Reading a distribution: learn to spot a coin flip

Read the whole distribution before summarizing a setup.

What followed, grouped by size

After a scan finds every time a condition was true, there is one honest way to show what followed: group the forward moves by size and draw how many landed in each group. Big losses on the left, big gains on the right, the ordinary middle in between. That chart is the distribution, and it is the whole result in one shape.

The even split is a finding

A roughly even pile on both sides of zero is one possible shape. Inspect what it contains before calling it a coin flip. It does not by itself establish useful directional information. Compare magnitudes, uncertainty and the eligible baseline; an even sign split can still have unequal outcome sizes. Learning to recognize it protects you from every backtest thread on the internet, and it is the payoff of the base-rate habit.

Skew versus one lucky day

When a distribution does lean one way, look at how. A broad lean, where most of the pile sits on one side, is a pattern. A symmetric pile plus one enormous outlier is one lucky day wearing a pattern’s clothes: remove that single match and the edge disappears. The chart makes the difference visible in a way an average never can.

A percentage needs its count

“65% finished up” sounds like the same fact whether it came from 20 matches or from 2,000 matches. It is not. With 20 matches, a 65% split is well within what coin flips produce by luck. That is why every rate on EdgeDepth travels with its count, and why the verdict sentence under a scan is really this chart translated into words, excursions and all.

A percentage without its count is a story. “65% finished up” from 20 matches and from 2,000 matches are different facts, and only the count tells you which one you have.

See it live

Run a scan and read its distribution chart.

Every published study carries the same charts over a pinned result, so you can read one before running anything.

Related: Base rates · MFE and MAE