Maybe you are right about being entirely unrelated. However, all it's close matches are with EIA, IA and Roman Hungarian samples. It probably has some Dacian mixture, otherwise it would not hover around Hungarian matches.
Reveal your samples used, so I can replicate it.
A little irony in your comment, since you lump Dalmatians together as if they are all one, many of which carry MENA ancestry, while claiming I am doing all encompassing overlaps.
isn't it obvious from the names of the samples? If you want an accurate model, you can't separate east Med admixture from IA one because there can't be any model which treats IA + Roman + Slavic as contemporaneous sources.
Albania_IA/Early_Medieval:ALB_Çinamak_Anc,0.123904,0.153200,0.028823,-0.012459,0.029148,-0.005379,0.001175,0.000396,-0.000292,0.024107,0.002598,0.006573,-0.018307,-0.007648,-0.009190,-0.002462,0.010989,0.001412,0.004741,-0.008665,-0.005633,0.001767,-0.003363,0.006593,-0.005474
Albania_IA/Early_Medieval:ALB_Mdv,0.120273,0.150298,0.015839,-0.027993,0.019696,-0.007251,-0.001410,-0.003154,0.002591,0.022355,0.000487,0.005296,-0.014519,0.000092,-0.015246,-0.010077,0.009257,0.000718,0.008506,-0.013256,-0.010856,0.005276,-0.001356,-0.003896,-0.003393
BGR_RomByz,0.114961,0.180764,0.000754,-0.055556,0.023081,-0.020080,-0.006110,0.000462,0.017998,0.043737,-0.008607,0.001649,-0.024678,-0.017616,-0.021444,-0.007160,0.020079,-0.013302,-0.006913,-0.004752,-0.013227,0.016075,-0.006655,0.023377,-0.008981
Dalmatia_early_medieval:HRV_Trogir_Byz,0.115872,0.146033,0.004827,-0.027261,0.021604,-0.012550,0.000047,0.002031,0.001145,0.017057,-0.000357,0.006564,-0.010466,-0.003468,-0.008062,0.012092,0.017732,-0.001140,0.001333,-0.002676,-0.003344,0.001212,-0.000887,0.000530,-0.000120
KAZ_Hun-Sarmatian,0.0387,-0.3879325,0.0771215,-0.0224485,-0.062781,-0.038766,0.017861,0.018922,0.003988,0.007472,-0.0261445,-7.5e-05,-0.0052035,-0.001789,-0.0012215,-0.000862,-0.004498,-0.0022805,-0.002074,0.0047525,-0.016533,0.005317,-0.0146665,-0.0028915,0.003892
TUR_Marmara_Ilıpınar_Byz3,0.130897,0.133034,0.065242,0.037791,0.043085,0.008646,0.008225,0.019615,-0.005931,-0.016401,0.000650,-0.007343,0.008771,0.019955,-0.008143,0.003447,0.009257,-0.004434,-0.004651,-0.002626,-0.007736,-0.005193,0.011216,-0.007712,-0.007664
Check how the individual range is always much bigger than the average:
Target: ALB_PostMdv:I14687
Distance: 3.3526% / 0.03352616
42.0 Dalmatia_early_medieval
32.2 BGR_RomByz
13.4 TUR_Marmara_Ilıpınar_Byz3
12.4 Albania_IA/Early_Medieval
Target: ALB_PostMdv:I15707
Distance: 2.8572% / 0.02857242
67.2 Albania_IA/Early_Medieval
20.2 Dalmatia_early_medieval
12.2 BGR_RomByz
0.4 TUR_Marmara_Ilıpınar_Byz3
Target: ALB_PostMdv:I14685
Distance: 2.4024% / 0.02402450
48.0 Albania_IA/Early_Medieval
26.6 TUR_Marmara_Ilıpınar_Byz3
19.4 Dalmatia_early_medieval
6.0 BGR_RomByz
In averages, individual variability is lost and because over a large number of samples many more samples will score very low or no BGR_RomByz you get no RomByz ancestry. Still, 0% or 10% it doesn't matter at all in any model.
Target: ALB_PostMdv:I15707
Distance: 2.9086% / 0.02908619
59.2 MKD_BA
38.6 ALB_Cinamak_Anc
2.2 ARM_Noratus_Anc
I15707 in your model doesn't pick G25 Albanians at all but ancient sources. That's because of the very low admixture and in fact this happens with many samples. It's ludicrous that you even suggest that because you can't model a sample on G25 with an average that it's not part of, the radiocarbon dating is wrong. It's utterly ludicrous.