Why trend channels work, and when they do not

A channel is not magic geometry. It works for one reason and fails for four, and both lists are short enough to check.

Last updated 2 August 2026

The one reason it works

A trend line is a record of where real buyers and sellers repeatedly changed their minds. Nothing about the line causes anything; it is a summary of behaviour that has already repeated three, five, eight times on that chart. Trading it is a bet that the same participants behave the same way at the same relative price — a much weaker claim than predicting the future, and weak claims are the ones that survive.

The useful consequence is that a channel gives you a price to act at, decided before you are emotionally involved. Most trading damage is done by deciding what to do while watching the quote move.

What the measurements actually said

Two findings shaped this site, and one of them was negative.

Buying in the lower part of a channel — the classic advice — reached an 8% gain within three months about as often as buying anywhere else did. There was no edge in it. What did separate was the overshoot: entries where price had fallen through the floor and then closed back above its recent levels reached 8% roughly four times in five, in about five weeks, with price typically moving less than 6% against the entry along the way.

That is why the buy level sits below the floor rather than in the channel, and why nothing is called until price turns up. The rest of the site is bookkeeping around that one measurement.

Fit quality is the whole game

A line drawn through two points is not evidence of anything. The fits are scored on how many swing points each line touches, how tightly price has held to them, and over how long a span — and stocks that score badly get no published numbers at all. If a chart is chaotic, the honest output is silence, and a service that always has something to say about every ticker is telling you something about itself.

The four places channels fail

Regime changes. A takeover, a fraud, a guidance cut or a product failure makes three years of price behaviour irrelevant in one session. Geometry has no view on news.
Structural decliners. An option-income ETF that distributes most of its return will drift down by design. Its channel points down and its "buy level" keeps following price lower. Price-only analysis reads a working product as a falling stock.
Illiquid and thin charts. Wide spreads and gappy bars produce swing points that are artefacts of the tape, not of anyone's decision.
Everything failing at once. If a portfolio is one theme in several tickers, every channel in it breaks the same week. That is not the method being wrong; it is diversification being absent, and no chart will warn you.

How to check any of this

The track record lists every call with the result, including the expired ones, and separates live calls from backtested ones. The data explorer shows the stocks that were considered and rejected, with the test each one failed. If a claim here is wrong, those two pages are where it will show.