I hope you take my post as a honest analysis of the data presented in those papers and how the admixture results you are showing do not definitively confirm the PIE is North of the Caucuses nor does it confirm that the PIE is South of the Caucus mountains.
It does show that PIE is not from Armenia. Even Wikipedia says literally on the third line at the top of the page:
"The most widely accepted proposal about the location of the Proto-Indo-European homeland is the steppe hypothesis."
https://en.wikipedia.org/wiki/Proto-Indo-European_homeland
Take Figure 4 for example from Lazaridis et al 2016. It clearly shows admixture from Iran Neolithic going North and EHG from North of the Steppes going South. But here is the question, what is the direction of causality so to speak. In a regression framework, it could be that the Y (Dependent variable) is South Caucus and the X (independent variable) is Steppe admixture. In this context, is the Steppe admixture with EHG moving south of the Caucuses bringing PIE language into Armenian and Iran, etc. Alternatively, if the Y (Dependent variable) is North of the Caucuses (Steppe) and the X is CHG admixture, etc. In this context, is is the CHG moving North and perhaps bringing PIE language to the Steppe. So what you have in this situation is a classic endogeneity problem in terms of the direction of causality, i.e. what is the Y and what is the X variable. So when Professor Reich in his 2018 book "How we Are..." talked about the Armenian Highlands are a plausible source of where PIE languages originated, I think the results in Figure 4 suggest that that is a plausible hypothesis. Just to be clear, this is not the Anatolian Hypothesis related to ancient Hittites being the source of PIE.
The northward migration was mainly of CHG type, not Iran_N (although this is present in a very small percentage). I also tend to rule out that PIE was brought by CHG-rich people, populations with the largest share of CHG do not even speak Indo-European. PIE cannot be from Armenia for the simple reason that for example Yamnaya and Corded Ware in the north are not descended from Areni and do not have its share of West Asian ancestry.
Target: ARM_Areni_C
Distance: 3.4176% / 0.03417649
39.6 TUR_Boncuklu_N:ZHAG_BON004___BC_7950
21.6 GEO_CHG:KK1___BC_7728
18.2 IRN_Ganj_Dareh_N:I1954___BC_8212
7.6 RUS_AfontovaGora3:AfontovaGora3___BC_16086
6.8 Levant_PPNC:I1699___BC_6750
4.4 Levant_PPNB:BAJ001___BC_6900
1.8 RUS_Veretye_Meso

ES001___BC_10761
Target: Yamnaya_RUS_Samara
Distance: 5.9363% / 0.05936292
34.2 GEO_CHG:KK1___BC_7728
28.0 RUS_Sidelkino_HG:Sidelkino___BC_9371
25.0 UKR_Meso:I1763___BC_8131
5.8 RUS_AfontovaGora3:AfontovaGora3___BC_16086
3.6 IRN_Ganj_Dareh_N:I1954___BC_8212
3.4 TUR_Boncuklu_N:ZMOJ_BON014___BC_7950
Target: Corded_Ware
Distance: 4.9219% / 0.04921941
42.2 UKR_Meso:I1763___BC_8131
24.6 TUR_Boncuklu_N:ZHAG_BON004___BC_7950
20.2 GEO_CHG:KK1___BC_7728
10.4 RUS_Sidelkino_HG:Sidelkino___BC_9371
2.6 IRN_Ganj_Dareh_N:I1954___BC_8212
So not sure what I wrote above makes any sense but right now at best your models show is an association/correlation in that yes Steppe Admixture is in those Ancient Armenian samples. But there is CHG in those Steppe_EMBA and Steppe_MLBA samples as well. Again, an association. But in my view those admixture results do not show one way or the other where the PIE homeland is.
Hopefully, the new Reich Team paper will be able to sort that out with this new samples. Again, hopefully my post is seen as a honest attempt to discuss civilly what is for many a very, very, very, contested question (PIE homeland). And as I have said numerous times, while I find the PIE an very interesting research question, I have no emotional feelings one way or the other where the PIE homeland is (North or South of the Caucus mountains).
Cheers, PT
I am of the opinion that many people do not want to accept the Steppe "hypothesis" because for some strange reason they link it to Nordicism or racism, but still...
What could Francisco in Portugal and Ram in India possibly have in common? Could it be a coincidence that they both speak Indo-European languages? Or that they both have haplogroup R1? ... and that they both have Steppe-related admixture, will that also be a coincidence?
Target: Portuguese
Distance: 0.7079% / 0.00707900 | R5P
37.0 IRL_MN:bally___BC_3181
21.8 Corded_Ware_CZE_early:KO1002.merged___BC_2583
20.0 BGR_C:I2519___BC_4287
15.6 Corded_Ware_CZE_late:STD001___BC_2550
5.6 MAR_EN:IAM.5___BC_5113
Target: Brahmin_Uttar_Pradesh
Distance: 2.0855% / 0.02085495 | R5P
58.0 IRN_Shahr_I_Sokhta_BA2:I8728___BC_2550
20.4 IRN_Shahr_I_Sokhta_BA2:I11459___BC_2749
15.2 Corded_Ware_CZE_early:VLI007.merged___BC_2789
4.6 CHE_LN:Aesch9___BC_2898
1.8 KAZ_Botai:BOT2016___BC_3446