Is your feature request related to a problem? Please describe.
I have a target variable that is the difference between x_t and x_t+500. This target variable is represented as y_t. I want to set up a model that looks like this:
Describe the solution you'd like
I want to support forecasters that can do 2 things:
Make a prediction like a regular ML model would and works in our backtesting framework. (A time series prediction that is not exactly forecasting).
Can use lagged y values from at least n timestamps ago (in this example n=500 because that is the most recent y we would have known at time t).
Describe alternatives you've considered
You could do a bunch of lags and stuff. Definitely goes against user friendliness.
Additional context
I am not sure if this is technically forecasting... But it is time series prediction and I feel like it is actually a better use case than forecasting. It will also allow us to set up time series classification and generally be more flexible. It should fit into the backtesting framework nicely.
Is your feature request related to a problem? Please describe. I have a target variable that is the difference between x_t and x_t+500. This target variable is represented as y_t. I want to set up a model that looks like this:
skew*B_0 + imbalance*B_1 = y_t(price_t - price_t+500)
Describe the solution you'd like I want to support forecasters that can do 2 things:
Describe alternatives you've considered You could do a bunch of lags and stuff. Definitely goes against user friendliness.
Additional context I am not sure if this is technically forecasting... But it is time series prediction and I feel like it is actually a better use case than forecasting. It will also allow us to set up time series classification and generally be more flexible. It should fit into the backtesting framework nicely.