Introducing Guidance Tracking

We just launched Guidance Tracking on Marvin Labs. Guidance Tracking automatically extracts, structures, and evaluates forward-looking statements executives make in company reports, earnings calls, press releases, and other primary sources.

What is Guidance Tracking?

Public company executives often make predictions about the future of their businesses. These statements can be official guidance statements "We expect revenue for the next quarter to be $10bn," but also off-the-cuff statements, such as "We are on track to surpass a revenue run rate of $5bn for the year."

Guidance statements fall into two categories: metric-based and event-based guidance statements.

Metric-Based Guidance Statements

The most prevalent kind of guidance statements are metric-based, which include numerical targets for financial metrics such as revenue, earnings, or margins, or non-financial metrics like user growth or product adoption. The metrics can be expressed in absolute terms, relative terms, or in precise or vague ranges or terms.

Examples include:

• “We expect revenue for the next quarter to be $10 billion”
• “We expect earnings for the next quarter per share to be between $1.50 and $1.70”
• “We expect annual user growth to be 20% year-over-year”
• “We expect next quarter's margins to be flat versus the previous quarter”
• “For the next quarter, we expect a similar install rate to the current period”
• “We expect costs to grow by mid-single digits over the next three quarters.”

All these statements can be converted into testable assertions, each having a clearly defined metric, an identifiable target value or range, and a forecasted period.

• Clearly defined metric: a financial or non-financial metric, such as revenue, earnings, user growth, or margin, applying either to the company as a whole, to a specific business unit or to any subdivision.
• Identifiable target value or range: a specific number, a range of numbers, or a relative term such as "mid-single digits" or "flat."
• Forecasted period: the time frame for which the prediction is made, such as "next quarter" or "next year." These can be multiple periods, such as "next quarter and the following quarter."

Event-Based Predictions

Event-based predictions do not specify a numerical target but rather predict the occurrence of a specific event or outcome in the future.

Examples include:

• “We expect to launch our new product line in Q3”
• "We will open a new manufacturing facility by July 2025"
• "We anticipate entering the Asian market by the end of the year"

Similar to metric-based guidance, event-based predictions can be converted into testable assertions, each having a clearly defined event and a target date range.

• Clearly defined event: a specific event or outcome, such as a product launch, market entry, or facility opening.
• Target date range: the time frame for which the prediction is made, such as "by Q3" or "by the end of the year." This can include specific dates or broader time frames like "by the end of the year" or "in the next six months."

Evaluating Guidance Statements

When the forecasted period arrives, we automatically check whether the company has met its guidance, giving us a clear indication of the quality and precision of the guidance statements made by executives.

For metric-based guidance statements, we compare the actual results against the target value or range specified in the guidance statement. This typically involves checking the company's financial reports, earnings releases, or other official communications to see if the actual results align with the guidance provided. Thus, we can determine whether the guidance was "exceeded," "met," or "missed,". Sometimes, companies provide guidance statements that feature metrics that cannot be evaluated with the available data, often when making predictions about non-financial metrics or when the company does not report the specific metric in its financial statements. In such cases, we mark the guidance as "not evaluated."

For event-based guidance statements, we check whether the predicted event has occurred within the specified time frame. The source for this is still often the company's official communications, such as press releases or earnings calls. However, we also rely on external sources, such as news articles or industry reports, to verify the occurrence of the event, as they often provide validation of the event's occurrence or non-occurrence.

If the event has occurred within the predicted time frame, we mark the guidance as "met." If it has not occurred, we mark it as "missed." If the event is not reported or cannot be verified, we mark it as "not evaluated."

Guidance Tracking in Action

We are adding guidance tracking for all companies in our coverage universe. Guidance tracking is available for all subscribers. You can find it under Period Guidance (for guidance provided for the current quarter in earlier periods) and Guidance Provided (for guidance provided during the current quarter) in the Company Overview and Earnings Overview section of the app.

You can also find a quick walk-through of the feature in the video above.

Using Guidance Tracking to Evaluate Management Quality

Evaluating managerial skill from the outside is challenging yet critical for analysts. Soft factors often dominate it and lacks consistency. However, research by Baik, Faber, and Lee (2019) and others shows that executives who make more accurate public predictions tend to demonstrate stronger managerial skill and their companies outperform peers in both operations and share price.

Why does forecasting quality matter? First, making good forecasts requires rigorous analytical capabilities and synthesizing data from diverse formal and informal sources. Second, most predictions executives share publicly are shared voluntarily. When executives choose to share forecasts, it reflects their ability to distinguish high-confidence predictions from low-confidence ones and a discipline to only share high-confidence predictions publicly.

That’s why it’s not enough to track only official guidance. We need to capture every public prediction executives make about the future. Are they diligent enough to craft high-quality forecasts? Are they disciplined enough to disclose only their highest-conviction views? By evaluating the quality of big and small management predictions, we provide investors with a consistent lens into managerial skills.