The Auto Stock Forecasting Application
Introduction
This standalone Python-based application is able to perform live extraction of stock data from Yahoo Finance, and from a set of predefined models, computes the daily forecast predictions for the months ahead. In the current build of the public version of the application, the program uses Facebookâs fbprophet Python library for forecasting. The amount of historical data the program takes in is at most 2 years back, from the current date. It will then proceed to perform forecasts for each stock, for a full year ahead. An example of the output for the stock FB is as follows.
User Instructions
Downloading the Application
The latest version of the program can be download via this link. You should be able to observe the following:
Download the desired version of the program (see version updates below). Once downloaded, unpack the archive and you should see the following folders inside the dist/main folder:
Running the Program
Scroll down in the main folder till you arrive at main.exe:
Double click on it to open (it may take over 20 seconds to boot up initially), and a window should appear:
Navigating the Program
There are instructions at each step of the program and it should be easy to follow. For simplicity, we illustrate the running of the program with performing a live extraction for the six default stocks, and compute their forecasts. So at step 1, we select y and the program proceeds to extract the data (internet connection required) for the six stocks:
Once all the data have been extracted, the model computes the forecast for each stock in the sequence it was extracted. An example of the program moving on to the next stock is shown below:
Finally, when all stocks chosen initially have their forecasts computed, the program terminates, as shown below:
The Output Files - Forecast Plots & Forecasted Data
When the program terminates, to find your forecasts for each stock in the form of a plot (daily), the forecast components and the more detailed forecasted data, navigate to the folder results in main/results:
Recall that if you have extracted stock data previously, they can be found in the data folder via main/data:
Hope the guide helps!
Version Updates
v1.2 (updated 03/10/2019): Added 17 models, which take into account holidays, weekends, seasonality (daily, weekly, monthly, yearly) and mode of seasonality (additive/multiplicative). The program first performs cross-validation on each model and uses the model with the lowest root mean square error (RMSE) obtained during cross-validation as the final model for computing forecasts.
Donate
If you will like to make contributions to the site (completely obligatory), click here.
You will also have the option of listing your profile under the list of contributors on the donation page.