0814 min

Starting late, or with no idea yet: the step-by-step guide

Seven steps from registering to hitting submit: finding an idea, validating it, the narrowest MVP, the submission materials, drafting the form and sending. For students and teams starting late, on the current rules.

Know this first: you're not too late

The competition started on 14 September and ends on 12 October. This guide takes a student with no idea, or a late start, from registering to the submit button, one step at a time. Whenever you start, the order is the same. Don't skip around.

Registration is still open. Under section 6(a) of the official rules, every member can register until 23:59 PT on 12 October. Joining late isn't a loophole, it's the rule.
Four weeks was only ever enough for one strong use case. A team that started on day one is building what you'd build: one user, one job, one screen. The difference is they could spread it over four weeks.
Speed is scored. Colosseum's evaluation looks at product velocity, and the final call asks what you shipped last week. Getting something working in a short time is good evidence for that criterion, not bad evidence.
It isn't only a coding contest. Colosseum runs it as a pre-seed funnel and the judges score you like investors. Five of the seven official criteria are business criteria. Clean code alone doesn't earn points.

The most expensive mistake is starting on something too big. Every step below also tells you what not to do.

Step 1: Register and form a team

Do this first. You can register without an idea; you fill in the project and team fields later.

01Create the account. Enter through Superteam Türkiye's referral link (below). You can sign up to Arena with email, Google, GitHub or X. The link matters: without it your registration still counts for you, but the attribution to us quietly goes missing.
02An account isn't a registration. Creating an Arena account and registering for the competition are two separate things. Press Register at the top right and finish the flow.
03Set your team's location to Türkiye. Choose Türkiye as the team location during registration. It's one of the conditions for the Superteam Türkiye pool.
04Every team member registers with their own account. A member missing from the project page is a silent disqualifier. A person can be on only one team, and a team can submit only one project (rules, section 7).
05Find a team. You can compete alone, but the judges prefer teams of three or more. If you don't have teammates, use Cofounder Matching or ask in the Telegram group.
06Fill in the Superteam Türkiye form. The orientation form puts you on the list for office hours and mentoring. Registering with Colosseum tells Colosseum you exist; this form tells us.
07Join the Telegram group. The other Türkiye teams, the mentors and the weekly calendar are there. If you ask a question, the answer helps everyone.

Step 2: Find an idea and validate it

The goal: by the end of this step you have a one-sentence idea and an answer to "has this been built before?" Don't let it drag on. If you already have an idea, skip to the validation part.

Learn the method first: start from signals, not from a product. "AI is hot, so I'll build something with AI" is the fastest way to land on the same idea as everyone else. The order is: signals, patterns, an opportunity thesis, product ideas, a filter, what to build.

Ecosystem signals. The Solana Foundation, the Colosseum team, founders and protocol teams say what they want on X, again and again. One tweet isn't a thesis; look for what recurs. Example: in June the Foundation explicitly asked for fully on-chain perps and derivatives, and named adjacent areas like vaults, aggregators and social trading.
Research and capital signals. Read reports from Blockworks, Messari and similar the way a founder would: what's growing, what infrastructure is missing? Example: according to Blockworks, tokenized-asset trading volume on Solana reached about $5.8 billion in Q2 2026, and Solana handled roughly 97% of tokenized-equity volume.
Onchain data. Look at volume, users and revenue on DeFiLlama and Blockworks Analytics. Look two ways: where Solana is already strong (what can be built around it?) and where there's demand elsewhere but little on Solana (why?).
Look outside Solana. Don't clone a product that works elsewhere; borrow the proof of demand. Why does it work, which user problem does it solve, and why doesn't it exist on Solana?
Use AI to gather evidence. Ask Grok or Claude to "find what these accounts have asked for repeatedly over the last six months, and don't generate ideas yet." Don't ask it for the idea. Two people reading the same data still arrive at different ideas, and that difference is your edge.

If you have no idea at all, start from ready-made material. The Requests for Products section on this site gives four ideas for Türkiye: provenance for exported goods, carbon MRV for industry and the EU's CBAM, used-car history and underwriting, and a portable credit history for professionals moving to Europe. Each one carries a "know this before you start" note: all four depend on data somebody else holds. If you're a student, work with synthetic but correctly shaped data and say so on the slide. You don't have to pick from the list, and nobody scores higher for it.

