two lines.
All three statistical models are quick, need little data, and land in the same neighbourhood.

Next: the same idea for the machine learning models, with a chart in Python.