Your age, their filters, and the ages people really marry. Three questions, real US data.
Most people set their age filter by habit, not by data. The range you swipe on decides who you can ever meet. This tool grades yours against who is really searching for you, and against who people your age actually marry.
I'm making introductions between people in my network and finding the people who actually clear your bar.
I'm making introductions between people in my network and finding the people who actually clear your bar.
Everything on this page is computed live in your browser from published data. There are three calculations, and each one is a simple, inspectable combination of real numbers. No AI, no black box, no server.
For every age of the opposite sex, we ask two questions: how many single, actively looking people are there at that age where you date, and does their typical dating app age filter include your age? The number of singles comes from the Census Bureau's American Community Survey (2024, table B12002: never married, widowed, and divorced, by sex and age), reduced by Pew Research Center's measured shares of singles who are actually looking to date (57% under 50, 36% at 50 to 64, 16% at 65 and up, with men looking more than women). The typical filter windows come from the largest published dataset of real dating app age settings: OkCupid's median search ranges by age and gender. A median 31 year old man sets 22 to 35. A median 42 year old man accepts a woman up to 15 years younger but no more than 3 years older. Women set windows roughly centered a little above their own age at every age. Filter edges are treated as soft (a couple of years of tolerance), because research shows people regularly date just outside their stated ranges.
American Community Survey microdata (2018 to 2022, via IPUMS, as computed for FlowingData's "Ages of the People We Marry" project) gives the full distribution of spouse ages for each age, sex, and marriage number. We blend the first marriage and remarriage curves as you age, since a third of American newlywed couples are remarriages and the blend point matters most after 40. To connect swiping today with marrying later, we add the typical American runway from meeting to wedding: about 3.3 years of dating before the proposal plus a 15 month engagement, shortened a little because couples who meet through online dating marry measurably faster. The runway varies mildly with your age, following the only government data that speaks to it: CDC survey data shows couples who start living together at 25 to 29 convert to marriage fastest (17 month median cohabitation, 58% married within 3 years), the youngest couples slowest (25 months, 31%), with 30 to 44 in between, and Stanford's relationship hazard models peak in the early 30s. So the calculator uses about 5 years if you match at 20, about 3.5 at 36, drifting back toward 4 at older ages. No study publishes this curve directly; ours is a disclosed blend of those sources, and the results barely depend on it. The age gap between you and a future spouse does not change over those years, which is what lets the math work backward to "the age they are today." When you pick a city, the panel also shows that state's median age at first marriage, because the Census does not publish that figure below the state level.
Every age from 18 to 75 gets a score that multiplies three things: singles supply at that age, the probability their filter points back at your age, and how often marriages like yours actually involve that age gap. Scores are shown relative to your best possible age (100). Strong is at least 55 of 100, fair is 18 to 54, weak is below 18. The coverage percentage is the share of your total realistic match weight that falls inside your chosen range, and the suggested window is the same width as yours, just placed where the weight actually is.
Reach alone rewards casting a huge net, so the panel leads with odds per swipe instead. Odds per swipe is your range's score per available single, divided by the best three year window's score per available single. It is a ratio, not a share: 39% means your average swipe is worth 39% of one aimed at your best years, not that 39% of your swipes land well. It falls as a range widens, which is the point, because widening is exactly how people hide a bad filter behind a good coverage number.
Because that number mixes two different mistakes, aim is shown separately. Aim compares your range to the best window of the same width, so it is width neutral: 100% means correctly pointed, however wide you cast. The two together tell you which fix you need. Low aim means move your range. Good aim with low odds per swipe means narrow it. Both matter, and neither can be gamed alone: a one year range scores perfectly on aim and terribly on reach, and the verdict calls that a keyhole rather than congratulating it. The three year benchmark is a deliberate choice of yardstick, and it means even a perfectly placed eight year range scores in the sixties on odds per swipe, so the thresholds for green and amber are set against what is actually achievable rather than against 100.
The last tile counts the waste: the share of the people you would swipe past who sit on ages in the weak tier. Those stretches are named in the same tile and shaded on the chart, so you can see exactly which years your filter is spending itself on.