A five-question filter. Run your candidate idea through these. If the answer to any is no, shrink it or change it:

Can I describe one type of user and their one job?
Will it work end to end on a single screen in the time I have left?
Could it exist without a chain? If it could, you'll have to defend your use of blockchain in depth; if you can't, don't use it.
Can I actually get the data or the permissions I need?
Can I explain it to a stranger in ten seconds?

Validate it: a gap analysis with Colosseum Copilot, in an hour. Copilot searches more than 5,400 past hackathon projects. Finding out late that someone already built your idea loses you the competition, so don't skip this.

01Generate a Personal Access Token in Arena.
02Set the environment variables: COLOSSEUM_COPILOT_API_BASE and COLOSSEUM_COPILOT_PAT.
03Add the skill to Claude Code or Codex: npx skills add ColosseumOrg/colosseum-copilot
04Ask: "Validate this idea for the hackathon: [your idea]. Has anyone built this before, what is the competitive landscape, and what is genuinely unsolved?"

Finally, write one sentence. One a stranger understands with no follow-up question: who it's for and what it does. Something like "Stablecoin neobank for SMBs". Don't change it; it goes in your bio, on your first slide, in your pitch's opening line and in every post. If you're stuck, Friday 19:00 office hours (Google Meet) is exactly for this: idea validation and picking a problem.

Colosseum Copilot

Step 3: Build the narrowest MVP

The goal: one user, one job, one screen. On a public domain, on devnet, with a real transaction. Overscoping is the most common way Türkiye teams lose, so the most important job in this step is cutting.

Cut the scope. Write the product as one sentence: "person X does job Y." Anything that isn't in that sentence (a second user role, a settings page, an admin panel, profile editing) goes on a "later" list, and stays closed for now.
Wire the chain first. Wallet connect, then a real transaction on devnet. If the product could exist without a chain, judges notice. Leaving the chain until the end is the most expensive surprise of the last night.
Use existing infrastructure; don't rebuild it. Helius or Triton for RPC, Privy for auth, Jupiter for swaps, Squads for multisig, Pyth for prices. Writing something Helius or Privy already ships is time stolen from what the judges actually score.
Work with an AI assistant. Pairing with something like Claude Code or Codex makes a real difference when time is short.
Get RPC support. Carbium is giving Colosseum teams 25 free RPC and Swap API grants (Developer RPC is 10M credits a month). The application and terms are on the Carbium card on this page; we check fit first. Don't demo on a public RPC: a transaction that drops mid-demo can't be undone.
Put it on a public domain. Vercel or Netlify will do. A localhost screenshot isn't a product, and a GitHub link isn't a demo. Judges want to click and see something work.
Post your first build-in-public update right away. What you're doing, why, and a screenshot. A post made early is worth more than one made late: it starts a record. If you don't have an X account, create one, verify it, and follow the judges.

A toolkit for building faster. Before you write anything from scratch, look at these:

Test after every work session. Before you stop, send the link to a friend and ask: "Could you connect your wallet and finish the transaction?" If the answer is no, the next job isn't a new feature, it's fixing that.

Step 4: Show it to people and collect numbers

The goal: real people used the product, and you have two numbers you can quote. A judge doesn't buy "a great idea", they buy "this many people used it".

Find ten people. Ten who look like your target user: classmates, club members, the other teams in the Telegram group. If the idea is for students, your easiest audience is already around you.
Watch in silence. Send them the product and watch what they do. Don't explain, don't say "press here". Where they get stuck is where the product is stuck. If you can, record the screen (with permission).
Write down what broke, and split it in two. List one: what stopped them using it. Fix those. List two: what only annoys you. Ignore those.
Pick two numbers. Users, returning users, transactions, volume. Choose the two you'd defend in front of an investor and start measuring now. For transactions, your program address on Solscan is enough. It's not too small to start: 10 people and 30 transactions beat zero.
Build the story around the numbers. Not "in the future this will happen" but "10 people tried it, 7 completed the transaction, here's where they got stuck, and we fixed it". Tell the result, not the vision.
Post a short update on X every day. A silent team looks abandoned. The panel is public now: five people from Colosseum and fifteen track judges from teams like Phantom, Base, Anza, Arcium, Drift, Ellipsis and MetaDAO. Follow all of them.
Tag us when you share a real result. Superteam Türkiye amplifies a real milestone ("10 users tried it"). We don't amplify launch announcements; we amplify results.

