AG2 Weather Data Buyer’s Guide

Start with the business decision, not the dataset.

Most buyers are trying to do one of four things:

• Build predictive models
• Trade weather-sensitive markets
• Improve operational decisions
• Research historical weather impacts

Once we understand the decision you are trying to improve, we can usually map you to the right AG2 dataset quickly.

AG2 provides a broad set of weather data products, including:

• Historical observations
• Current conditions
• Deterministic forecasts
• Probabilistic forecasts
• Precipitation history
• Renewable energy weather data
• Agriculture-specific weather data
• Historical forecast archives

We also support custom datasets, model data extraction, and delivery through APIs, Snowflake, Databricks, and other workflows depending on the use case.

Historical data describes what actually happened and is best for research, model training, validation, and backtesting.

Current data describes what is happening now and supports monitoring, operations, logistics, and real-time decision-making.

Forecast data describes what is expected to happen and supports planning, trading, energy, agriculture, risk management, and operational decisions.

Recommended product: History on Demand or Cleaned Historical Observations

These are typically the first datasets we recommend for machine learning, AI, and quantitative research because it represents observed weather rather than forecast model output.

• Historical observations back to 1990
• Cleaned and corrected records
• 8,000+ stations
• Consistent data for training, validation, and backtesting

For more details about the Cleaned Historical Observations dataset, see the Cleaned Historical Observations documentation .

If the timeframe desired is prior to 2016 and the parameters desired are available with History on Demand, we recommend that. Otherwise, Cleaned Historical may be the best fit.

Recommended product: Historical Forecast Archive

Use this when your analysis depends on forecast information rather than actual observed weather.

• Forecast skill studies
• Trading strategy backtesting
• Model verification
• Historical decision simulation

This is the right category when the question is not just what happened, but what a market participant could have known at the time.

For most machine-learning applications, AG2 recommends training on observations because observations are the best available truth dataset.

Historical forecasts are useful when measuring forecast skill, recreating historical decision environments, or testing a strategy that depends specifically on forecast information.

A simple rule of thumb: train on observations when you need truth; use historical forecasts when you need to understand what was knowable at the time.

Recommended products: Enhanced Current Conditions and History on Demand

These products are useful when observation stations are sparse or unavailable and the user needs weather data for a specific point location.

• Renewable energy assets
• Transmission infrastructure
• Agriculture
• Insurance
• Logistics
• Site-specific operational analytics

Recommended product: Enhanced Forecast API

Enhanced Forecast is AG2’s flagship forecast product for users who need forecast quality, resolution, and reliability. Plus, it's the most skillful forecast out there, according to an independent 3rd party evaluation.

• Proprietary skill-weighted forecast blends
• AI-enhanced guidance
• High-resolution data
• Hourly forecasts out to 15 days

Core is the entry-level weather package and is best suited for basic applications that need standard weather data.

Enhanced Forecast is the premium forecast product and is the better fit when forecast quality, resolution, and skill matter.

For data buyers, the practical question is: do you need basic weather context, or do you need the strongest forecast signal available for decisions, models, or trading workflows?

Recommended product: Probabilistic Forecast API

Probabilistic forecasts help users understand the range of possible outcomes rather than relying on a single deterministic forecast.

This is valuable for traders, utilities, risk managers, energy planners, and anyone who needs probabilities, percentiles, scenarios, or risk-aware planning.

Recommended starting point: Enhanced Forecast, Probabilistic Forecast, and Renewables API

The value is not just better weather data. The value is making better decisions faster and turning weather intelligence into a repeatable market advantage.

Raw model data can be useful, but data alone rarely creates value.

AG2 combines weather data, meteorological expertise, forecast interpretation, custom datasets, flexible delivery, and direct access to experts who help clients act on the information.

This matters especially for firms that know weather affects their market but do not want to build a full internal meteorology operation before extracting value.

Perfect. That is how many conversations start, and how many of our customers begain their journey.

Tell us what you are trading, forecasting, optimizing, researching, or trying to protect against.

We will help map your problem to the right weather solution, whether that is an existing API, a historical archive, a custom extraction, or a tailored dataset.