Inside bar breakout up event

Marks an upward break: a close above the high of the inside bar just before it.

An inside bar is a candle whose high is below the prior candle's high and whose low is above the prior candle's low, so its whole range sits inside the bar before it. This indicator marks the candle right after an inside bar when its close is above that inside bar's high, an upward break of a tight range. confirm_days is meant to require the break to hold for several bars in a row, but a break candle can never itself be an inside bar, so that requirement cannot be met above 1.

How to read it

Each marker sits on the candle whose close is above the high of the candle just before it, which was an inside bar.
On one-minute bars markers are frequent, about 32 a day on the SPY sample, so read them against the trend and nearby levels.
Only upward breaks are drawn.

What it outputs

Breakout up: True on the candle whose close is above the high of the inside bar just before it. Drawn on that closing candle.

When it fires

Fires on the candle whose close is above the high of the inside bar just before it, one marker per break. On the default SPY sample it fired 674 times in a month, about 3% of bars. With confirm_days above 1 it never fires.

Settings

confirm_days: Bars the break must hold for in a row. At the default 1, one close above the inside bar's high marks the bar. At 2 or more nothing fires, since a break candle is never an inside bar. Minimum is 1, so no looser setting.

Pitfalls

confirm_days above 1 looks broken: a break candle can never be an inside bar, so the streak never builds. Zero markers at 2 and 3 on the SPY sample and on valid bars. Treat it as unusable above 1 until fixed.
Markers are frequent on one-minute bars, about 3% of candles and 32 a day on the sample, so most are small noise breaks. Pair them with a trend or level filter.
The test is a close above the inside bar's high only. A close a fraction of a cent above counts the same as a wide move, and the size of the inside bar is never checked.