To give this a bit of perspective - I already posted about my running VO2max, LTHR and pace being a bit on the low side here https://forums.garmin.com/forum/into-sports/running/forerunner-935-aa/1283968-low-lthr-pace-on-fr935
Since that time there was a period when I virtually did not train at all, and when I restarted my running VO2max calculation sequence seemed to reboot itself, first jumping to around 49 but then gradually reducing to previously seen levels of around 46, so this seem to be a consistent result of Firstbeat algorithms application to my specific case. I also made a LTHR test in the meanwhile and am occasionally getting LTHR updates from normal training, with the result being again consistent 172bpm and around 5.50/km pace (all this despite my recent HM time of 1.56 (which is almost 2 hours at average pace of 5.33/km and heart rate in 176-186bpm range). To sum up, my 935 appears to consistently underestimate my running LTHR and pace, and most likely VO2max, though the latter is hard to say for sure.
Moving on to cycling, which is a bit new for me, I found 935 showing quite generous and constantly growing VO2max and FTP during last half year. Now (per Garmin/Firtsbeat) my cycling VO2max is 52 and my FTP is almost 200w (that's 2.9w/kg for me). In fact, my FTP is lower, I believe. My recent FTP test result in Zwift was around 186w and that looks much more plausible. The VO2max of 52, when applied to other activities (like running) is also an huge overestimate for me.
Long story short - I see my running physiological metrics (VO2max, LTHR and pace) being consistently underestimated while cycling metrics (VO2max and FTP) are, on the contrary, too high. To make things vorse, I see these deviating VO2max estimates messing up the training status algorithm which essentially tracks VO2max development, in the attached screenshots you may see the training status report from the "cycling" viewpoint (showing cycling VO2max consistently growing, which is at least consistent with what the watch says) and the "running" viewpoint (showing running VO2max all over the place due to high cycling VO2max values occasionally picked).
Do you guys also see any major deviations in your cycling and running VO2max estimates and is there any real world explanation for this (other than Firstbeat estimates being just, well, modelled estimates)?
If yes, do you also see deviating VO2max picked up inconsistently for Training Status purposes?