A plain guide to patterns that describe what gold did, not what it will do.

How a page of this site is built

Each pattern page opens with a chart drawn from real daily bars, directly under the heading. A candle that closed above its open is hollow and one that closed below is filled; the pale band is the figure; dashed lines are the lines the definition uses, the floor and ceiling of a rectangle or the lines of a flag; and a labelled circle is a point the rule used. Under the chart a caption names the instrument, the source and the date the history was read.

Below the chart the page defines the pattern in plain words, then repeats the definition in the rule's own numbers, then says what the pattern cannot tell a reader. A count follows: how many times the rule found the model in ten years on the two gold series. Nothing on a page is a signal and no page promises what gold will do next. One page ends with a finding instead of a chart, and the reason is given there.

Two gold series, both in dollars

The history behind the site is two series. The first is COMEX gold futures (GC=F), priced in dollars per troy ounce, 2,437 daily bars after cleaning. The second is the SPDR gold fund (GLD), a fund listed on a stock exchange and priced in dollars per share, 2,513 bars. Both run from 7 November 2016 to 5 November 2026 and were read once, on 6 November 2026.

They are different objects. A futures contract and a listed fund follow the same metal but print different bars, which is why a rule can find a pattern often on one and rarely on the other. The site draws no rupee price. A chart in rupees would mix the move in gold with the move in the exchange rate, and the patterns here are meant to describe gold itself. Each chart says which of the two it is drawn on.

Eight models, eight written rules

The site covers the weekend gap, the bear flag, the rectangle, the marubozu, the tweezer top, the tweezer bottom, the three white soldiers and the three black crows. Each has a rule written before the search. The gap needs a Monday open at least 1 ATR from Friday's close. The bear flag needs a pole of at least 8 ATR within 15 sessions and a flag of 5 to 20 sessions that gives back 10 to 50 percent of it. The rectangle needs two highs and two lows within 1 ATR of each other and a box at least 3 ATR high.

The marubozu needs a candle of at least 1.5 ATR with a body of at least 95 percent of the range. The tweezers need two candles whose highs or lows differ by at most 0.1 ATR. The soldiers and the crows need three candles of at least 0.6 ATR each with strict opening and shadow conditions. ATR is the average true range of the last 14 sessions, and a rule written in ATR means the same thing in a calm year and a wild one.

What the counts say

In ten years the rules found 64 weekend gaps (6 on futures, 58 on the fund), 3 bear flags (all on the fund), 14 rectangles (7 and 7), 21 marubozu candles (19 on futures, 2 on the fund), 7 tweezer tops and 2 tweezer bottoms, and no soldiers and no crows on either series.

The spread is itself informative. A gap of a full ATR is a visible event but there were 64 of them; a bear flag, which needs a violent fall followed by a gentle recovery, appeared three times in a decade. The counts say how demanding each definition is on these two series. They do not rank the patterns, they say nothing about how often a pattern was followed by a move, and they would change with other instruments or other thresholds. The fund and the futures follow the same metal, so the counts are not independent either.

Why a pattern counts only when a close completes it

A candle that is still trading is not yet a candle. A session that looks like a marubozu at midday can end with a long shadow, and a break through a flag's lower line can reverse by the close. For that reason every rule counts a model as formed at a close: the last candle's close for a candle model, a close below the flag's lower line within 3 sessions of the last flag bar, a close outside the box of a rectangle within 25 sessions of the last pivot.

The session on which the rule was satisfied is marked on the chart, and the chart stops there or shortly after. Nothing drawn after that session was used to decide anything. A reader who finds a pattern on a live chart and acts before the close is acting on something that may not exist at the close, and the school of mistakes this site teaches against starts there.

The page with no chart

The page about the three white soldiers and three black crows has no chart. Rules 3W-1 and 3W-2 were run over every bar of both series and found no occurrence in ten years. The site could have loosened the rules until an example appeared, or borrowed one from another market, and either would have put a chart on the page. Neither would have been honest.

The finding is narrow. It says that these rules, written in advance, produced zero on gold futures and the gold fund. It does not say that three long candles in a row never happen on gold, and it does not say they mean nothing. Readers who remember such runs on a chart remember a looser shape than the one defined here, and the page shows the difference between a shape recalled and a rule applied.

Bars with no range

The source sometimes prints a bar whose high equals its low: an exchange holiday or a day with a single price. Such a bar carries no session. Left in, it would pass for an inside bar or sit as a flat neighbour beside real candles and would pull down the ATR that every threshold uses. Before any rule runs, these bars are removed from the stored history.

On the gold futures 77 bars were removed in this way, which is why the futures series has 2,437 bars against 2,514 stored. On the fund none were removed. The step is the same for every rule and happens before any count, so it cannot favour a result. A reader who lines up a futures chart here with a platform's chart may find a date missing; the stored history shows no trading range on it.

What a drawn example leaves out

A chart shows one occurrence, the cleanest by the rule's own measure when there are several. It leaves out what happened next in any way that could be counted, because counting outcomes needs a definition of success, and this site publishes none. It leaves out volume, because the stored history holds only open, high, low and close. It leaves out every time frame except the daily one, and every instrument except these two.

These omissions are stated where they apply. A reader who wants to know how often a flag was followed by a fall, or how often a gap closed, would need a counting definition and far more occurrences than the three bear flags and the two tweezer bottoms found here. This site teaches the shape, the rule and what the shape cannot say, and stops there.

Where the broker page fits

The site links to one broker, FxPro, from the button in the header and from the call to action at the foot of each page. The link is an advertising relationship: if a reader opens an account through it, the site is paid. The broker page explains in plain words what a contract on the price of gold is and is not, what is not delivered and what is not paid, and what to read in the terms before an account is opened from Pakistan.

None of the eight pattern pages needs the account: the charts, the rules and the counts for gold futures and the gold fund are on the pages. If a reader does open one, the order ticket asks for a size, a stop and a take profit, which no chart pattern can choose.