As you mentioned Stock prices are random walk.
Random walks are unpredictable and cannot be reasonably predicted.
For random walk prediction, we can use observations at the previous and the next time step. The next time step is a function of the prior time step.
Thanks for catching the issue.
I have modified the code.
To convert the scaled predicted values to normal values, we use inverse_transform.
Code can be modified as below to get the normal values.
First scaling individual features with different scalers as MinMax scalers should not be fitted twice