Which inputs does False Breakout Filtering use?

Direct answer

False breakout filtering uses a set of rule inputs that convert raw market information into a yes/no label for whether a breakout is likely to be false. The key inputs are the price levels that define the prior range, the rule that defines “breakout,” and the follow-up observation window that decides whether the price quickly returns back into that range.

Because “False Breakout Filtering” can be implemented in multiple ways, you should treat its inputs as the parameters of the specific rule set you are using. The stable concept is: compare a breakout attempt against a prior reference range, then require a later confirmation that the price does not sustain outside the range.

Mechanism and definition (what inputs are used)

At a high level, a false breakout filter needs three groups of inputs.

Reference range inputs (what you filter against)

Lookback window: the number of bars/time units used to define the “before” region.
Range definition: what counts as the upper and lower bounds (for example, recent highest high and lowest low, or a tighter band using swing points).
Reference price type: whether the range is defined using highs/lows, closes, or another consistent method.

Breakout detection inputs (when you label an attempt)

Breakout threshold: the rule for “outside the range” (for example, crossing above the upper bound, or closing beyond it).
Price sampling rule: whether you require the condition intrabar (touch/cross) or based on bar close.
Optional buffer: sometimes a small offset is used to reduce sensitivity to minor boundary breaches.

**False confirmation inputs (how you decide it was “false”)

Confirmation window: how many subsequent bars you watch after the breakout attempt.
Re-entry condition: the rule that defines “failure,” such as returning back inside the original range by close.
Confirmation price type: again, whether the re-entry must be measured by close, low/high, or another consistent choice.

Put simply: the filter consumes (a) historical price to set bounds, (b) current price to detect a breakout attempt, and (c) future price behavior to judge whether the attempt fails quickly.

Evidence or example (a concrete, checkable rule set)

Example of a fully specified input set (no real-time data assumed):

Data: a sequence of bars with high, low, and close.
Lookback window: 20 bars used to compute an upper bound and lower bound from the range.
Breakout definition: a breakout attempt occurs when the close is above the upper bound.
Confirmation window: the next 5 bars.
False condition: the breakout attempt is labeled “false” if any of those 5 bars closes back inside the original range (for example, close falls below or equal to the upper bound and above or equal to the lower bound, depending on your exact rule).

In this example, the inputs and parameters are explicit: 20-bar lookback, close-based breakout, 5-bar confirmation window, and a close-based re-entry rule.

A material detail is that the inputs depend on consistent data handling. If one implementation uses highs/lows for the range and another uses closes, or if one checks breakout on intrabar touch while another checks only closes, then they are different filters even if both are described as “false breakout filtering.”

Limitations and risks (what can go wrong)

At least four limitations commonly affect false breakout filtering:

Parameter sensitivity Small changes to the lookback window, threshold, or confirmation timing can change labels dramatically. That means “the filter” is not a single universal input set; it is a family of rules.
Market regime changes A method that assumes range-like behavior may perform poorly when volatility structure changes (for example, when the market trends strongly rather than oscillates within bounds).
Execution and cost assumptions Historical evaluation can be misleading if costs are ignored. Even if your logic only uses prices, real outcomes depend on spreads, slippage, and how orders would be filled.
Ambiguous boundary behavior If the market repeatedly hovers near the range boundary, the filter may oscillate between “breakout” and “false” labels based on minor differences in bar-close timing.

Also note: historical relationships do not establish future results. Treat any observed edge as conditional on the exact inputs and conditions used in testing.