If you're stuck: the most common blocker at this stage is finding users. Post in the Telegram group, "I need ten people to try this, anyone?" Teams can start as each other's users.

Step 5: Prepare the submission materials

The goal: two videos, a repo, a first slide, and a clear paragraph about your team. Record the video before the slides. Superteam Balkan counted all 139 winning submissions from the last three competitions, and about 88% of winners uploaded a recorded video rather than a static deck. Colosseum also says the pitch video is the first thing judges look at.

Pitch video, at most 2 minutes. In your own voice and on your own face, Loom or YouTube. The first ten seconds decide: say the problem and the solution straight away, not your own story. A suggested order: problem and solution, the product working, the team, your numbers and what's next. The structure is always the same: Problem, Solution, Demo, Team. The Superteam Türkiye Earn listing still accepts up to 3 minutes, but if you're uploading the same video to both places, keep it to 2 minutes and it fits both.
Demo video, at most 3 minutes. Show the live product. Not slides, and not a code walkthrough. Go technical here and don't make a second pitch: this is the only place the judge sees it working.
First slide: the product name and one sentence, nothing else. The name at medium size, the sentence large and five words at most. No image, logo, website, contact details or year. According to the hackathon mentor Josip Volarević, a judge or investor remembers at most 0 to 3 things the next day: a feeling, a key takeaway, and sometimes a second one. Make them remember your sentence, not your name.
Don't undersell your team. Don't list names and titles; say concretely why each person can do this. The weak line: "we're a passionate, hardworking team." A good line (an example): "Aylin wrote the backend of an app that reached 400 users at the last hackathon." No adjectives, no years of experience, write outcomes. Claim only results you were directly responsible for. If you don't have such a result yet, the user and transaction numbers you produce now are your first proof.
Publish the repo. Give the direct GitHub link to the repository, not an organisation or profile page. If it stays private, give access to hackathon@colosseum.com. Locked Google Docs and private repos are the most common, and most expensive, mistake.
Write the business model in three sentences. How does it make money? Don't put it in the video; write it in the matching field on the form. A business model and a roadmap read as padding in a two-minute video, but the form scores you on it.
Do it all in English. The videos, the form answers and the repo description. It's also a condition of the Türkiye pool.
Have someone watch the video before you upload it. Ask a friend, "did you understand what I'm doing?" Bring the rough cut to Friday 19:00 office hours; getting feedback before you polish saves a week.

Step 6: Draft the form in a doc beforehand

Colosseum's project form is full of required fields with character limits. Fitting your business model into 500 characters for the first time with the form open costs far more than writing it in a doc first. Fill this doc in while you do steps 3, 4 and 5.

Tier 1, the public page. Anyone with your project link sees these; write for a stranger.

Project name, brief description (up to 500 characters; put your one sentence here), website (optional), product category, the country your team is based in.
Logo or graphic (JPEG, PNG, WEBP or GIF, up to 20MB before compression).
GitHub repo link (the repository itself, not an organisation).
Pitch video (up to 2 minutes) and your X profile.

Tier 2, seen by your team, the judges and Colosseum, but not public.

What are you building, and who is it for? (1,000 characters)
Why this, and why now? (1,000 characters)
Which technologies and AI tools are you using?
Which chains do you use, and how? (500 characters)
A team Telegram contact. Prize distribution and accelerator interviews go through it, so make it someone who checks it.
Did anyone not on the team do meaningful work on this? (600 characters; answer honestly)
Anything else the judges should know? (500 characters)
Demo video (up to 3 minutes), live product link, and access instructions if needed.

Tier 3, the accelerator screen. Only the organisers and Colosseum see it, but it's required to submit even if you don't want the accelerator. It's really where Colosseum decides who to look at, so take it seriously.

How do you know people actually need this product? (1,000 characters)
How far along are you, and do you have users? Be specific. (1,000 characters)
Who else is building in this space, and what are they getting wrong? (1,000 characters)
How do you make money, or plan to? (500 characters)
How long has each of you been working on this, and is it full time? (500 characters)
Where is the team based, and do you work in person together? (500 characters)
Four yes/no questions: have you formed a company, taken investment, are you fundraising, do you have a live token. There's no right answer, and "no" is fine. Leaving the open questions thin reads worse.

