What Is an Andrews' Pitchfork?
Andrews' pitchfork is drawn from three chosen pivots: a line runs from the first through the midpoint of the other two, with parallel lines from each of those. The central median line is the tool's actual claim, since price is expected to return to it more often than to either outer line.
An Andrews’ pitchfork looks like a channel and is not one. Its subject is the line up the middle, and everything worth knowing about the tool concerns whether price returns to that line.
How it works
Choose three pivots. Conventionally a significant low, the high after it, and the higher low that followed — labelled A, B and C.
The median line runs from A through the midpoint of B and C. That is the handle of the fork and the tool’s central object.
Two parallels complete it, one through B and one through C, forming the outer tines that give the drawing its shape.
What the claim actually is
Alan Andrews’ stated proposition was specific: price returns to the median line a large proportion of the time. The outer tines are context; the median line is the assertion.
That is mean reversion, drawn. The median line is roughly the middle of the recent swing structure, and “price returns to the middle” is a claim that can be examined directly.
Which is a useful reframing. Asking “does this market revert to its recent middle?” is answerable; asking “does the pitchfork work?” is not, because the answer depends on which three points you chose.
A worked example
Take this site’s shared series. 95% of bars sit below a prior peak, the maximum drawdown is 3.76%, and the longest stretch below a prior peak runs 73 bars.
That 95% figure is the interesting one here. In a series that rises over time, price is almost always below its recent high — which means a median line drawn through the middle of a swing has price beneath it far more often than above.
So “returns to the median” is partly a statement about where the median sits. Draw the line through the middle of a range and price will cross it frequently, because that is what a middle is.
The test that matters is different. Not whether price touches the line, but whether touching it predicted anything about the next move — and the drawing itself cannot answer that.
Choosing the three points
A is the anchor and does the most work. Move it and the whole fork rotates, so the choice of starting pivot determines the tool’s entire output.
B and C set the width. A wide swing produces a wide fork whose tines sit far from price; a narrow one produces a tight fork that price crosses constantly.
Convention says use the most recent clear swing, which helps — but “clear” is a judgement, and two traders looking at the same chart will disagree about which swing qualifies.
Which is the honest limitation. Three chosen inputs is three opportunities to fit, and a tool with three degrees of freedom can be made to describe almost any chart after the fact. The discipline that makes it useful is choosing the pivots before the move rather than after.
The original data
On this site’s shared series: 95% of bars sit below a prior peak, maximum drawdown 3.76%, longest stretch below a prior peak 73 bars, finishing +3.61%. Median bar range 0.493, direction runs averaging 2.01 bars. A round trip costs 0.0098, about 2% of the median bar.
The 73-bar stretch is what a median line has to survive. A tool premised on returning to the middle has to be held through periods where the market stays on one side of it for a long time.
And every touch acted on costs the spread. In a series where direction changes every two bars, a line through the middle is touched often, and treating each touch as a signal is expensive well before it is informative.
How it compares to a channel
A parallel channel is drawn from two pivots, so it has fewer degrees of freedom and is harder to fit to the past.
A pitchfork adds a third point and a median line, which is both its contribution and its extra flexibility.
A Gann fan shares the fitting problem differently — one pivot, many angles, with the added complication that its angles depend on chart scale, which a pitchfork’s parallels do not.
Of the three, the pitchfork makes the most specific claim. Channels describe containment; a pitchfork asserts a tendency to return to a particular line. Specific claims are better, because they can be wrong in a way you can notice.
When it fails
The characteristic failure is a trend that never comes back. The fork is drawn, price leaves the median line and keeps going, and each attempt to fade it back toward the middle loses. The tool did exactly what it says — it identified where the middle was — and in a trending market the middle is behind price and stays there. A mean-reversion drawing applied to a trending market is not malfunctioning; it is being used against the only condition that reliably defeats it.
A second failure is redrawing from new pivots whenever price escapes, which fits the tool to history.
A third is treating the outer tines as targets. They are parallels, not levels anything reaches for.
A fourth is using it on a timeframe where the swings are smaller than the bars, which makes every line meaningless.
And a fifth is forgetting that you chose the pivots. The fork describes a structure you nominated, and nominating a different one produces a different answer from identical data.
Related
Trend lines covers the simpler construction this is built from. Mean reversion covers the claim the median line is actually making. And Gann fan covers another multi-line tool drawn from chosen pivots.
Strip away the shape and this tool makes one claim: price tends to come back to the middle of its own recent range. That is a mean-reversion argument, it is testable, and it is a good deal more modest than the confident geometry on the screen suggests.
— Michael Whitman, from this video
This page is educational, not financial advice. Test every idea on your own charts before risking money.