OBVSlopeN is calculated as:

The OBV is calculated, using the past 100 values, for the given offset. This resuls in a list of OBV values like:
OBV(0)
OBV(1)
OBV(2)
OBV(3)
...

The linear regression is calculated for the past N-OBV values:

OBVSlope = LinearRegression(OBV(0) .. OBV(N-1))

The normalized rate of change of the OBVSlope is calculated as:

OBVSlopeN = OBVSlope * 100 / Abs(OBV(0))

The OBVSlope gives the rate of change of N-values of OBV. If the OBVSlope is 500, then the OBV is increasing at a rate of 500 per bar. Similarly, a OBVSlope of -500 would indicate that the line that best fits N-values of the OBV is falling at the rate of 500 per bar.

When comparing two instruments, one that's trading 1M shares/day and another that's trading at 50K share, then it's not fair to compare the two rates of change of the OBV (OBVSlope). OBVSlopeN is the normalized OBVSlope which is the percent change of the OBV per bar of the regression (best fit) line. OBVSlopeN normalizes the data for the sake of comparison among other OBVSlopes that trade in different ranges.