Every field and what it's looking for: What Colosseum's form actually asks

Step 7: Submit

The goal: everything uploaded at least two days before the deadline. The platform slows down on the last night every single time, and any mistake made then can't be undone.

Convert the deadline to your own time. The deadline is 23:59 California time on 12 October. In Türkiye that's the morning of Tuesday 13 October, 09:59. Teams that assume midnight in Türkiye submit ten hours late, and there's no appeal.
Submit on the Colosseum portal. Make sure the team leader has uploaded the project and every member is attached to it. A missing member can invalidate the submission.
Submit the identical project a second time to the Superteam Türkiye Earn listing. The Colosseum portal and the Earn listing are two separate places; doing one doesn't count as doing the other. The Türkiye pool is 10,000 USDG (5,000, 3,000 and 2,000 for 1st to 3rd), with results on 28 October. The conditions: Türkiye as the primary base country on your Colosseum profile, the Superteam Türkiye form, the Colosseum Telegram group, submission to both places, and videos and repo, all in English.
Don't run the link test yourself; have someone else do it. Ask somebody outside the team to open the GitHub link, the demo video and the pitch video. A link only you can open is a link that doesn't open.

Right before you send, tick off these nine one by one:

01The one-liner and short description are written, and someone has read them.
02The MVP is live on a public domain.
03The business model is written in three sentences.
04The chain integration really works: wallet connect and a real transaction.
05The code is published, or the judges have been given access.
06A verified X account, posting every day, following the judges.
07Pitch materials (slide, pitch video, demo) are ready and someone has seen them.
08The project and every team member are registered on Colosseum.
09The same project has also been submitted to the Superteam Türkiye listing.

Four mistakes that lose after the work is done

01Selling the future. A big vision with no record reads as a pitch, not a product. Instead of "in 2027 we'll do this", say "this week we did this, and this many people tried it".
02Forced crypto. If the product could exist without a chain, you have to defend your use of one in depth. If you can't, don't say "we use blockchain".
03A weak demo. A GitHub link isn't a demo. There has to be something a judge can watch working.
04Missing permissions. A private repo, a locked Google Doc, a Loom link that won't open: the most common fatal mistake. If it stays private, give access to hackathon@colosseum.com, then have someone outside the team open every link.

Two rules that get teams disqualified:

A team may submit only one product, and a person may appear on only one submission.
You may start before the window and reuse your own old code, but only work done inside the window is judged. You have to disclose all prior development on the submission form.

Prizes and the accelerator: the current numbers

These figures come from Colosseum's official rules (section 14), not from the launch announcement. The numbers from earlier competitions no longer apply.

Total prizes: $840,000, across the general awards and eight ecosystem tracks.
Grand Champion: $30,000, one team. The next 20 teams: $15,000 each, $300,000 in total. That band is what feeds the accelerator.
Special awards: Public Goods and University, $5,000 each, one team apiece.
Tracks: Solana, Tempo, Hyperliquid and Zcash at $100,000; Ethereum L1, Base, Arbitrum and Robinhood Chain at $25,000. You pick up to three at registration, and every submission competes for the general awards too.
Accelerator: $250,000 per team, a twelve-week programme with the first two weeks in San Francisco and the remaining ten remote-friendly. There's a $2.5M fund behind it.
Türkiye pool: 10,000 USDG (5,000, 3,000 and 2,000), for teams based in Türkiye only. Results on 28 October.
Winners are announced on or about 5 December.

A note for students: the University Award goes to one team and is $5,000. A Türkiye team, Lexicon, won the University Award at the Radar hackathon, so it can be done.

Where to get help

Every weekly event is free and none is compulsory. The current times are on the Luma calendar.

Wednesday 19:00, Colosseum Community Call (X Spaces): the week gets set here, and the room is where teams find each other.
Friday 10:00, Coworking Fridays (Kolektif House Levent, İstanbul): all day, bring the thing you're stuck on.
Friday 19:00, Colosseum Office Hours (Google Meet): one to one. Idea validation, architecture, deck and video review.
Saturday 14:00, Shipyard (Nişantaşı, and livestreamed): the vertical deep-dive.

For general questions, use the Telegram group: the answer helps everyone. Or write to Caner directly at @trench_survivor. If you get stuck on something, stop reading this guide and write.

Stuck on this?

Superteam Türkiye reads every message from a Türkiye team.

Join the group Next: Geç kalanlar ve fikri olmayanlar için adım adım rehber