An iBuyer's single most expensive mistake isn't overpaying — it's buying a home that takes 120 days to sell instead of 40. At a 12% annualized holding rate on a $400K home, that extra 80 days costs $10,500 in pure carrying cost. Add the price cuts you'll take after day 45 on market and you're looking at $18K+ in margin destruction — on a deal that looked profitable at acquisition.
The problem: most acquisition models optimize for price accuracy but ignore time-to-exit. A home priced perfectly at $400K that takes 4 months to sell is worse than a home where you overpaid by $5K but it sells in 3 weeks.
What Makes a Property Liquid
Liquidity in residential real estate is driven by how many active buyers exist for a specific property at a specific time. That's a function of:
| Feature | Source | Effect on DOM |
|---|---|---|
| Price band alignment | MLS price distribution | Homes at $399K sell 2× faster than $415K in same zip |
| School district rating | GreatSchools API | 8+ rating → 35% faster absorption |
| Bedroom count | Property records | 3-bed is max liquidity; 2 and 5+ sit longer |
| HOA monthly fee | MLS / disclosures | > $350/mo cuts buyer pool by 40% |
| Active inventory ratio | MLS snapshot | Supply/demand at zip level, weekly |
| Seasonal month | Historical DOM by zip | May listings sell 2× faster than December |
| Listing photo count | MLS listing data | > 25 photos → 20% faster sale |
| Distance to highway | GIS data | < 0.2 mi kills liquidity; 0.5–2 mi is optimal |
The Model
Days-on-market is a time-to-event variable with right-censoring — active listings haven't sold yet but aren't dead. This is another survival analysis problem. An Accelerated Failure Time (AFT) model with a log-normal distribution:
The coefficients directly tell you the percentage change in expected DOM for each feature. A coefficient of −0.25 on school rating means a one-unit increase in school rating reduces expected DOM by .
The output is a full distribution over DOM — not just a point estimate. The liquidity score is the expected DOM mapped to a 0–100 scale:
Score of 90 = sells fast. Score of 30 = expect it to sit. The score is computed at acquisition time using only pre-purchase features.
How to Use the Score
1. Risk-adjusted bidding
The acquisition model already computes an optimal offer price. Apply a liquidity discount:
Where calibrates how aggressively you penalize illiquid homes. A home scoring 40 gets a 3–4% haircut on the offer. Either the seller takes the lower price (compensating you for the carrying risk) or they don't — and you avoid a slow asset.
2. Portfolio-level liquidity management
Set a constraint: no more than 20% of active inventory can have a liquidity score below 50. This prevents the slow tail of the portfolio from dragging down the capital efficiency of the entire operation. If you own 500 homes and 150 of them are scoring below 50, your capital is trapped.
3. Listing timing
The seasonal component of the liquidity score tells you when to list, not just how to price. A home scoring 55 in December might score 78 in April. If the holding cost of waiting 4 months is less than the DOM improvement, wait.
Backtesting
12-month backtest on 4,200 closed transactions: Quantile accuracy: P50 coverage: 52.1% (target: 50%) ✓ P90 coverage: 88.7% (target: 90%) ✓ Discrimination: Top quartile (L > 75): median 22 DOM Bottom quartile (L < 40): median 78 DOM Separation ratio: 3.5× Portfolio impact (simulated): Avoiding bottom-decile acquisitions → -6% deal volume, +14% capital efficiency
The non-obvious insight: liquidity is more predictable than price. AVM errors are ±5–12%. DOM prediction errors are ±20% — but the ranking is stable. You don't need to know exactly how long it'll sit. You need to know which ones will sit longest.