"Best admixture calculator" is the most-asked question in this hobby and the most malformed. A calculator is an optimiser over chosen references — so best depends on which references can speak for your ancestry and on what kind of statement you want to walk away with. Disclosure before the guide: we build one of the tools below and sell a formal alternative, so this post names where the free classics genuinely win.
If the machinery is unfamiliar, start with what an admixture calculator is; if the GEDmatch landscape is, the project map.
By background: the home-field table#
Every calculator fits best the populations that trained it — the calculator effect never sleeps. The community's settled matchups, which our reading supports:
| Your ancestry is mostly… | Run first | Worth a second look |
|---|---|---|
| European (north, east, central) | Eurogenes K13 | K15; K36 for texture |
| Southern European / Mediterranean | K13 + Dodecad K12b | G25 era calculators (below) |
| South or Central Asian | HarappaWorld | Dodecad K12b |
| West Asian / Caucasus | K13 and K12b together | G25 era calculators |
| East European / Baltic | K13 or K15 | MDLP World-22 / K23b |
| African | puntDNAL; Dodecad Africa calculators | Ethiohelix (African-focused) |
| Mixed continental | Two projects, compared | G25 era calculators |
Read any of them with the Oracle rather than the bare percentages — the Oracle guide is short — and remember that the absolute numbers are that calculator's coordinates, not portable facts.
By question: the honest sort#
"Which populations does my genome resemble?" Best tool: not a mixture calculator at all. A plain Global25 distance ranking answers resemblance directly — nearest reference populations by Euclidean distance, across six eras, no components in between. Free, in the browser, no account.
"What are my deep ancestry proportions — farmer, forager, steppe?" Best tool: an era-scoped ancient calculator. The GEDmatch classics predate most ancient genomes; the free Global25 admixture calculators fit your coordinate against curated panels of dated ancient populations era by era, with a stated fit distance. This is the question the 2012 tools were reaching for with the samples of their day.
"How do I compare to my relatives / a forum thread?" Best tool: whichever calculator the people you are comparing against used — comparisons are only valid within one tool. For GEDmatch threads that means K13; for coordinate threads it means the same G25 panel they ran.
"Is this component real? Am I actually part X?" Best tool: none of the above. This is a testing question, and calculators cannot test — no percentages tool can reject its own model. The method built for it is qpAdm: explicit sources, stated outgroups, a p-value that can fail, standard errors and z-scores per source. That is a paid, hand-built analysis here because each model is composed and checked by a person — and for many questions it is more instrument than you need, which is exactly why the free tier of this ecosystem exists.
The three-step routine that beats any single tool#
- Rank first. Closest populations by distance, modern era and two ancient eras. No mixing assumptions; this is your genome's actual neighbourhood.
- Fit second. One curated era calculator for proportions, noting the fit distance; one GEDmatch classic from your home-field row above, for the Oracle view. Agreement between two unrelated tools is the signal worth keeping; disagreement is expected and diagnostic.
- Test only what matters. Most numbers can stay exploratory forever. The one claim you would defend in an argument — that is the candidate for a formal model.
The best calculator, in other words, is a short pipeline rather than a product name — and two thirds of the pipeline is free. If you have raw DNA but no coordinates yet, getting Global25 coordinates is the one setup step the coordinate tools need.
Terms used here are defined in the glossary.
