How to Read Kunkafa Forecast Results and Accuracy

You open Kunkafa’s results page and see three columns of numbers. Your eye goes straight to the biggest percentage. Start somewhere else.

Open Kunkafa forecast performance results alongside this guide. The figures update daily, so the examples below use invented round numbers. The definitions follow Kunkafa’s published methodology as checked on 7 September 2026.

Start with what Kunkafa forecast

A forecast describes four things. How far a price could move, which direction, how long it has, and Kunkafa’s confidence in reaching the level. Seventy experts stand behind it, and both directions are always shown.

Those details set what counts as a result. A level reached after the duration has ended does not answer the original question.

Imagine an invented market at 100, with an upward level at 103 over ten days. It reaches 103 on day four and closes at 99 on day ten. It reached the level but did not finish in the upward direction. Now imagine it closes at 102 without ever touching 103. It finished in the upward direction but did not reach the level. The record measures whether the level was reached within the duration.

The three columns, in order

• Forecasts evaluated. How many forecasts finished inside the window and could be counted at all.
• Confident enough. How many of those cleared the published confidence filter. This is the number the rates are counted on.
• Reached the level. How often the price travelled the whole distance inside the duration. The harder question.

The first two columns are counts. The last is the target-reaching rate, a fraction of the second column, not the first.

Read the count before the rate

Suppose an invented window holds 1,000 evaluated forecasts, and 300 are confident enough for the filter. If 240 of those 300 reach their levels, the rate is 80%: 240 divided by 300. It does not mean 800 of all 1,000 forecasts reached their levels.

That is why Kunkafa puts the count beside the percentage. A high rate from a narrow subset answers a different question from a rate across the whole group. Read the denominator first, and the rate will mean what you think it means.

What Kunkafa’s confidence filter changes

The confidence filter sets which forecasts enter the selected record. Raising it leaves fewer forecasts to count. It is not a promise that the next forecast will succeed.

The public summary labels the confidence it counts at. Keep that setting in view when you compare results. The threshold itself is not the observed success rate.

In the app the same control reads “Only show forecasts Kunkafa is at least this confident about”. It filters the list, and the record beside it is recounted on the same setting. The rate you read always belongs to the forecasts you are looking at.

Why both directions remain useful

A market can reach an upward level and a downward level within the same duration. In the invented example it could visit 103, fall to 97 and finish at 101. Kunkafa keeps both directions visible so you weigh the range of paths rather than treat one as certain.

Compare windows carefully

Read the update date and the named window together. A one-day result describes a different collection from a longer window. The windows overlap, so adding their counts does not produce a count of unique forecasts. A forecast still running has not yet supplied its outcome.

Before comparing two snapshots, check that the filter, the window and the outcome definition match. A shift in one headline percentage does not by itself say whether the mix of markets, durations or selected forecasts changed.

A practical sequence: choose your market and duration, look at the up and down levels, then open the record and read its count and definitions before its rates. Write down what would count as reaching the level and when the duration ends. The Kunkafa market forecasting methodology defines every term, and the Kunkafa market forecast app shows the record beside a live forecast.