The question investors ask us most is not what we invest in, but how we decide. And the honest answer starts with an uncomfortable fact: most of a manager's work consists of saying no. Every deal that ends up becoming a vehicle has left behind many others that fell by the wayside — because of the land, the numbers or the exit.
That discarding is not a formality. It is where capital is truly protected, because the mistakes of a real estate investment are almost never made during construction or management: they are made on the day of purchase, when assumptions are accepted that only hold if everything goes well. This article sums up the order in which we analyse, and why that order matters.
We start with what cannot be changed
A construction cost can be renegotiated, a project can be redesigned and a rent adjusts to the market. The location and the planning status cannot. That is why they are the first and hardest filter: if the surroundings do not generate their own demand —transport, employment, universities, retail— or the land carries uncertain permitting, the deal is discarded before the spreadsheet is even opened.
Once the asset clears that bar, the rest is detail work: which uses the planning allows, which administrative timelines are realistic and which product that specific place calls for. Not the other way round. We distrust deals that start with the fashionable product and then look for land to fit it into.
Stress the assumptions, don't dress them up
With the location validated, we build the base case. And here the discipline is arithmetic: every variable is loaded against the deal, not in its favour. In practice, we stress four blocks:
- Costs and timelines. Budgets checked against real contractors, headroom for overruns and calendars that assume the administration will not be in a hurry.
- Rents and occupancy. We underwrite below the rents the market publishes and with occupancy below what the comparables show. If the numbers only work with the rent of the best building in the area, they do not work.
- Financing. Moderate leverage, a cost of debt with room to rise and structures the deal can sustain in the slow scenario too.
- Exit. The price at which a comparable asset sells today is not the one we use: the yield required at exit is always more demanding than at entry.
If a deal only works in the optimistic scenario, it is not a deal: it is a bet. And bets are not made with other people's money.
The result of this exercise is not an exact return —nobody has one— but a range: what the deal returns if things go reasonably well, and what happens if timelines stretch, costs rise or rents take longer to arrive. What we are looking for is a downside scenario that remains defensible.
The exit is decided at the entry
Every real estate investment ends the same way: selling the asset or keeping it for income. That is why the last question of the analysis is the first one you need to be able to answer: who will buy this, and why? A well-located asset, with a clear use and a size that appeals to both private wealth and institutions, has a market at any point in the cycle. A singular asset, however brilliant the project, only has one in the good years.
That logic explains the portfolio: professionalised product on structural demand, like the flex living of Mad Río Hco, or income from solvent tenants on long leases, like the retail units of Cesaraugusta Prime. Different assets, same condition: the exit was clear before going in.
The final filter: our own money
When a deal clears everything above, the definitive test remains: investing in it. We commit a minimum of 5% of the capital of every vehicle, on the same terms as our investors. It is the simplest way to check that the analysis has really been done: assumptions are stressed differently when the first euro at risk is your own.
We do not promise to always be right; we promise method. Explaining the assumptions before asking for the capital, showing the bad scenario next to the good one, and discarding everything that cannot survive that conversation. The rest —the return— is a consequence.