THIS IS AN ONGOING PROJECT

I’ve been interested in investing in the stock market lately. I started this projects to help me learn more about valuing companies and deciding what stocks to invest in. This is a short project where I analyse Nvidia stocks using past information, and the project answers some important questions like:

• What was the change in price of stock over time?
• What is the daily and monthly return of the stock?
• What is the stock’s moving average?
• How much risk do we take by investing in this stock?
• Can we predict stock behavior with statistical models?

The last question bears the most interest to me. I used an ARIMA model as I wanted to compare its performance between a statistical model and an AI model (in my other project).

The ARIMA model

ARIMA models only work with stationary time-series data. So it’s important to check for that first. If the data isn’t stationary, multiple iterations of differencing might be needed:

{{ result = adfuller(nvidia_data.Close) }}

Once we’re sure that our data is suitable for the model, we can build the model using the following:

{{ for time_point in range(N_test_obs):
model = sm.tsa.arima.ARIMA(history, order = (4,1,0))
model_fit = model.fit()
out = model_fit.forecast()
y_pred = out[0]
model_preds.append(y_pred)
y_orig = test_data[time_point]
history.append(y_orig) }}

View the full project repository here