How Trade The Path works, step by step

Six steps take a ticker from three years of daily closes to one buy price, one sell price and one date — with the evidence for each printed beside it.

Last updated 2 August 2026

1. Three years of the stock's own history, cached

Everything starts from daily open, high, low and close bars going back three years. They are stored the first time a stock is looked at and reused afterwards, so a chart is drawn from our own cache and only the newest missing days are ever fetched again. That matters for more than speed: every claim on the site is re-measurable against the same bars a month later, which a service that re-downloads and re-fits invisibly cannot promise.

2. Find the turning points, not the noise

A day counts as a swing high only when its high is the highest across the ten trading days either side of it, and a swing low only when its low is the lowest. Most days fail that test, which is the point — what is left is the handful of turns that actually mattered on that chart.

3. Fit a ceiling and a floor

Two lines are fitted independently: one through the swing highs, one through the swing lows. The slope of each comes from the median of the slopes between every pair of points, so a single freak spike cannot tilt it, and the line is then pushed out to touch the most extreme swing so it is a true wall rather than an average. The fit is done in log-price space, which is why a channel on a $9 stock and a $900 stock mean the same thing proportionally. The lines are not forced parallel: a channel that widens or narrows gets drawn that way.

Four lookbacks are tried — three years, two years, eighteen months and one — and the one price has actually respected is kept, preferring the more recent window when the fits are close.

4. Project both lines 3 months forward

Each line keeps its slope, but the forward projection is drawn at half of it. That is not caution for its own sake: measured across the stored history, about half of a channel's fitted drift is what actually carried into the following six months. Drawing the full slope forward produces a flattering picture and a target price nobody reaches.

5. Turn the geometry into one buy price and one sell price

Buying anywhere inside a channel measured no better than buying at random, so the site does not do it. What measured differently was the overshoot: price that has dropped below its own floor and has already turned back up. So the buy level sits just under the projected floor, and no call is made until price has closed above where it was five and ten sessions ago.

The target is then chosen rather than assumed. Every comparable entry in that stock's history is replayed and, for 8%, 10%, 12% and up, three things are measured: how often the gain arrived inside three months, how long it took, and how far price went against the entry first. The published target is the largest gain that still lands often enough and that the projected ceiling leaves room for. The date is how long comparable setups took — a typical duration, not a deadline.

6. Publish the evidence, and the failures

Each target carries the number of comparable setups behind it and the share of them that reached it. Where a stock has too few to be worth anything, the market-wide rule stands in and the page says so instead of quietly borrowing another stock's confidence. If the fit is not clean enough, if the stock has never overshot and recovered, or if no target above 8% is likely enough, nothing is published and no alert is sent.

Every call then appears on the track record whether it worked or not, and a live call is cut short when the run loses its 50-day trend and stays below it — you are told the day that happens rather than discovering it in a chart three weeks later.

See it applied: today's picks, the three-month predictions, or search any ticker from the stock finder.