Error message "Column not found in access or evaluation of 'LnRV.SPY Recursive'

[#1]

dt = Open( "::Desktop:assignment10.jmp" );
window = 4321; //recursive window beginning 

rp = dt << New Column( "LnRV.SPY Recursive", numeric );

For( i = 0, i <= N Rows( dt ) - window - 1, i++,
	dt << clear row states();
	If( i,
		dt << Select Rows( 1 :: window + i ),
		dt << Select Rows( 1 :: window )
	);
	dt << Invert Row selection();
	dt << Exclude(); //exclude rows outside window
	
	fm = dt <<Fit Model( invisible(1),
	Y( :LnRV.SPY ),
	Effects(
		:L1RV.SPY, :L2RV.SPY, :L3RV.SPY, :L4RV.SPY, :L5RV.SPY, :L6RV.SPY, :L7RV.SPY,
		:L8RV.SPY, :L9RV.SPY, :L10RV.SPY, :L11RV.SPY, :L12RV.SPY, :L13RV.SPY, :L14RV.SPY,
		:L15RV.SPY, :L16RV.SPY, :L17RV.SPY, :L18RV.SPY, :L19RV.SPY, :L20RV.SPY, :L21RV.SPY,
		:L22RV.SPY
	),
	Personality( "Generalized Regression" ),
	Generalized Distribution( "Normal" ),
	Run(invisible( 1 ),
		
		Fit( Estimation Method( Lasso( Adaptive ) ), Validation Method( AICc ) )
	)
); //replace with your Fit Model script
	
	fm << (fit[1] << Save Prediction Formula);
	pred = column(dt, ncols(dt));
	rp[i + window + 1] = pred[i + window + 1];
	
	dt << delete columns( pred );
	fm << close window()
		
	);

dt << clear row states();

I am using JMP 17 Pro and can not seem to figure out this error. 

[#2]

Hm, the code seems to work.

Perhaps some of the columns in the Fit Model doesn't exist such that the fit platform doesn't start?
Without a fit platform, new column with a prediction formula. Then 

pred = column(dt, ncols(dt));
dt << delete columns( pred );

will delete LnRV.SPY Recursive.

You could try to remove the 

invisible(1),

and 

fm << close window()

to check the outcome of the fit ...

How many rows does your data table have?

[#3]

In addition to what hogi suggested you could also add a Show(i) inside your for loop to see when it errors out. If you remove your just created column it would most likely be second loop (i = 1). Also log might have some additional information than just that error message.