I'm currently developing a model based on the current macroeconomic scenario in the world to predict the price of copper 1, 2 and 3 months ahead. That's my idea and I'd like to know what are your thoughts:
My dataset is composed by Copper, which is the variable I'd like to forecast and roughly 30 macroeconomic indicators, like USA CPI, China PMI, COT and so on. Data are monthly and go from 2010-01-01 up to today. Via Web Data Scraping and several API's I obtained the historical and real time macroeconomic data, so that when there's a new release, my dataset is updated.
From that, I computed the correlation of the returns between each macrovariable and Copper returns, defining the most significant the ones which in absolute value are greater than 0.3. (This threshold it's my choice, do you have any other suggestion?)
Thus, after defining the sample of significant macrovariables, for each of the macrovariable I checked what happened in January, i.e if the variable accelerated or delecerated, and I retrieved all the past years in which this same acceleration or deceleration happened in January (Acceleration > 0, Deceleration < 0). I applied the same process for every month of the 2024 for which, up to today, these Rate of Change are available. Important: I took into account that if a variable can have negative values, if it goes from -0.3% to -0.6% is an increase of 100%, which is the same as if it goes from 1% to 2%, so I double checked the acceleration and the sign for that month.
Once found all the years, I basically counted the number of times a year is occurring and divided this counter by the number of variables for which the data are available, finding the percentage of similarity with a certain year.
Once the best year is find( in terms of similarity) I then compute the mean and the standard deviation at 1,2 and 3 months and use this data to simulate a MonteCarlo simulation and retrieve all the data I need.
Surely there are several flows. For examples, concerning inflation, an increase from 0,5% to 1% is 100%, but it has not the same impact on the market and the economy as if the inflation goes from 2% to 4%. But that's something I might work on. Do you have any idea? If you want to contact me I can also provide the code, which is really huge.
