Fake-Breakout Filter
Every breakout of a box edge, swing, session level, pivot or trendline, tagged at the bar it happens with how often similar breakouts on earlier days failed, with n and a 95 percent range.
mdtFakeBreakoutEach breakout tagged with its own failure rate.
Five kinds of level
Consolidation box edges, swing highs and lows, prior-day and overnight highs and lows, the prior day's pivot, R1 and S1, and the chart's trendlines as mdtTrendlines finds them. Each is on by default and each switches off.
What counts as a breakout and a failure
A close beyond the level by more than 0.1 ATR while the bar before closed on the level's side; each level breaks once. Failed means a close back across the level within 5 bars; otherwise the breakout held.
Buckets built from the breakout bar
By default the bar's range (under 1, 1 to 2, over 2 ATR) and where it closed (the third toward the break, the middle, the far third). Part of the day, level type and volume against its norm can be added.
Rates from earlier days only
The tag on a breakout uses the breakouts that resolved before it, never the ones after, so a tag on a past bar is what it read at the time. A bucket with fewer than 30 earlier breakouts is marked "small n".
The fake-rate panel
The overall failure rate with n and its 95 percent range, then each bucket's rate, as of the bar in view. Held breakouts and the broken levels can be shown or hidden.
A module in two other tools
The same tag appears on each break in Break & Retest and on each breakout in the Breakout Box. Optional alert on a breakout.
What it looks like in a session.
Most breakouts fail; the question is which.
Breakout traders live with a known problem: a large share of breaks of an obvious level close back through it within a few bars. The usual answer is a filter read off the breakout bar itself. A wide bar closing near its extreme is a different thing from a narrow bar closing in the middle. The Fake-Breakout Filter turns that instinct into a count. It sorts every breakout on the chart into a bucket by those features and reports how often the earlier breakouts in that bucket failed.
Each breakout gets a tag at the bar it happens: the failure rate of its bucket, the number of earlier breakouts behind it and the 95 percent range of that rate. A tag marked "small n" rests on too few earlier cases to lean on. When the breakout resolves it is marked failed or held, and that result joins the statistics from then on. The panel lists the overall failure rate and the rate per bucket for the chart, as of the bar in view.
The tool does not say a breakout will fail; it says what share of earlier breakouts like it did. The buckets are settings. Adding part of the day, level type or volume splits the sample further and leaves fewer breakouts per bucket, which the n and the range make visible. The breakout buffer and the failure window are settings too, and a change rebuilds the statistics live.
Every setting, explained.
Read from the indicator's own settings, so this table always matches the current build.
| Group | Setting | What it does |
|---|---|---|
| Levels | Box edges | |
| Swing highs / lows | ||
| Prior day and overnight highs / lows | ||
| Pivots (prior day P, R1, S1) | ||
| Trendlines | Breaks of the chart's trendlines (as mdtTrendlines finds them). | |
| Swing strength (bars each side) | ||
| Swings kept (each side) | ||
| Bars to form a box | ||
| Box height at most (x ATR) | ||
| Breakout | ATR period (chart bars) | |
| Close beyond the level (x ATR) | ||
| Failed if back across within (bars) | ||
| Buckets | By breakout bar range (x ATR) | Under 1, 1 to 2, over 2 ATR. |
| By close location in the bar | The third toward the break, the middle, the far third. | |
| By part of the day | The cash session's first hour, midday, its last hour, outside it. | |
| By level type | Box edge, swing, session level, pivot, trendline. | |
| By volume | The breakout bar's volume against its norm: low (under 0.8 x), normal, high (1.5 x or more). Time of day on minute charts, the last 50 bars otherwise. | |
| Mark as a small sample under (breakouts) | A rate from fewer earlier breakouts than this is marked \"small n\": too few to lean on. | |
| History and alerts | 1-minute days (session levels) | The chart's loaded days with 1-minute data for the session levels and pivots (a change needs a reload). |
| Alert on a breakout |
One license. Every tool.
Fake-Breakout Filter is one of 76 tools in the Suite.