Utilize input predictions for Supervised Learning

[donnertrud]

Hi guys,

Currently I am working on a paper in order to research the usefulness of ML in order to predict Forex Exchange Rates using Supervised Learning techniques. So far, my model is based on monthly predictions utilizing economic indicators like GDP, Inflation or Unemployment.
Now, there exist forecasts for each of the indicators listed above, e.g. it is predicted that the GDP is going to change by 4,2% next month due to certain events. This information is obviously useless for training the model, but can it be used for predictions in the future? As far as I know, some traders rely on upcoming political / economic events in order to improve their chanes to make profit. If it is possbile, how can I achieve that?

[hussainmujtaba]

Usually, in your case. Two factors are really important, viz Technical Analysis and Public mood.The public mood definitely depends on upcoming political/economic events. One way to go forward is to use twitter to know about the latest news that can affect the Forex exchange rates. You need a sentiment classifier that can classify tweets about the financial market. You also need to cluster the tweets into various categories such as tweets related to customer service are clustered separately from the tweets that actually affect the exchange rates such as financial news.
Then you can use sentiment analyser on the tweets related to finance. Next step is to integrate it with your technical analysis.There are various approaches in that too.
You can either use the sentiment along with the result from the technical analyser to train a different model.But there are other ways too.

[donnertrud]

Thanks for your response!

What you describe makes sense, however, I want to focus on fundamental analysis due to various reasons. I would like to use an economic calender (https://tradingeconomics.com/calendar) that shows previous and future events and their impact. I just need an approach to include predictions of a certain feature in the final prediction, if that makes sense.