--- title: "Personalized G25 calculator: every Ancestry Map is built around you" description: "Why a standard Global25 calculator can describe a population well and one person badly, how our analyst builds a calculator around your own coordinates for every Ancestry Map, and what your report shows you." canonical: https://www.ancestrify.io/blog/personalized-g25-calculator date: 2026-08-28 updated: 2026-10-10 author: "Ancestrify" --- # Personalized G25 calculator: every Ancestry Map is built around you Every Global25 admixture result is only as good as the source panel it was fitted against, and every published calculator was built for a population, not for you. Most of the time that is fine: a well-curated panel for a region describes most people from that region well. Some of the time it is not, and the result is a breakdown that fits with a large distance, leans on sources that make no sense for the person, or spreads a real component across three proxies because the right one was never offered. That is why every [Ancestry Map](/g25), our Global25 analysis, is built the other way round. Our analyst builds a calculator around your own coordinates, starting from the populations of the region you declare and testing every source against your row, and a person reviews and approves it before your report opens. It is part of the report, at no extra cost. This article explains why it is needed, how it is built and what your report shows. We are not aware of another service that offers it. ## Why a standard calculator can fail one person A G25 admixture fit searches for the weighted average of source rows that lands closest to your row. Three properties of that search explain most disappointing results: 1. **A missing source is redistributed, silently.** If the population that actually explains part of your ancestry is not in the panel, its share is absorbed by whichever offered sources are nearest: plausibly, and wrongly. 2. **Panels are built for the typical member of a group.** A calculator curated for a country optimises for its most common ancestry profiles. Someone with an unusual combination (two distant regions, a minority background, recent migration) sits at the edge of what the panel was built to cover. 3. **Fit distance hides the problem.** Offer enough sources and the distance falls regardless. A fit can look acceptable while assigning ancestry to populations that are only standing in for the absent one. Reading those signs is covered in [How to read a Global25 report](/blog/understanding-global25). A calculator built around your row is the answer to them, and it is where every Ancestry Map starts. ## How our analyst builds it The discipline is the one behind every curated calculator, applied to one row: **Check the row.** The coordinates are checked as an official, unmodified Global25 projection before anything is fitted, because a converted or simulated row would make every later number meaningless. **Start from your declaration and your row.** The country and region you declare set the first candidates: the populations of that region, era by era. Your row's closest populations, and the individual samples nearest it in PCA space, show where the panel has to reach beyond them. Your declaration decides where the search starts, never the percentages. **Compose era-coherent panels.** Sources are drawn from one era at a time, from the same curated reference of 30,386 samples in 1,535 populations the standard calculators use. Candidates that plausibly precede or constitute your ancestry are offered; ones that merely sit nearby are not. Simpler panels are tried beside richer ones. **Solve and test.** Each panel is solved with the same engine as the standard calculators. Then every source is tested against your row: removed in turn and the fit solved again. A source whose share collapses when a neighbour is present is dropped; a source that survives the removal of its neighbours has earned its place. Repeated solves check that the answer is stable rather than the accident of one run. **Apply the bar.** The panel kept is small enough to be honest, with every source earning a real share and the fit distance within the band accepted for the era. **Lower distance is not better past that bar**: distance falls with panel size, so the search optimises defensibility, not the number. Where a standard calculator covers your region, the one built for you must fit your row at least as well. **Write the note.** Our analyst writes a short note in plain words on what the calculator is and why its sources are there, so the result can be argued with rather than believed. **A person approves it.** Nothing reaches your report until a person has reviewed the calculator, its fits and the note, and approved it. ## What your report shows Your Ancestry Map opens on the calculator built for it. Every era lists its full source panel, split into the populations the model used and the ones it was offered and did not use, with the fit distance and the place the calculator was built for. Beside the results sit three things: - **The populations it used**, era by era, with their shares. - **A note from our analyst** on why the calculator holds the sources it does. - **The panel itself**, yours to copy into any Global25 tool, so you can re-run the fit yourself in Vahaduo or the free [admixture calculator](/lab/admixture). The distance and PCA lenses, and the free Notable Matches lens against 254 published ancient individuals, are unaffected: they never depended on a calculator. ## When your report opens An Ancestry Map opens after a holding period, once a person has approved the calculator built for it, and we email you when it is ready. The optional fast-compute add-on removes the holding period, so your report opens as soon as its calculator is approved. If we are obtaining your coordinates for you through the Coordinate Concierge, our analyst starts once they arrive, and fast compute is not offered with that service. Reports delivered before 10 October 2026 keep their original result. A calculator bought for one of them earlier stays on that report as its own version. ## A worked example Take an imaginary customer with one parent from the western Balkans and one from the Levant, who declares the Balkan side. A standard calculator for either region would fit with a distance well outside its usual band and give the "other" half of the ancestry to whatever sources sit nearest: for the Balkan panel, an Aegean and an Anatolian population standing in for the Levant. The percentages look tidy and describe the panel's limits, not the person. A calculator built around the row starts from the Balkan populations the declaration points to, sees from the row's own distances that a Levantine component is missing, and offers both regions' era-coherent sources side by side: Balkan Iron Age and Roman-period populations, Levantine Bronze and Iron Age populations, and the deeper components each shares. Solved and tested, the panel that survives has a handful of sources, a fit distance inside the band, and a breakdown that reads as a Balkan and Levantine mixture because that is what it is. The note says so in plain words, and the panel is there to copy and check. ## The free G25 Admix report, built the same way The free [G25 Admix report](/blog/free-g25-admix-report) runs on a calculator built the same way: our analyst builds it around the Global25 row you give, starting from the country and region you declare, and a person reviews it before your report opens. You get a notification and an email when it is ready. The free report shows the note from our analyst beside your admixture in every era, your three closest populations and your three closest notable matches. The panel to copy is part of Ancestry Map. ## The Calculator Explorer is no longer sold The Calculator Explorer, which re-solved a report against the other published calculators of an era, is no longer sold. Customers who bought it keep it on their report. ## What it does not do It does not change your coordinates, which come only from the independent Eurogenes service; see [What are Global25 coordinates?](/blog/what-are-global25-coordinates). It does not add a p-value: G25 remains a coordinate fit, and the method that can reject a model is [qpAdm](/qpadm), compared in [qpAdm vs Global25](/blog/qpadm-vs-global25). And it does not make any source population your ancestor: a source is a reference the model tests against, never a statement about who anyone's ancestors were. If you want to see how much a panel changes an answer before buying anything, the free [admixture calculator](/lab/admixture) lets you fit your row against curated panels and your own pasted sources in the browser. The terms are in the [glossary](/glossary).