We are building the place where weather predictions are priced by how right they turn out to be

The founder, Eugen Mihulet, has spent close to a decade inside Romania's National Meteorological Administration, first in the Weather Forecasting Department, later in the Meteorological Satellites Laboratory. It is the kind of place where you watch the whole chain up close: the observation networks, the models, the moment a forecast leaves the building and lands in somebody's decision.

And you notice something that is hard to un-notice.

A solar producer schedules tomorrow's output on a forecast. A farmer decides whether to spray or wait. A logistics planner reroutes a fleet. When the forecast is right, it is worth many times what they paid for it. When it is wrong, it costs them a trading position, a spraying window, sometimes a harvest.

The invoice is identical either way.

This is not a failing of the people who make forecasts, it is simply how the industry grew up. Services are sold on a fixed fee or a subscription, agreed before anyone knows how the forecast will turn out, because until recently there was no practical way to do it differently: no shared ground truth in most cases, no automatic verification, no settlement mechanism that both sides could trust. The result is that the outcome risk sits entirely with the buyer, and the forecaster gets no credit for being better than the alternatives.

The numbers are not subtle: between a quarter and a half of the global economy is directly sensitive to weather conditions. Weather-related losses run at roughly 2–3% of global GDP every year, and extreme events alone cost the world economy around $2 trillion over the past decade. Research consistently finds that a single euro invested in weather services returns up to thirty in avoided losses and better decisions.

And yet meteorology remains one of the most chronically under-monetized sciences there is. The internet and decades of public funding turned most weather products into free goods. That was genuinely good for the world, as nobody should have to pay to know a storm is coming, but it also produced a market with almost no price signal, too little private investment, and a discipline that stays largely dependent on public budgets.

The field is not short of capability as there is a great deal of excellent atmospheric science, national services running world-class observation and modelling infrastructure, and private companies delivering genuinely high-quality operational services. What is missing is the mechanism that lets that capability be priced, compared and paid for.

The problem is not a shortage of skill, it is a missing market.

What if the price of a forecast depended on how right it turned out to be?

It sounds obvious when you say it out loud. It is awkward the moment you try to build it, because it requires four things at once:

• Trusted ground truth — an independent record of what the weather actually did.
• An automatic verdict — accuracy scored by algorithm, on rules published in advance.
• Settlement nobody needs to argue about — money that moves on the result, not on a claims process.
• A scoring method that respects uncertainty as forecasts are probabilistic by nature.

The fourth is the one that shapes everything else, as weather models do not produce certainties, they produce likelihoods. A genuinely excellent forecast can still be overtaken by an outcome it correctly called unlikely, and a mediocre one can look brilliant for a week.

1. You buy a forecast. Choose fixed pricing, or accuracy-based pricing. With accuracy-based pricing the amount is held in a virtual wallet rather than paid out immediately.
2. Reality settles the question. When the forecast period ends, TWP compares what was predicted against trusted reanalysis data, automatically, on rules published before the forecast was bought.
3. The money follows the result. An accurate forecast, and the provider is paid in full. A less accurate one, and a share of the amount returns to the buyer. Below an agreed threshold, compensation applies.

No claims form, no negotiation, no waiting on an account manager. The same comparison is applied to everyone, and both sides know the rules before they commit.

Once accuracy has a price, several things that were previously impractical become natural.

• Protection for buyers. If a forecast can go badly enough to hurt, a third party can take on that tail risk for a premium, i.e. a small insurance layer, built directly into the purchase.
• Protection for forecasters. A forecaster who needs certainty can give up part of the upside to secure the rest, or take payment up front on a narrower accuracy band. Confidence becomes something you can express in the contract rather than something you absorb privately.
• Democratized weather derivatives. The weather derivatives market is worth roughly €25 billion and has grown sharply as climate volatility increases, but in practice it is open only to institutions with a trading desk. TWP opens index-based weather instruments to everyone else, in a simple and gamified interface: a farm, a festival organiser, a small energy producer or a private individual can hedge against a rainy May the same way a utility hedges against a very cold winter. Democratizing these instruments is not a side feature, it is how weather risk management stops being a privilege of scale.

We are not here to replace existing private or public services, and we do not think free weather should stop being free.

TWP is infrastructure, not a competitor. An independent forecaster with a good local model, a private weather company, a research group, a data owner sitting on a valuable archive, a national meteorological service, each can use the platform to reach markets that are hard to reach today, and to try business models the current structure makes difficult. Our aim is to bring private capital into meteorology without pushing out anything that is already working.

An enabler, not a disruptor; a distinction that matters to us.

• Make it easy to find the forecast that is right for you. No forecasting model is best everywhere, or best all year. One may perform well over flat terrain and struggle in a mountain valley; another may be strong in summer convection and mediocre in a winter inversion. Working out which model is genuinely best for your location, your variable and your season is a research problem in itself, and few have the resources to run it. TWP runs it for you: the platform tracks how each model has actually performed under conditions like yours, and puts the strongest candidates in front of you.
• Make accuracy count in what a forecaster earns. Performance measured against reality, verified automatically, reflected in the settlement, so that being better at forecasting shows up in the revenue, and buyers can tell providers apart on evidence rather than reputation.
• Open weather risk management beyond institutions. Not just to organisations with a trading desk, but to the farm, the small energy producer and the individual, the people who carry weather risk today with almost none of the tools to manage it.
• Turn dormant assets into income. Datasets, models, algorithms and research results that currently sit unused become products, with their owners paid when they create value and without their intellectual property leaving their control.
• Let the public take part, not just watch. Weather risk is currently something the public absorbs and institutions trade. On TWP anyone can choose to stand on the other side of it: taking on a share of forecast-uncertainty risk, or of the long-term risk of adverse weather, in exchange for a share of the returns. Capital contributed this way funds better models and better observation, and cover losses related to the inherent forecasts uncertainty. It also does something very valuable: it gives a far wider public a direct, personal reason to care how good a forecast actually is.

And underneath all five, the long-term objective: to help make meteorology a field that can sustain itself, where the value it creates for the world circulates back into the science that produces it.

TWP is being built now. The core mechanics, e.g. the marketplace, secure models training, accuracy-based pricing, are in development, starting with renewable energy forecasting. We are working alongside early adopters, including data owners, forecasters and developers who will shape what the platform becomes.

If you forecast, hold data, build models, or carry weather risk, we would like to talk to you. We are also speaking with investors and partners who want a position in the infrastructure of the weather economy as it forms.