---
title: "Ancient vs modern admixture calculators: which reference panel answers your question"
description: "A calculator against modern populations answers 'who do I resemble today'; one against dated ancient samples answers 'which deep ancestries formed me'. Mixing up the two produces the classic misreadings."
canonical: https://www.ancestrify.io/blog/ancient-vs-modern-admixture-calculators
date: 2026-08-31T08:04:00+00:00
author: "Andi Thomaj"
---

# Ancient vs modern admixture calculators: which reference panel answers your question

Every admixture calculator fits you against references — and the single most consequential fact
about any calculator is *when its references lived*. A panel of modern populations and a panel of
dated ancient samples produce different-looking percentages from the same genome, both correct,
answering different questions. Most calculator arguments online are two people comparing answers
to questions they did not notice were different.

The machinery itself is covered in
[what an admixture calculator is](/blog/what-is-an-admixture-calculator); this post is only about
the reference axis.

## Modern panels: resemblance among the living

Fit against modern populations — as [Eurogenes K13](/blog/eurogenes-k13-explained),
[K36](/blog/eurogenes-k36-explained) and the testing companies mostly do — and the output reads
as *affinity to present-day groups*: how your allele frequencies or coordinates sit among people
alive now. That is genuinely useful. It is also built on a moving target, because every modern
population is itself a mixture of the same deep sources. When a modern-panel calculator says
"32% Tuscan-like", it is describing your position relative to one *blend*, not naming an
ingredient. Assigning ancestry to a modern country label — the most tempting reading — is
exactly what this panel type cannot support: modern borders postdate the mixing by millennia.

## Ancient panels: ingredients with dates

Fit against ancient samples and the components stop being blends: a
[Neolithic Anatolian farmer](/blog/neolithic-farmer-ancestry-explained) reference *is* the
ingredient, excavated and radiocarbon-dated. A European genome fitted against
[hunter-gatherer](/blog/hunter-gatherer-ancestry-test), farmer and
[steppe](/blog/yamnaya-dna-steppe-origins) sources returns proportions of actual formation-era
ancestries — the decomposition the 2012 GEDmatch calculators were groping toward with the labels
of their day ("West Med", "Gedrosia") before the genomes existed to anchor them.

The date cuts the other way too: ancient panels cannot see anything *recent*. A person with an
Irish parent and a Greek parent fits cleanly as farmer-plus-steppe-plus-forager proportions that
resemble neither parent's country, because the panel's resolution ends where the deep ancestries
stopped differing. Recent admixture is a modern-panel (or better, a matching-based) question.

## One more axis: when the panel was frozen

GEDmatch's ancient-flavoured calculators (Eurogenes Hunter-Gatherer vs Farmer, puntDNAL K12
Ancient) were built from the handful of ancient genomes available in 2012–2015 — before Caucasus
hunter-gatherers were published, before the [AADR](/blog/aadr-allen-ancient-dna-resource-explained)
existed. They are historically interesting but structurally incomplete: an ancestry stream absent
from the panel gets [silently reassigned](/blog/why-admixture-calculators-disagree) to whatever
is nearest. A current ancient panel, by contrast, draws on tens of thousands of published dated
samples.

## Era-scoping: the resolved version of the choice

The clean modern form of this whole distinction is not "ancient versus modern" but *era by era*.
Our free [Global25 admixture calculators](/lab/admixture) are scoped to dated windows — Late
Bronze Age, Iron Age, Imperial Antiquity, Middle Ages, Early Modern, Modern — so "which sources
explain me" is answered within one time slice at a time, and the
[distance rankings](/lab/g25-distance) run across the same six eras. Reading your genome against
1200 BC and against the present are both legitimate; the era label keeps you honest about which
question you asked. (Coordinates are the one prerequisite —
[here is how to get them](/blog/how-to-get-global25-coordinates).)

## The matching table

| Your question | Right panel |
|---|---|
| Who do I resemble among the living? | Modern — [distance ranking](/lab/g25-distance), modern era |
| What deep ancestries formed my genome? | Ancient — [era calculators](/lab/admixture) |
| Was my great-grandparent from X? | Neither — recent genealogy needs matching, not admixture |
| Is a steppe source *required* for my genome? | Ancient + a test — [qpAdm](/qpadm), which can reject the model without it |

The last row is the general rule in miniature: panels choose the question, and only a formal
method turns an answer into a claim that can survive argument —
[the difference in kind is here](/blog/qpadm-vs-admixture-software#is-qpadm-worth-it).

Terms used here are defined in the [glossary](/glossary).