The Chart Pattern Bundle — 10 Guides, Every Claim Measured
You have been shown a head and shoulders and told price falls. A triangle and told the trend continues. A cup and handle and told to buy the handle. Then you went looking and found them everywhere, or nowhere, depending on how hard you squinted - and no way to tell whether any of it was true.
This bundle is ten complete guides, 450 pages, that measure the claims instead of repeating them.
Across 35,944 bars of real market history on ten instruments and three timeframes, every chart pattern that promises a direction was raced against a control: does price travel one average true range the promised way before one average true range the other way? 13 patterns made a claim. About 4,507 cases were tested. The corrected bar for 13 tests is a z of 2.891.
Nothing cleared it.
The previous bundle measured eighteen candlestick patterns the same way and found one survivor - pointing the wrong way. Thirty-one pre-registered tests across two waves, one borderline result.
WHAT MAKES THESE DIFFERENT
Half these shapes have a published algorithmic definition - Lo, Mamaysky and Wang, Journal of Finance 2000 - used exactly as written. Half do not, and are defined in this series. Every guide says which it is holding on its own first page. No other source draws that line, and it changes what you should believe.
Six of the 19 shapes measured carry no direction at all. Triangles, broadening formations and rectangles promise nothing in any published definition, so three of these guides are about patterns the literature never made a claim for. Those three end with a use rather than a caution.
And under eight of the ten sits a number nobody prints: every count is a count of a smoothed line, and how hard you smooth moves the counts from twice to 32 times. By hand it is simply your zoom level.
WHAT'S INSIDE
Head & Shoulders (49pp) - The most famous pattern in the subject, and the one place charting has a peer-reviewed definition to test.
Double Tops & Bottoms (49pp) - A condition in the published rule that almost no source repeats.
Triangles (47pp) - Two of the three triangles you were taught cannot satisfy the published rule.
Broadening Formations (45pp) - The rarest published shape here, and a measured answer to an argument.
Rectangles (48pp) - The strictest levelness test in the published set, and the condition that fails silently.
Wedges (45pp) - The commonest shape in the wave, and the clearest null in it.
Flags & Pennants (43pp) - The largest edge in two waves, reported with everything that qualifies it.
Cup & Handle (42pp) - Why one pattern's count is far less stable than any other here.
Rounding Tops & Bottoms (42pp) - The one shape found by fitting a curve, and the trap that comes with it.
The Three Methods (40pp) - The closing guide, and the one lesson worth more than any pattern.
ONE MEASURED FINDING FROM EACH
Head & Shoulders: It reached its promised downside 4.84 points LESS often than a random bar, on 457 cases - and the inverse leans the same way, which a directional shape cannot do.
Double Tops & Bottoms: How far apart must the two tops be? The paper says 22 bars. Nobody else says anything, and moving it takes the count 9.1-fold on identical bars.
Triangles: It needs strictly falling highs AND strictly rising lows, so an ascending triangle fails the first and a descending one the second. No published count covers either.
Broadening Formations: Expansion against contraction, settled by counting: 303 triangles against 238 broadening formations on the same bars, and triangles win at every smoothing tested.
Rectangles: Three conditions, and the third - the lowest top must sit above the highest bottom - is quoted almost nowhere. Without it two overlapping bands qualify, which is not a range at all.
Wedges: 679 cases - the largest sample any test here was given - and an edge of +0.78 points, under a third of one standard error. Not a failure to measure; a measurement of nothing.
Flags & Pennants: +9.62 points on 213 cases, z = 2.81. The corrected bar for 13 tests is 2.891, so it misses by 0.081 - and the guide says so on the cover.
Cup & Handle: It needs SEVEN turning points where every other shape needs five, and a window holds 7.03 on average. The count swings 32 times across the smoothing range.
Rounding Tops & Bottoms: A straight line fits a curve perfectly. Requiring only the fit score gives 13,771 detections; adding a bend requirement removes 50.4 per cent of them. Half of what a fit alone finds is a trend.
The Three Methods: +5.73 points on 13 cases and +0.78 points on 679 cases have the SAME z. One edge is seven times the other and they carry identical evidence.
WRITTEN FOR SOMEBODY WHO KNOWS THE BASICS AND NO STATISTICS
Every guide opens with a two-page plain-English primer: average true range, the control, the sample size, the standard error, z, the correction, smoothing and thresholds - all defined before the first number appears, with no mathematics anywhere. If you can read a candlestick chart you can read these.
WHAT YOU'LL BE ABLE TO DO
Mark every one of these shapes by a written rule instead of by eye, and get the same answer twice.
State the thresholds you used, and know what moving each one costs.
Read a published pattern statistic and know in four seconds whether it can mean anything.
Tell the difference between "this was tested and failed" and "this was never tested" - three of these guides are the second kind.
Apply the mirror test, which costs nothing, needs no statistics, and settled more of this wave than any significance test in it.
AN HONEST NOTE
These are clues, not crystal balls. This bundle contains no win rates, no backtest results and no promises about what you will make, because nobody can verify those and you should not trust anyone who prints them. It also does not claim chart patterns are worthless - what these shapes record about where price turned is exact, and several of these guides are built on that. It is the directional claims that did not survive. Educational, not financial advice.
Instant download. 10 guides, 450 pages, PDF.