Andrews Pitchfork: What the Three Points You Pick Decide

Four widely read explanations of this tool were read in full while preparing this page. All four describe it as a drawing of three parallel lines running forward from three points on the chart. The platform documentation describes something looser than that, and the difference is not cosmetic.

The Three Lines Are a Default, Not the Definition

Ask any of the four comparables what a pitchfork is and the answer is a median line with one parallel above it and one below. That is what the tool draws when it is placed on a chart, and it is what every worked example on those pages shows.

MetaQuotes documents it differently. In the MQL5 object reference the pitchfork is a graphical object, and among its documented properties are how many levels the drawing carries and what value each one takes. A user who wants more lines than three sets a property; nothing about the object fixes the count.

The same reference documents forward and backward continuation as two independent ray settings. The drawing can be told to carry on past the anchor points in one direction, in both, or in neither, which means the familiar picture of a fork opening to the right is a configuration rather than a shape the tool has.

One of the four comes close to noticing this. It mentions in passing that another charting platform offers a version with five lines, then continues describing the three-line version as the tool. The conclusion sitting in that sentence is that the line count is a setting, and that is a different claim from the one the page makes.

Everything the Fork Says Comes From Three Points You Pick

The construction is not in dispute. Three points are placed on the chart, a median line is drawn from the first through the midpoint between the second and third, and the outer lines are drawn parallel to it through those two points. The cTrader knowledge base and the MQL5 reference agree on this, and so do all four comparables.

What none of the four supplies is a rule for placing the three points. They describe them as pivots, peaks and troughs, or reaction highs and lows at the start of a confirmed trend, and every one of those descriptions leaves the reader to decide which swing on the chart qualifies.

The most careful of the four is explicit about it. Rather than giving a rule, it advises using judgement about the first point and experimenting with alternatives until the resulting channel looks reasonable. That is honest, and it also concedes the point: the tool has no opinion about where it should be anchored.

This is the difference between a drawing tool and a computed one. A channel that reads highs and lows from a fixed window returns the same lines for everyone looking at the same data, which is why a channel computed rather than drawn behaves differently as evidence. A pitchfork returns whatever its three anchors imply.

The selection problem also has no natural stopping point. A chart holds many swings that could pass as a peak or a trough, and the number grows as the timeframe shortens, so a reader on a five-minute chart faces more candidate anchors than one on a daily chart looking at the same week.

None of that makes the choice arbitrary in the sense of careless. A reader who anchors to the swing that started the move being studied has made a defensible decision and can say what it was. The gap in the four comparables is that they present the placement as a step to be performed rather than as the assumption the whole drawing rests on.

One Chart, Two Correctly Drawn Pitchforks

Follow the construction rule exactly, twice, from two different sets of pivots on the same chart, and both results are correct. The median lines will sit at different angles, the outer lines will fall in different places, and the support and resistance levels a reader takes from them will not agree.

Neither drawing is a mistake, which is what makes the situation awkward. There is no procedure inside the tool that resolves the disagreement, because the disagreement is upstream of the tool, in a choice made before the first line appeared.

A reader who wants the same discipline applied by hand faces the same question one line at a time. Deciding which swing anchors drawing a trend line by hand is the identical judgement, made once rather than three times.

Charts can also differ for a reason that has nothing to do with the reader. Where two brokers show different highs and lows for the same period, the levels move regardless of what was drawn on them, and that half of the question belongs to the page on the Donchian channel named above rather than to this one.

The Schiff Variants Exist Because Point One Moves

Two named variants appear in the comparables. In the Schiff version the first point is relocated vertically, to the midpoint between the first and second points, so the median line runs at a shallower angle. The modified Schiff version moves it horizontally as well.

Both are described as alternatives a reader can select, in the same register as choosing a colour. They are better read as evidence about the original: a tool whose first anchor can be moved to two other places, with the drawing still considered valid, is a tool that does not depend on that anchor being anywhere in particular.

The variants are useful for exactly that reason. Drawing all three on one chart shows how much of the resulting picture came from the market and how much came from the placement decision, which is information the single drawing hides.

The comparables treat the variants as a menu and give no rule for choosing between them either. One page lists them and moves on; another describes the shallower median line as suited to steeper moves, without saying what counts as steep. So the same open question appears twice, once for where the first point goes and once for which construction to use.

A reader can close the second question the way the first one closes: by picking a construction, recording why, and treating the result as that decision rendered as lines rather than as a reading the chart supplied.

The Percentages Attached to the Median Line, and Why They Are Absent Here

Two of the four state a pair of percentages: one for how often price is said to return to the median line, another for how often it reaches the outer lines. Both pages present the pair as established, and both attribute it to the person who devised the tool rather than to any study, dataset or publication.

No primary source for either figure was obtainable. The platform documentation describes how the object is drawn and says nothing about how price behaves around it, which is the correct scope for a vendor document and leaves the claim unsupported.

The standard applied here settles what follows. Where two or more explanations repeat a number and no primary publication carries it, the number does not appear on this page in any form, including as a range or as an approximation with a hedge attached.

The absence is worth stating rather than passing over, because those two figures are the only quantified reason any of the four gives for using the tool at all. Remove them and what remains is a construction method and a set of lines to watch, which is a smaller claim than the pages make.

What the Two Platform Documents Actually Say

Two vendor documents were read for this page, and the useful thing about both is how little they claim. The table below separates what each one settles from what neither of them addresses.

The last row is the one that reframes the tool. Vendor documentation describes drawing behaviour and stops, and no other document was found that fills the gap, so the behavioural claims in circulation rest on the pages that repeat them.

Reading the Fork Without Treating It as a Forecast

What a drawn pitchfork does supply is a record of a decision, made visible. The median line states where the middle of a move would sit if the three chosen pivots described it, and the outer lines state how wide that move was at the moment the anchors were placed.

Read that way it is a frame for watching what happens next rather than a statement about what will. Price staying inside the outer lines means the swing chosen at the start still describes the movement; price leaving them means it no longer does, and neither event was predicted by the drawing.

That reading also survives the problems described above, which is its main advantage. It does not require the anchor placement to be the correct one, only that it was recorded, and it does not require any claim about how often price returns to a line, because nothing in it depends on a frequency.

It does require a reader to say in advance what would count as the frame failing. A drawing that is redrawn from new pivots each time price leaves it can never be wrong, and a tool that cannot be wrong is not telling you anything about the market.

The same reading applies to any hand-drawn geometry, including the Gann fan and its angles and a channel built from two parallels. Each one turns a chosen reference into lines, and each one is only as informative as the choice behind it.

A Four-Step Check Before Trusting the Fork You Drew

Four checks are enough to establish how much of the drawing came from the market. Each takes seconds and none needs a source.

First, redraw the fork from a different set of three pivots on the same chart and see how far the median line moves. A small shift means the swing is well defined; a large one means the levels belong to the choice rather than to the chart.

Second, add the Schiff placement and compare. Third, check the level count and the continuation settings in the object properties, so you know whether you are looking at the default or at a configuration someone changed. Fourth, note that the lines are drawn from prices on one broker feed, so anything you plan around them is measured against that feed.

What none of the four checks can do is tell you whether the tool works, because no document reviewed for this page makes a testable claim about that. A reader who wants a line to act on rather than a frame to watch is better served by trading against a line you drew deliberately, with the same question about anchors asked once.