Skip to content
Fottory
Start free
Menu

Writing

Tinder Elo score: retired in 2019, still being sold in 2026

Search for your Tinder Elo score and you will find a large body of advice on how to raise it. Almost all of it is optimising for a system Tinder said it stopped using in 2019. The strange part is not that the advice is wrong. It is that the correction is published by Tinder, in plain English, on its own press site, and has been for years.

What Elo actually is

Elo is a rating system from competitive chess, named after Arpad Elo. Every player holds a number. When two play, the winner takes points from the loser, and how many depends on the gap between them: beating someone rated far above you moves your number a lot, beating someone far below moves it barely at all. It is a way of extracting one comparable number per participant out of a long series of pairwise results.

Transplanted to a swipe app, the idea is that every swipe is a match in the chess sense. Being liked by someone with a high number is worth more than being liked by someone with a low one, and your number determines who you get shown to. It is an elegant fit for the problem, which is part of why the story has been so durable.

It was real, and Tinder confirmed it

This is not a myth that was invented from nothing. Around 2016 Tinder acknowledged using an internal score referred to as Elo, and Austin Carr of Fast Company was shown his own score of 946 during a visit to the company in January 2016. It is a journalist's account of what he was shown, which makes it a secondary source, and it should be weighed as one. But it is a contemporaneous, on-the-record account, and Tinder did not dispute it at the time.

So people who insist Elo existed are right. They are describing a decade-old system.

Then Tinder published that it stopped

In Powering Tinder: The Method Behind Our Matching, published on 15 March 2019 and last updated on 11 July 2022, Tinder writes: "Elo is old news at Tinder. It's an outdated measure and our cutting-edge technology no longer relies on it."

The replacement is described in the same piece. Tinder says it now has "a dynamic system that continuously factors in how members are engaging with others on Tinder through Likes, Nopes, and what's on members' profiles". The factors it names are activity level and whether two people are active at the same time, location and distance, age and gender preferences, interests listed on the profile, anonymised cues drawn from photos, and Likes and Nopes. It also states that the algorithm "doesn't track social status, religion or ethnicity".

Treat that for what it is. It is a platform statement, not an independently verified one, and there is no way for anyone outside the company to audit it. What it does establish beyond argument is that Tinder has publicly disowned Elo, which is a strange thing to do if you are still running it and a strange thing for advice pages to ignore.

The word does not appear in their filings either

Match Group's Q2 2026 prepared remarks, filed with the SEC on 4 August 2026, discuss Tinder's recommendation systems at length. Elo does not appear in the document. Neither does the word score, nor the word ranking, in any context.

What appears instead is a different vocabulary entirely. The filing talks about Sparks, which Match Group defines as "the number of users engaging in six-way conversations on Tinder in a given week", and Sparks Coverage, "the percentage of active Tinder users who experience a Spark in a given period". It says updates to the recommendation algorithms for straight women went out earlier in 2026, that a further version rolled out in mid-July, and that the update is being extended to straight men and LGBTQ+ users.

That is a company measuring itself on conversations, not on a desirability number. You cannot prove a negative from a filing, and a private ranking value could exist under any name. But if a hidden desirability score were the organising idea of the product, this is the document where you would expect to find its fingerprints, and it has none.

Tinder's own product keynote on 12 March 2026 used the same vocabulary: Chemistry, described as a daily curated recommendation, and Learning Mode, described as a real-time recommendation system designed to more quickly understand what you are looking for, alongside Face Check, Camera Roll Scan and Events. The pitch is about inferring preferences faster, which is a different thing from sorting people by rank.

Why the advice survives anyway

Bad advice does not persist because people are gullible. It persists when it has useful properties, and "raise your Elo" has three.

Look at what the advice actually reduces to and it collapses further. "Swipe more selectively", "complete your profile", "use the app regularly", "improve your first photo". Some of that is reasonable on its own terms. None of it needs Elo to justify it, and none of it becomes more true when Elo is attached. The score is decoration on advice that would stand or fall on its own.

The same thing has happened on Hinge

Hinge has its own stale claim doing the rounds. TechCrunch reported on 11 July 2018 that Most Compatible used the Gale-Shapley algorithm, the Nobel-winning solution to the stable marriage problem. The "eight times more likely" figure that travels with that story came from Hinge's own early market tests and has never been independently verified.

Hinge's most recent statement about its own system does not mention Gale-Shapley at all. In Evolving Together: How Daters Helped Shape Hinge in 2025, published on 10 November 2025, it says "Our updated algorithm uses deep learning to deliver more thoughtful recommendations by better predicting mutual compatibility, contributing to a double-digit increase in matches overall". That double-digit increase is Hinge's own internal figure, presented without published data. The pattern is identical to Tinder's: a specific, quotable, seven-year-old mechanism outlives the company's own account of what it is running.

What to do instead

Nothing, in the sense of nothing aimed at a score. There is no number to raise, no reset, and no ritual that restores one. Deleting your account to clear an Elo you cannot observe costs you your matches and your conversation history in exchange for a mechanism its operator says it retired.

The things worth doing are the ones that stand up without the story. The ranked version, separated by how well each one is evidenced, is in Why am I not getting matches. If your results changed abruptly and you want the documented explanation, Tinder matches suddenly stopped has it. And if the suspicion is that you have been quietly hidden rather than merely ranked low, Shadowbans: three different things wearing one word covers what can and cannot be verified from outside.

Written by Fottory, which generates dating photos from photos of your own face. The first two are free, and none of the advice above depends on using it.