All insights

The thousand lives of a single house

A sale price is a number. A future is a branching set of chances, delays and decisions.

On the last morning of the fourth month, a seller refreshes her bank balance in a parked car. Her house closed on Friday. The proceeds are not yet usable, and another payment comes due today. This is a hypothetical scene, not a reported transaction. It contains the difference that matters: a property can be sold and still fail to produce cash by the date its owner needs it.

Go back four months. The house has an estimated value, a fresh coat of paint and a price she would be pleased to get. What she needs to know is different: how much can this asset produce, by when, and with what chance of coming up short? One number cannot carry that answer. An estimate of value says little about the time required to find a buyer, whether an offer survives inspection and financing, or how much remains after the costs of a sale. We tell transactions as straight stories because straight stories are easy to draw. List. Offer. Close. Funds. The arrows bend in real life.

Many runs, one decision

Monte Carlo simulation is a way to study that bending. Start with a model of how a transaction might unfold. Give uncertain inputs plausible ranges or probability distributions. Draw one combination of inputs and follow it through the model. Then do it again, many times. The collection of outputs reveals a spread of possible results. NIST defines Monte Carlo sampling as a computer experimental method using random numbers to estimate distributions of simulator outputs.

The random draws are the least interesting part. The difficult work comes earlier. What counts as an outcome? Which inputs move together? If mortgage rates rise, do buyer arrivals also slow? If a seller reduces an asking price, does the buyer pool change? How often does a contract fail, and what happens to the listing afterward? These are not details to hide inside an engine. They decide whether its results mean anything.

Our hypothetical owner could compare two policies. The first is to ask for a higher price and be prepared to wait. The second is to start lower in the hope of attracting more buyers before the deadline. A single predicted sale price invites the first policy. A simulation might show something less comfortable: more upside in some higher-price paths, but more paths in which the calendar runs out. The lower ask might reduce the best outcome while making timely cash more likely. Neither policy wins without knowing what the owner can afford to risk.

To make that comparison fair, run both policies against the same simulated market conditions. Let the same broad shifts in demand, rates and competition confront each choice. Then changes in the distribution are easier to attribute to the policy being tested, rather than to a lucky collection of draws. Even this does not prove a causal effect if the rules connecting price and buyer behavior are wrong. It simply makes the comparison more disciplined.

THE SECOND CLOCK

A sale has two dates.

01 / TRANSACTIONWhen a buyer closes

A contract reaches settlement. Some paths never get this far.

02 / LIQUIDITYWhen funds can be used

Costs, disbursement and timing determine the cash that actually arrives.

The paths that stop

There is a path a polished chart often omits. The house is listed, the deadline arrives, and it has not sold. That path belongs in the result. If we compute the average only among successful sales, we quietly answer a different question: what happened to the homes that managed to sell? We do not answer what might happen to an owner who tries.

Housing research shows why that distinction matters. An NBER study of housing market liquidity describes a sharp fall in the chance of a listed home selling during the Great Recession, alongside falling prices. The sale rate is not an incidental statistic attached to value. For an owner with a deadline, it may be the central fact.

Another path reaches an accepted offer, then turns back. A buyer's financing fails. An inspection changes the bargain. A title issue introduces a delay. The home returns to market with fewer weeks left and perhaps a different reception from buyers. A realistic simulation treats this as another state in the process, not a footnote after a successful offer.

The timing of money matters too. Closing and disbursement are stages of the transaction. The Consumer Financial Protection Bureau's description of mortgage closing explains how funds reach a settlement agent for delivery to the seller under the transaction's terms. The exact sequence varies, but a model that stops its clock at the offer has stopped before the owner's question is answered.

What a thousand futures cannot do

Running a model many times does not make its assumptions true. If buyer demand is inferred only from completed sales, unsold listings may be invisible. If yesterday's conditions are treated as permanent, a new rate or insurance shock may catch every simulated path unprepared. A million draws through a mistaken model repeat the mistake a million times.

The right response is not to abandon simulation. It is to ask better questions of it. Show the inputs. Compare the simulated distribution with outcomes the model did not train on. Change uncertain assumptions and see whether the preferred decision survives. Preserve the paths that end without a sale. Separate sale date from funds available date. Where evidence is thin, widen the range or state plainly that the model does not know.

A Monte Carlo result is not an oracle about this particular house. It is a structured conversation about the futures the model considers possible, their consequences and the evidence required to trust them. Its most useful line may not be the bright one through the middle of a fan. It may be the faint path that misses the deadline.

Return to the car. The owner could have made every sensible choice and still lived one of the paths where the closing clock stopped in time and the cash clock did not. Seeing that path beforehand would not have predicted the morning. It might have changed the schedule, the financing reserve or the commitment made against the expected funds.

The house will live one life. The others are worth drawing so that the owner can decide with more than one life in view.

Source notes