Machine Learning, Regression and Data Analysis

Hello to all,

Recently I have been exploring machine learning, linear regression models and data analysis in python for predicting forex prices...On paper it seems like a holy grail but it reality it isn't. Or at least maybe I am still a noob when it comes to machine learning and statistics...Take for example the r correlation coefficient of todays closing price to tomorrows closing price: 0.98. So in theory you should be able to predict tomorrows closing price pretty accuratly using todays closing price however this is not the case in reality. Does anyone know why?

Thanks

Cus forex prices are random and cannot be predicted!!

Maybe you can gain a small edge through seasonality, correlation and statistics.
But relaisticly you wont, it can be Part of a strategy But there is no hohly grail.
Data can be Bad and markets Go up, seasonality can be bearish like it should be end of february, markets still went up during that Time.

Currently its all about Inflation, a Weaker currency means assets adjust up, you need more currency to buy that asset (higher prices)

Either Inflation continius and markets Go up further or something is done to fight Inflation, causing disinflation or Deflation and markets would Fall.

Past data doesnt matter at this point, its about what happens to currency (usd) now in this unique end Game Type of Situation with the usd having lost 96% of its original value

"According to Ramsey theory, pure randomness (in the sense of there being no discernible pattern) is impossible, especially for large structures. Mathematician Theodore Motzkin suggested that "while disorder is more probable in general, complete disorder is impossible"." Source: wikipedia - randomness

Have you looked into Arima models? They supposedly take into consideration seasonality, noise, trend and supposedly this is the best method in forecasting markets as opposed to regression. Haven't dove deep into it myself yet...