I have noted the stabilisation is not used during active probe measuring. Another thing I have noted is that it may happen to have put a pattern before the measure. Worthess to add that precision in such a measurements is quite important as it reflects on probe correction on measures, profiles and caracterisation. I am also sensitive beacues I calibrate mainly OLED's

In order have some assessment of that I have preparared a patch file, with 3 Black, 3 R, 3 G, 3 B, 3 W, 3W (for test), 3 R, 3 G, 3 B, 3 W repeated another time, but introducing before each reading the reading of a Black pattern. The Black pattern emulates a stabilisation for a time equal to the integration time.
I have done all that on my E8 and my I1 display pro. Integration time is set to .75 sec.
Here what I have measured for at around 100 and 200 Nits.


What it can be seen (red marking) is that white insertion introduce less variation on XYZ if stabilisation is applied. So somehow the results are less dipending on the display history.
On the right part of the picture I have assessed for the two patterns (No white insertion, white insertion) the differences between the cases without and with stabilitation. Also there stabilisation introduce some benefits.
Interesting/good that the standard deviation stays alsways white low

PS: it is not difficult, starting fom XLM format of LS profiles, use Excel to elaborate them. Basically you might take my CSV file, run it, and, with my excel file, have the outcomes by a simple import. Also based on file the generation of a bpd is quite easy. This may be of interes if for same reason you want averaging on a value greater than 3.
If somebody is interested, ask for it, and I will post a link to my files with a little of explanation.

This is one of beauties of LightSpace - flexibility.
You are not restricted to set workflows for any process, and can combine ideas and alternative approaches as wanted.

Steve

Yes Steve, I fully agree and I have followed on that:
I have created a patch sequence done by 3 x 5 WRGB readings. I can easily assess the results just by importing its outcome in Excel and automatically generating average, deviance, xyY in the format needed for being exported to .bpd file.
I could achieve that setting a stabilitation with black of .75 seconds and adding a pre-patch sequence done by the points used for characterising a 3 cube (Only replaced black with white).

Then I started generating some corrections. The same template I have used for assessing and generating a .bpd file has been used also as input for checking the outcomes of the correction. I have virtually generated two XYZ matrixes: one at 100 Nits with a rec709 Color space and another at 200 Nits with a rec2020 One (accepting negative Z)
Here their assessments based at (100 Nits, rec709), (100 Nits, rec2020), (200 Nits, rec709), (200 Nits, rec2020):


I have to tell that when I have generated the correction on rec709 I was surprised about how well it fits 200 Nits at rec709, and this is the reason I have generated a correction based on higher luminance. Starting from a rec2020 color space looks definitely as good compromise for everything.