Position in the walk-forward sequence, oldest first.
First index of the optimization window, into the prepared session.
How that winner scored on the window it was chosen on.
End of the optimization window, exclusive — and where scoring begins.
End of the scoring window, exclusive.
The parameter vector that won this fold's optimization window.
Vectors this fold evaluated in-sample before picking its winner.
What one fold concluded. The out-of-sample row is the answer; the in-sample figure is only there to be compared against it, since any grid produces a flattering in-sample winner — that is what optimizing does. The gap between them is the whole reading.