Choosing a Hackathon Is a Guess
Most people pick hackathons the same way: sort by the biggest prize pool, check the deadline, and register. Maybe glance at the theme.
That's a guess, and it's usually a bad one. The $100,000 hackathon with 8,000 participants and a theme you've never touched isn't a great opportunity — it's a lottery ticket with extra steps. Meanwhile, a $5,000 hackathon that's squarely in your wheelhouse, ending in two weeks, with a few hundred competitors, is where you actually get paid.
The problem isn't a lack of information. HackRadar already knows the prize pool, the deadline, the themes, and — since this week — the participant counts for most events. The problem is that no human can weigh all of it against their own skills for 400+ live hackathons.
So we built that calculation for you: skill-based recommendations.
How It Works: Pick Skills, Get a Ranking
On the HackRadar homepage you'll find a Skills dropdown above the search bar. Select the skills that describe what you build with — 45+ skills across 15 groups:
The moment you select a skill, the list re-ranks: a new Recommended sort order pins your best matches to the top, each item shows a Match badge with the skills it matched on and an estimated expected return, and the feed still respects your search, prize filters, and platform toggles.
Your selection is saved in your browser — no account, no signup. Come back tomorrow and the ranking is still there, recomputed against fresh data (the crawler refreshes every ~6 hours).
The Formula: Expected Return per Day
No black box. Every recommendation is one transparent calculation:
expected return per day = prize × skill match × competition ÷ days to deadline
Here's what each factor means:
Skill match
Your selected skills are matched against the hackathon's themes, title, description, and prize details. Themes count most (an event explicitly themed around your stack is your event), followed by the title, then the body text. Hackathons that don't touch any of your skills are left out of the ranking entirely.
Competition
We now track participant counts for most platforms (Devpost, Devfolio, DoraHacks, Taikai, HackQuest, Luma). Fewer competitors means a higher win chance, so an event with 100 participants scores higher than an identical one with 1,000. Where counts aren't published, the factor stays neutral.
Time, honestly discounted
This is the part we care most about: the expected return is divided by the days left until the deadline. A $50,000 prize 300 days away is worth $83/day of expected value; a $5,000 prize ending in 20 days is worth $125/day. Same weekend of your time — very different return. The formula doesn't bury far-future events, it just prices them honestly per unit of your time.
Prize, normalized
Prize pools in rupees, euros, or pounds are converted to USD with a static exchange table before ranking, so a ₹10,000,000 pool isn't mistaken for ten million dollars.
Why Expected Return — and Not "Most Likely to Win"?
Nobody can predict winners — hackathons are judged by humans with taste. What we can compute honestly is expected value: how much prize money a weekend of your time is statistically worth.
Ranking purely by "biggest prize" overweights lottery-ticket megathons. Ranking purely by "highest win chance" overweights tiny events with meaningless pools. Expected return per day sits in the middle: it's the metric a rational participant optimizes when their time is the scarce resource — which, for most of us, it is.
It also composes with everything else on the site. Pair skill matching with the Known system and the loop is: open HackRadar, your best opportunities are already on top, mark the ones you've reviewed, and the recommendations quietly re-rank to surface what you haven't seen.
Under the Hood
The matching engine runs server-side against the full hackathon dataset — titles, themes, descriptions, and prize details — with a curated skill taxonomy whose aliases are drawn from the real vocabulary platforms use ("Machine Learning/AI", "capture the flag", "Bittensor", "time series"). Skill selection lives entirely in your browser, and recommendations exclude anything you've marked Known, so the top slots always go to fresh opportunities.
The whole thing runs on the same serverless stack as the rest of HackRadar — Next.js on Cloudflare Workers, data from R2 — with zero tracking of your skill choices. Your profile never leaves your device.
FAQ
How do I use skill matching?
Open the HackRadar homepage and use the Skills dropdown above the search bar. Pick one or more skills — the list re-ranks immediately under the new Recommended sort order, and matched items show a Match badge with the matched skills and estimated expected return.
Is my skill selection stored anywhere?
Only in your browser's local storage. Nothing is sent to any server for tracking, and there's no account. Clearing your site data clears your selection.
Why don't some hackathons appear in my recommendations?
Three things keep an event out: no skill match at all, no parseable cash prize (for example MLH events, which don't publish prize amounts), or it's already over / closing within a day. Hackathons you marked Known are excluded too.
Does the ranking guarantee I'll win?
No — and we'd distrust any tool that claimed otherwise. Skill matching ranks by expected value: the prize money statistically worth pursuing for your time. It surfaces events where your skills fit and competition is low, which is the honest, measurable part of picking a winning opportunity. The building is still on you.
Try It Now
Head to hackradar.win, pick your skills from the dropdown, and watch 400+ upcoming hackathons re-rank around you. No signup, no email — just your best opportunities on top.