P10, P50, and P90 are percentile values that represent three levels of confidence in a forecast or estimate. The “P” stands for percentile, and the number tells you the probability that the actual outcome will meet or exceed that estimate. P90 is the conservative (low) estimate, P50 is the middle estimate, and P10 is the optimistic (high) estimate. These three values together give you a picture of how much uncertainty surrounds a prediction, whether you’re forecasting oil reserves, solar energy output, or the cost of a construction project.
What Each Value Means
These percentiles are defined using what’s called “probability of exceedance,” which simply means the chance that reality will beat the estimate. Here’s how each one works:
- P90 is the conservative figure. There’s a 90% chance the actual result will equal or exceed this number. It’s the “we’re almost certain we’ll get at least this much” estimate.
- P50 is the median estimate. There’s a 50/50 chance the actual result will come in above or below this number. It represents the most likely outcome.
- P10 is the optimistic figure. There’s only a 10% chance the actual result will reach or exceed this number. It represents the best-case scenario.
A quick example: if a solar energy project has a P50 of 10,000 kWh per year, that means there’s a 50% likelihood the system will produce more than 10,000 kWh in any given year. A P90 of 10,000 kWh would mean the system is likely to produce over 10,000 kWh about 90% of the time, a much more confident prediction.
Why the Numbers Seem Backwards
The most confusing thing about this system is that P90 is the lowest number and P10 is the highest. That feels counterintuitive until you remember what the number after “P” actually measures: the probability of exceeding the estimate. A high probability of exceedance (90%) means you’re setting the bar low enough that reality will almost always clear it. A low probability of exceedance (10%) means you’re setting the bar so high that only the best outcomes will reach it.
Think of it like setting a goal for a road trip. P90 is saying “I’m 90% sure I’ll drive at least 200 miles today.” P50 is “there’s a coin-flip chance I’ll drive at least 350 miles.” P10 is “only if everything goes perfectly will I hit 500 miles.” The confidence goes up as the estimate goes down.
One Term, Two Conventions
There’s an important wrinkle: not every field uses P90 the same way. In the oil and gas industry, renewable energy, and project management, P90 typically means a 90% probability of exceedance (the conservative, low number). But in some statistical and environmental contexts, the “90th percentile” means the value that 90% of observations fall below, which makes it the high number. Water quality testing, for instance, uses the 90th percentile to flag when more than 10% of samples exceed a safety threshold.
These two conventions are essentially mirror images of each other. A P90 on an exceedance curve equals the 10th percentile on a standard cumulative distribution, and vice versa. If you’re reading a report and the meaning isn’t clear, check whether a higher P-number corresponds to a higher or lower value. That tells you which convention is in use.
How They’re Used in Oil and Gas
The oil and gas industry was one of the earliest adopters of this framework, using it to classify how much petroleum a reservoir is likely to produce. The Society of Petroleum Engineers defines three reserve categories tied directly to these percentiles:
- Proved reserves (1P) correspond to P90. There’s a 90% probability the well will recover this volume or more. This is the number companies can report with the highest certainty.
- Proved plus probable reserves (2P) correspond to P50. There’s a 50/50 chance of recovering either more or less than this amount.
- Proved plus probable plus possible reserves (3P) correspond to P10. Only the top 10% of wells will recover this volume or more.
As one industry presentation puts it, the only difference between proved, probable, and possible reserves is uncertainty. The underlying geology doesn’t change. What changes is how confident you are in the estimate.
How They’re Used in Renewable Energy
Banks and investors financing wind and solar projects rely heavily on P50 and P90 values to assess financial risk. The P50 estimate represents the expected annual energy output in a typical year. The P90 estimate represents the output that the project will likely meet or exceed in 9 out of 10 years.
Lenders care about P90 because they need to know the project can service its debt even in a below-average year. If a wind farm’s P90 energy output still generates enough revenue to cover loan payments, the project is considered financeable. The gap between P50 and P90 also reveals how variable the resource is. Wind tends to have a wider spread between P50 and P90 than solar, because sunlight is generally more predictable than wind speed from year to year.
Calculating these values requires many years of historical weather data. The longer the dataset, the more reliably the P50 and P90 estimates capture real-world variability.
How They’re Used in Project Cost Estimation
Government agencies and large organizations use the same framework to set budgets. Australia’s Department of Finance, for example, requires a P50 confidence level for early-stage project cost estimates and P80 for final approval. A P50 cost estimate means there’s a 50% chance the actual cost won’t exceed that number. A P80 estimate means there’s an 80% chance the cost will stay within budget.
This is not the same as “cost plus or minus 20%.” It’s a statement about confidence. And the relationship between confidence level and cost isn’t linear. Getting from P50 to P80 might add a modest contingency to the budget, but pushing toward P100 (certainty that you’ll never exceed the budget) would require an impractically large contingency that accounts for every conceivable risk. That’s why organizations pick a practical confidence level rather than aiming for P100.
On the flip side, using a bare estimate with no contingency at all might land you around P15 confidence, meaning the actual cost will exceed your budget roughly 85% of the time.
What the Spread Tells You
The gap between P10 and P90 is just as important as any single number. A narrow spread means the outcome is relatively predictable. A wide spread signals high uncertainty. If a project’s P90 energy estimate is 8,000 kWh and its P10 is 12,000 kWh, the range of likely outcomes is manageable. If the P90 is 5,000 kWh and the P10 is 15,000 kWh, you’re dealing with far more risk.
P50 sits in the middle and, for a symmetrical (normal) distribution, equals both the median and the mean. In skewed distributions, which are common in fields like oil exploration, P50 still represents the median but can differ from the mean. This matters because the mean can be pulled toward extreme values, while P50 stays anchored at the midpoint where half the outcomes fall above and half below.