Équinoxe Pulse

• 

• 

• 

• 

Inspiration

Rent forecasting gets complicated when portfolio mix, concessions, renewals, relocations, and provincial rules distort simple averages.

What it does

Équinoxe Pulse forecasts 2026 rent growth using same-unit lease history, effective rent, renewal vs relocation behavior, and asking-rent signals.

How we built it

We corrected for portfolio mix bias, matched consecutive leases for the same units, separated renewals from relocations, integrated market context, and backtested multiple forecasting approaches on 2023–2025.

Concessions and Effective Rent

We compared contractual rent (sRent) with effective rent (sRentEffective) to measure the impact of concessions.

The gap became especially important from 2024 onward:

• 2024: contractual growth = 3.43%, effective growth = 1.87%
• 2025: contractual growth = 7.11%, effective growth = 3.64%

The main reason is that concessions became more significant. In 2025, the median discount reached roughly 10.3%, and more than 85% of leases had an effective rent below the contractual rent.

Because of this divergence, we selected effective rent growth as our main forecasting target. It better represents the rent actually realized after concessions.

B3 Forecast Formula

For relocations, B3 starts with the asking-rent gap:

Asking Gap = (Latest Asking Rent / Previous Effective Rent) - 1

Then we measure how much of that gap was historically realized:

Capture Difference = Realized Relocation Growth - Asking Gap

where:

Realized Relocation Growth = (New Effective Rent / Previous Effective Rent) - 1

The relocation forecast is:

Forecast Growth = Asking Gap + Historical Capture Adjustment

If the interval between leases is not exactly one year:

Annualized Forecast = (1 + Forecast Growth)^(1 / Years Between Leases) - 1

For renewals, B3 keeps the historical median same-unit growth for the corresponding jurisdiction.

In simple terms: B3 = historical renewal behavior + current asking-rent signal for relocations.

Result

Our best explainable method achieved a 0.542 percentage-point MAE and forecasts 4.19% rent growth for 2026.

What we learned

More complexity does not always improve accuracy. The best-performing method was also the most interpretable.

Built With

• css
• jupyter
• matplotlib
• notebook
• numpy
• pandas
• python
• react
• recharts
• scikit-learn
• tailwind
• typescript
• vite