Twelve places out of twenty-five thousand
Cohort selection is where most programmes pretend to be mysterious. It is not mysterious. It is arithmetic, and the arithmetic is public if you ask for it.
Take the shape of it: 25 places per cohort, and roughly 25,000 applications across a year. That is about 12 admitted against roughly 25,000 expressions of interest in the widest reading, and a rate near four hundredths of one per cent. The denominator is the only part of that sentence that is fixed. Everything else is a decision made by the applicant before the form is opened.
The acceptance rate is not the useful number
The useful number is the ratio the reviewer applies. It is almost never the whole application, because nobody reads 25,000 files. In practice a reader takes between two and four minutes on a first pass, and in that window a few things decide whether the file moves forward.
Three of them are countable. Whether the person has built something that can be opened in a browser. Whether there is a link that still works. Whether the writing states what was done rather than what the person hopes to do.
The rest is cost. Every claim without evidence makes the reader do work, and readers do not do work they were not asked to do. So the practical question is not "am I good enough". It is "how much imagining does this file demand of the person reading it".
What the file looks like when it works
A portfolio is not a list of courses. It is a short set of things that ran: a page deployed, a dataset cleaned, a client who paid, a system that stayed up. Each one has a link and a date. Each one can be explained in 2 sentences, which is roughly the length of attention available on a first pass.
The strongest files carry 3 to 5 of those and nothing else. They do not explain the applicant's childhood, they do not describe a passion in the abstract, and they do not use the word "passionate" at all. They state what exists, and they let the reader draw the conclusion.
There is a second-order effect that matters more than most people accept. Programs that place people into operating companies are not selecting for potential. They are selecting for the smallest possible gap between hiring someone and getting output. The file that removes that gap is the file that gets a second read.
Why the model is project-based
The reason a programme runs on live projects rather than on a curriculum is that a curriculum has a half-life. The tools change inside a year, the market moves twice as often, and a syllabus written 18 months ago is a description of a market that no longer exists.
So the material is rebuilt weekly, and the assessment is replaced with output. No grades, no exams, no textbooks. You are placed into an operational role inside a company that has clients, deadlines and a budget, and your work either runs in production or it does not.
That design changes what a selection process has to look for. If the answer to "how will we assess this person" is "by what they shipped", then the entry filter is looking for evidence of shipping, at any size.
The people who came before you
The alumni list is the argument. 210 or more alumni across 19 to 26 nations, from cohorts that have run through the same filter. Some have gone into investment banking, audit and advisory; some stayed inside the ventures they were placed in. A cohort of 25 people who each spent 3 to 6 months inside live companies produces a network that is small enough to know each other and large enough to matter.
That is the real asset of a narrow door. It is not the acceptance rate. It is the density of the room you enter once you are through it, and the number of people who can vouch for your work in one sentence.
Count backwards from the intake date
Selection is a date, not a judgment. There is an intake date, there is a review date, and there is a date by which the file has to be complete. Working backwards from the intake, the file has to be closed several weeks before. Working backwards from that, the artefact you are proudest of has to be finished a month earlier again.
So the preparation order is fixed. Build the thing first, publish it, then write the application about it. Not the other way around. The candidate who writes the application and then looks for a project is trying to borrow evidence, and reviewers notice borrowed evidence faster than applicants expect.
I have built companies across twelve countries and hired into teams in most of them, including a EUR 75 million industrial group, 210 energy systems, 200 wood-gasification machines and a listing in Frankfurt. Sixteen years of field work sits behind that. The people who were in the room when a door opened were not the ones with the longest list of adjectives. They had 3 links, a folder and a date.
The probability that a strong candidate is passed over by a cohort that fills in a week is not zero. The probability that a candidate with no published work and no working link is admitted because the essay was well written is close to zero, and most applicants are pricing it as though it were much higher.
The Summit runs 18 to 20 November 2026, online, in Las Palmas and in Bali. If you want a conversation about what a selection process actually filters for, or about advisory, board seats, or media appearances, write to media@exventure.co.