Inspiration
Data-Driven Developments (a fictional fund) must put $10M CAD into BC short-term rental property. Where, how many properties, what mix, and how confident are we?
What it does
We recommend 14 one-bedroom condos: 9 Downtown, 4 in Gastown/Chinatown, 1 in the West End ($9.7M deployed). Each unit runs on two calendars: nightly stays May to August, furnished stays of 30+ nights the rest of the year. Expected 2027 net operating income is $270K (1 year in 10 below $213K, 1 in 10 above $340K), a 2.8% return on all-in cost.
Why this works:
- Houses earn almost nothing after property tax and upkeep (about 0.4% net yield, 49% lose money). Condos earn about 2.2%.
- Renting nightly all year invites the Empty Homes Tax and the Speculation and Vacancy Tax, which would turn income into a loss of about $152K.
- Furnished months also cushion a possible Vancouver opt-out of the principal-residence rule: $232K at a 20% fall in nightly revenue, versus $153K for nightly-only.
How we built it
A Python pipeline over the 15 June 2026 Inside Airbnb Vancouver snapshot plus public rent, price and tax sources. 6,509 listings were cleaned to 1,037 comparables (licensed, nightly, a year of reviews). Unit economics, a 5,000-draw Monte Carlo per strategy, a validation suite and a decision log. We also built an interactive dashboard (map, sliders, in-browser simulation) and a pitch deck. Every number on the slides is generated from one metrics file by one script.
Challenges
- The June price field is a quote for one stay in World Cup weeks: the same listings cost 16% more than in September, so we rescaled.
- Inside Airbnb occupancy is reviews x 6, so new listings look worse than they are. We required 12 months of history.
- Gastown/Chinatown towers are filed under "Downtown Eastside" but trade like downtown.
- The tax and tenancy reading is ours and needs counsel's confirmation.
Accomplishments we're proud of
We reproduced Inside Airbnb's occupancy formula for 99.9% of listings. A re-run on the September snapshot gave a core yield of 2.2% vs 2.3%. Top segments stayed on top in 99% of 300 split samples. And we list what we tested and dropped: the Superhost premium, one-night minimums, a five-month nightly season, two-bedrooms, and a house with suite and laneway home.
What we learned
Return per dollar, not revenue, is the right lens: costs that scale with price are what sink houses.
What's next
Tax counsel review, building-level screening (strata bylaws, licences, real bookings), and a re-run on each new snapshot.
Use of AI
Claude (Anthropic) helped write the pipeline, dashboard and slide text under our direction. We checked it by reproducing Inside Airbnb's own formulas, re-running on the September snapshot, and a test that compares every published number with the script output.
Team
Chitransh Motwani, Divya Kapoor, Pranay Ratan (Team Two Calendars)
