Flag pattern
Also called flag
A short continuation pattern: after a sharp move (the flagpole), price drifts in a small parallel-sided channel sloping against that move. A breakout in the direction of the pole is treated as confirmation.

- Level
- Intermediate
- Reading
- 2 min
- Sources
- 5BookOriginal authorReferenceResearch
- Updated
On this page· 5 sections
A flag needs a pole. First comes a sharp advance or decline on heavy volume; then price drifts against that move for a short while, inside two parallel lines[3].
Chartists read the drift as a pause in the middle of the move, not the end of it[3][4].
How it works
The flagpole is the sharp move. StockCharts measures it from the first break of resistance or support to the high or low of the flag[3]. Bulkowski adds that the pole should be unusually steep and last several days[2].
The flag itself is a small rectangle that slopes against the pole: down after a rally, up after a decline[3]. Flags are usually too short to have clear reaction highs and lows, so the price action is contained by two parallel trendlines instead[3].
Duration sets the limits. StockCharts’ glossary puts a flag at generally less than three weeks[4]. Its pattern page allows one to 12 weeks, ideally one to four, and calls a flag older than 12 weeks a rectangle[3].
Volume should be heavy on the pole, lighter during the flag and expanding again on the break[3].
A break in the direction of the pole completes the pattern. The length of the pole, added to the breakout level, gives the conventional target[3]. Edwards and Magee give such measuring rules a chapter of their own[1].
Example
Say a share breaks resistance at $50 and runs to $60 in six sessions on heavy volume.
- Flagpole: $60 − $50 = $10.
- For the next two weeks it drifts lower between two parallel lines. The upper line falls from $60.00 to $58.50, and the flag’s low is $56.00.
- A close at $59.00 breaks the upper line at $58.50.
- Pole convention: $58.50 + $10.00 = $68.50.
Bulkowski scales his version down. He multiplies the height of the price swing by the share of his flags that met their target, 46%, and adds the result to the flag’s low[2]. Here that is $56.00 + $4.60 = $60.60. The two conventions sit $7.90 apart, and neither is a forecast.
Watch out
A slow drift is not a flag. Without a sharp move before it, StockCharts calls the formation’s reliability questionable[3].
Long flags lose their character. StockCharts calls patterns of eight to 12 weeks debatable, and some analysts treat eight weeks as the limit[3].
The research is limited. Park and Irwin’s review cites studies in which bull flags on the NYSE Composite Index beat buy-and-hold before transaction costs. One of them, by Leigh, Paz and Purvis (2002), was the only chart pattern study in the review to run out-of-sample tests[5]. Bulkowski’s bull market statistics, last updated 27 August 2020, show a break-even failure rate of 44% for upward flag breakouts and 45% for downward ones[2].
Sources
- BookOpen Library (Internet Archive)Technical Analysis of Stock Trends, 9th ed. (Edwards & Magee, 2007), record with table of contentsChecked November 9, 2026
- Original authorThePatternSite.com (Thomas N. Bulkowski, author of Encyclopedia of Chart Patterns)Bulkowski on FlagsChecked November 9, 2026
- ReferenceStockCharts ChartSchoolFlag, PennantChecked November 9, 2026
- ReferenceStockCharts ChartSchoolGlossary - FChecked November 9, 2026
- ResearchUniversity of Illinois at Urbana-Champaign (farmdoc)The Profitability of Technical Analysis: A Review (Park & Irwin, AgMAS Project Research Report 2004-04, October 2004)Checked November 9, 2026
