Great video! Can we use the same approach for forecasting currency exchange rate price? By taking input as open high and low, and the target variable is close? will it be correct.
@@KDcodeyt thank you do you have an email where i can contact you about your video? I've got a question i couldn't find answer for and it would be helpful if i get your opinion on it?
This is like most ML videos on financial time series - it misses how financial people use models. The lowest MSE/MAPE or similar metric is not good enough. You should look at out of sample results for an equity curve using the model with a decision such as go long GLD (buy) if the model predicts tomorrows price > today's price, else go to cash and compare that to the equity curve of GLD. Hint: it will be a terrible equity curve because financial markets are very noisy. You need more features and a more complex model. Thank you.
Great tutorial to compare classical ML with NN!
Please subscribe for more videos :)
Great video! Can we use the same approach for forecasting currency exchange rate price? By taking input as open high and low, and the target variable is close? will it be correct.
Hi, yes you can try creating a model using past open, high and low data to predict the close value.
@@KDcodeyt thank you do you have an email where i can contact you about your video? I've got a question i couldn't find answer for and it would be helpful if i get your opinion on it?
Hi there, another great video of course.
I was wondering if you could perhaps make a web app video for this one?
Hi, will post a video on the web app verry soon
This is like most ML videos on financial time series - it misses how financial people use models. The lowest MSE/MAPE or similar metric is not good enough. You should look at out of sample results for an equity curve using the model with a decision such as go long GLD (buy) if the model predicts tomorrows price > today's price, else go to cash and compare that to the equity curve of GLD. Hint: it will be a terrible equity curve because financial markets are very noisy. You need more features and a more complex model. Thank you.
in my case extra_reg_train_score ==random_reg_train_score and extra_reg_test_score ==random_reg_test_score.
is this possible ?
Yes, it might be possible, you can try running with different parameter values to get best model out of them
please Subscribe for more videos :)