P50 and P90 are exceedance values used to communicate uncertainty in an energy-yield estimate. Under this convention, P50 is an energy value expected to be exceeded with 50% probability, and P90 is expected to be exceeded with 90% probability under the assessment's probability model. P90 is therefore the lower, more conservative value when both describe the same distribution.
A probability statement needs a target and period
The label alone is incomplete. An assessment must state whether the quantity is energy for a single future year, an average over several years or another defined period. It must also identify the plant configuration, measurement boundary and assumptions about losses and availability. A percentage attached to an unspecified yield is not a reproducible result.
The period changes the uncertainty question. Weather variation can affect one year's production differently from a longer-period average, while systematic uncertainty in resource estimates or modelling may remain. Do not compare assessments with different periods as if their P90 values describe the same risk. Ask which uncertainty components were included and how the stated period affects them.
Exceedance differs from a percentile label
A 90th non-exceedance percentile is a high value: outcomes fall below it with the stated probability. P90 under the energy-yield exceedance convention is a low value: outcomes exceed it with the stated probability. These labels point in opposite numerical directions. Ask for the probability statement in words whenever a report uses P-values or quantile labels.
The same caution applies to short-term forecasts. A quantile for tomorrow's production interval is not interchangeable with a long-term project's annual energy exceedance value. They have different targets, time scales and uncertainty models, even if a chart happens to use similar notation.
What creates uncertainty in yield estimates
Resource data has limited coverage and may require adjustment to represent the site and longer period. The conversion from weather to electricity includes model parameters and assumptions about the installed system. Losses, availability and operating conditions also affect expected energy. Different studies can produce different results because they use different evidence or define different uncertainty scopes.
For wind, examine how the assessment represents the resource at the relevant height and the plant's power conversion. For solar, examine irradiation, array configuration and the conversion and loss assumptions. Neither technology's uncertainty should be reduced to a single generic percentage without explaining where that percentage came from.
P90 is not a guarantee
An exceedance statement describes a probability model. It does not promise that every year will exceed the value, nor does it establish project revenue or debt repayment. Changes in equipment, restrictions or operating assumptions can make actual conditions differ from those modelled. Commercial outcomes also depend on contractual and market conditions beyond energy yield.
A financing or investment discussion therefore needs the full assessment, not only its headline P50 and P90. Review the central estimate and uncertainty treatment separately. A seemingly conservative result can still rest on an optimistic central resource or loss assumption. Greater numerical precision does not remove uncertainty about the underlying evidence.
Compare studies on a common basis
Request the resource period and sources, site adaptation method, current plant design, loss assumptions and energy boundary. Confirm whether the study models a single year or a multi-year average. Ask which uncertainties are combined and whether dependence between them is represented. Do not combine several uncertainty percentages informally without a documented method.
For a portfolio, ask how dependence between sites is handled. Shared weather and common modelling assumptions can mean that project uncertainties do not cancel as an independent calculation would suggest. The sum of individual P90 values is not automatically the portfolio's P90 value.
enercast describes YAS as a tool for pre-investment yield and forecastability assessment. Confirm the specific inputs, methods and outputs available for your project before relying on a named probability value. The useful outcome is a traceable estimate with clear assumptions and uncertainty definitions, suitable for review by the project's technical and commercial specialists.