How to read a P10–P90 range

Every Coneview forecast is bounded by two numbers: P10 and P90. They are percentiles of the 20,000 futures we simulate for an instrument. P10 is the level only 10% of those futures finished below. P90 is the level only 10% finished above. Between them sits the middle 80% of everything the model considers plausible.

What the edges claim

• •P90 (upper edge). Roughly one simulated future in ten finished above this. It is an unusually strong outcome — not an impossible one.
• •P10 (lower edge). Roughly one in ten finished below it. Likewise unusual, likewise entirely possible.
• •The band between.About four times in five, the price should finish somewhere inside. That "about" is doing real work — it is a claim we test, not a promise we make.

The mistake almost everyone makes

The edges are not a floor and a ceiling. Price is not trapped between them, and reaching P90 is not a signal that the move is over. One time in ten — which, over a year of daily forecasts, is many, many times — the market simply keeps going. Treating a percentile as a barrier is how people find themselves stunned by an outcome their own model told them to expect roughly monthly.

Nor is the band a forecast of direction. A wide symmetrical cone around the current price is the model saying, quite explicitly, that it does not know which way this goes.

Width is the message

The single most useful thing about a P10–P90 range is not where its edges sit but how far apart they are. A band spanning 0.4% of price over the next hour and one spanning 12% over the next month are describing completely different situations, and the second is not "more bullish" or "more bearish" — it is simply less knowable.

Width scales with the square root of time, so a forecast four times further out is roughly twice as wide. It also scales with the instrument's own volatility: a quiet index and a volatile altcoin with the same midpoint will have wildly different bands. Comparing widths across timeframes and instruments tells you where the uncertainty genuinely lives.

When direction is a coin toss

Coneview reports the odds of finishing higher alongside the range. Frequently that number sits near 50%, and we say so rather than manufacturing a lean. When direction is noise, the range is the forecast: it tells you how much room the market plausibly has, which is information, even though it is not the information most people wanted.

A forecast that always has an opinion is not a better forecast. It is a less honest one.

Read next: what a forecast cone is and what makes a probability trustworthy.