How a Hotel Could Price a Rainy Weekend Using a Prediction Market

When the Weather Decides Your Weekend Revenue

Independent hotels and guesthouses in weather-exposed destinations live with a specific kind of uncertainty: two weeks out, nobody knows if a coastal or lakeside weekend will be sunny or wet, but the general manager still has to decide today whether to release rooms to OTAs at a discount, order extra breakfast staff, or hold the rate. Get the call wrong in either direction and it costs money — either empty rooms at full price, or a discount handed out on a weekend that would have sold out anyway.

This is exactly the kind of decision a prediction market is built for. A market doesn't replace a weather forecast; it turns a forecast, plus everyone else's information and opinions, into a single number — an implied probability — that can be plugged directly into a revenue decision.

The Two Numbers a Hotel Actually Needs

Cut through the noise and a discount-or-hold decision comes down to two numbers: the expected revenue if you discount, and the expected revenue if you don't, each weighted by how likely bad weather actually is. Whichever expected value (EV) is higher is the better call — not the outcome that "feels safer."

The missing input is usually the probability. Internal weather apps give a percentage, but it's a single source, updated on its own schedule, with no cost attached to being wrong. A liquid prediction market on the same question aggregates many forecasts, local knowledge, and updated information continuously, and the price only moves when someone is willing to back their view with a position.

Worked Example: A 40-Room Hotel

Take a hypothetical 40-room hotel on the Adriatic coast, one weekend in September, rack rate €140/night.

If the weekend is sunny, history says the hotel fills to 90% occupancy anyway (36 rooms), for €5,040. Discounting to €120 in that scenario just fills the last few rooms it would have sold regardless — occupancy goes to 100% (40 rooms) but revenue drops to €4,800. Discounting into strong demand is a net loss.

If the weekend is rainy, occupancy without a discount falls to 55% (22 rooms), or €3,080, as last-minute and walk-in demand dries up. Discounting to €120 in that scenario pulls occupancy up to 80% (32 rooms), for €3,840 — meaningfully better than doing nothing.

Set up the two expected values as a function of p, the probability of rain:

• EV(hold rate) = p × €3,080 + (1 − p) × €5,040
• EV(discount) = p × €3,840 + (1 − p) × €4,800

Solving for where the two lines cross gives a breakeven probability of about p = 0.24. Below roughly a 24% chance of rain, holding the rate wins. Above it, discounting wins. The manager no longer needs a gut feeling about "does it look like rain" — they need one number, and a rule for what to do once they have it.

Where the Probability Comes From

This isn't a new idea in principle. Commodity exchanges have run tradeable weather instruments since 1999, when CME Group launched degree-day futures so that utilities, retailers, and other weather-exposed businesses could hedge revenue swings tied to temperature. That market has kept growing — CME reported that trading volume in its weather futures and options surged more than 260% in 2023 alone as more firms looked to manage weather exposure. The instruments there are built around temperature indexes across a couple dozen cities, not a specific hotel weekend, so the exact contract a small hotel needs doesn't exist on an exchange.

A retail prediction market fills that gap at a smaller scale: a simple, resolvable question — "Will it rain more than X mm in [city] this weekend?" — with a live, tradeable probability attached, similar to the questions covered in our piece on hedging weather risk for outdoor events. The same logic that ski resorts use to plan around snowfall, described in our ski resort snow-hedging case study, applies to any weather-sensitive property.

Setting the Threshold and Acting On It

The practical workflow is simple: build the EV table once (it barely changes season to season for a given property), find the breakeven probability, then check the market price against it whenever a decision window opens. If the market-implied probability sits meaningfully above the threshold, discount and staff up for lower-margin walk-ins; if it sits below, hold the rate. Because the threshold and the market price are both plain numbers, this is a rule a duty manager can apply consistently, not a judgment call that changes with whoever is on shift.

The Limits of This Approach

None of this removes the underlying uncertainty — it just prices it. A market can be wrong, liquidity on a niche local-weather question may be thin, and the EV table itself rests on historical occupancy patterns that can shift. This is a framework for making a discount decision under uncertainty more consistent, not a guarantee of the outcome.

If you want to see how these implied probabilities move in real time before applying the logic to your own numbers, Outcomer lets you follow and trade weather and event markets with virtual money — a low-stakes way to get a feel for how a market price behaves before you build a real decision around it.