These are patterns, not predictions: the age gap data tells us how often marriages like yours have landed on each age, not that any one relationship failed because of the gap, and part of that pattern reflects who was available rather than who was chosen.
Two outcome layers sit under the marriage panel, both descriptive. Divorce risk is the chance a first marriage beginning at your projected wedding age ends within ten years. The level comes from the official NCHS life table (first marriages beginning at 25 or over survive ten years 78% of the time), rescaled to how often Americans divorce today, since the first divorce rate fell from 18.7 per 1,000 women in 2008 to 13.4 in 2023. The slope with age is shown as a range, not a single number, because good studies disagree by roughly fourfold on how steep it is: a large Census-based analysis finds about 3% lower odds per year of age at marriage, while a recent longitudinal study finds about 13%. Outside ages 20 to 40 the curve is held flat rather than extrapolated, because the surveys behind it thin out past the late thirties.
The curve is monotonic: marrying later is associated with less divorce, with no penalty for waiting. The widely repeated claim that risk climbs again after 32 is not modelled here, and that is deliberate. It comes from a survey that stopped interviewing at 44, so people who married late had barely been observed; its own author flagged the censoring; no study has reproduced it on newer data; and the fitted turning point is 28 rather than 32. Five independent sources, including the Census and the NCHS, find no upturn at all.
Nothing on this page attaches an outcome to the age gap between partners. The figures circulating online (a 10 year gap meaning 39% more divorce, a 20 year gap 95%) come from a chart whose own author withdrew it after the study's authors objected. In the underlying data the raw association is not statistically significant, the effect appears only once controls are added, and the strongest US national probability sample concludes that an age gap in itself does not raise the odds of divorce. What it does track is who tends to marry across a gap, which is a different thing entirely.
Children are shown as plain description: the share of American women at that age who have not had a child, and the average number of children women that age have, both from the Census Bureau's June 2024 fertility survey. Because it describes the mother, the figure follows the woman in the couple whichever partner is using the calculator. There are deliberately no clinical numbers here: no monthly conception odds, no IVF success rates, no fertility prognosis. Those are medical questions, they depend on a person's own history, and a dating calculator has no business answering them.
The research is unusually consistent on one point: stated filters are habits, not destiny. In a 2025 study of 4,542 real blind dates (published in PNAS), people's stated maximum partner age had no measurable effect on how attracted they were on the actual date. Men's messaging on OkCupid concentrated at the bottom of their stated range and below it, yet the marriages that actually happen cluster within a few years of the couple's own ages: 35% of American married couples are within one year of each other, and just over half of husbands are two or more years older. The filter you set decides who you can ever meet. The data says most people aim it somewhere other than where their real matches are.
Choosing a city replaces the national singles supply with the same Census table pulled for that metropolitan statistical area (ACS 2024 1-year, table B12002), so both the counts and the local balance of single men to single women are that city's own. The fifteen metros available are Atlanta, Boston, Chicago, Dallas and Fort Worth, Denver, Houston, Los Angeles, Miami, New York, Philadelphia, Phoenix, San Diego, the San Francisco Bay Area (the San Francisco and San Jose metros combined), Seattle, and Washington DC.
The differences between them are real and they change the answer. Counting singles aged 25 to 34, Seattle has about 129 single men per 100 single women, Denver 125, Phoenix 123, the Bay Area 121, Los Angeles 112, and New York and Philadelphia only 106. That is a 23 point spread between the tightest and the loosest market for the same person at the same age. Every metro then flips the other way in later life, to a large surplus of single women past 60, because women live longer and remarry less. The city you are in is a real input, not decoration.
Two things deliberately stay national: the dating app search window curves and the marriage age gap curves. No metro level source for either exists, and inventing one would be guessing. So a city changes who is available around you, not how people in that city set their filters or how much older or younger they marry.
Vintages vary because no single modern source covers stated filters, behavior, and marriage outcomes together. Where sources conflict, the calculator uses stated search filters for "who is searching," population data for supply, and marriage records for outcomes, and says so above.
Everything runs in your browser. Your answers are never sent, saved, or shared.