Education
Contents
By Arizmic LLC10 min readPublished Last reviewed

QFL: base breaks, bounces, and risk limits

Turn a broken support base into a causal rebound experiment before adding layers, leverage, or hindsight.

QFL turns a broken base into a rebound hypothesis

Quickfingers Luc (QFL) is a community-developed, usually long-only approach to trading a sharp move below a prior support reference called a base. The thesis is not that every support break must recover. It is that a sufficiently clear earlier bounce can provide a reference, and a later break below that reference may create a bounded opportunity to test for a rebound.

That makes QFL a particular kind of mean-reversion study. Its anchor is not a rolling average or a fair-value estimate; it is a previously defined price level. Its hard problem is not merely choosing a percentage drop. It is defining a base without using information that was unavailable when the trade would have been considered.

Why a broken base might rebound

The rebound hypothesis is a liquidity hypothesis, not an assertion that a chart level is inherently valuable. A rapid decline can contain temporary selling pressure that exhausts near a previously observed support reference. The competing explanation is repricing: the support may have become irrelevant because the information, liquidity, or broader trend changed. QFL therefore needs a failure path as carefully specified as its entry path.

The popular term covers many implementations. Some use pivot patterns, bounce size, and volume filters to qualify a base. Some place several resting orders below it. Some use no hard stop. Those are materially different risk profiles, not interchangeable settings on one strategy.

A base must be a rule, not a line drawn after the bounce

A researcher needs to decide what makes a base eligible before testing the subsequent break. A simple, reproducible formulation can use a confirmed swing low as the candidate reference, then require a predeclared bounce magnitude and minimum time before the level can become eligible. The confirmation delay matters: a pivot recognized only after later bars have formed cannot be traded as though it were known at the original low.

DecisionA declared baselineWhy it changes the result
BasePrior confirmed swing low with a minimum bouncePrevents support from being selected after the outcome
CrackCompleted price observation below a fixed percentage from the baseSeparates context from an executable decision
EntryNext eligible event, or a later reclaim/retestTrades earlier price against more information
ExitPartial return toward the base, timeout, or invalidationDefines what counts as a rebound worth holding
RiskOne entry with a fixed loss budgetMakes the initial test interpretable before adding concentration
ReuseLimit the number of attempts per baseAvoids treating one prolonged decline as independent trades

The base itself is not proof of demand. It is a reference for an experiment. When price stays below it, the result may be a failed reversion, not a reason to redraw the level until the trade looks better.

A worked base-break sequence

Illustrative example. Suppose a previously confirmed base is 100. A later completed observation closes at 94, a 6% move below the base. A subsequent completed observation is 97: 3.19% above the crack, but still 3% below the original base.

This path supports a causal sequence:

  1. The base was confirmed before this decline and was already eligible under the study rule.
  2. The close at 94 meets a predeclared 5% crack condition. It creates a candidate; it is not backfilled as an order at the intrabar low.
  3. A one-entry baseline decides at the next eligible event, or waits for a separately declared reclaim/retest condition. Record the decision price and the actual fill rather than assigning the trade the 94 close automatically.
  4. A rebound target might be a fixed fraction of the distance back to 100. A timeout or a further, predeclared invalidation closes the position if the original support does not regain relevance.

The 97 observation is useful precisely because it is incomplete. It is neither proof that the base has been respected nor proof that the selloff is over. The study needs the whole future path to measure what happened, while the strategy may use only the information available at each decision.

One entry and layered entries answer different questions

Layering or dollar-cost averaging is common in QFL descriptions. It can lower the average entry price when price recovers, but it also increases exposure as the thesis is under the most pressure. A backtest that reports only the average fill and final profit can hide the capital, liquidation, and tail path required to reach that outcome.

Start with a one-entry version. It exposes the base-selection and rebound hypothesis directly. Then compare a layered version only with all of the following fixed in advance:

  • maximum number of layers and total notional;
  • layer distances and order type;
  • capital reserve and the outcome when an order cannot be placed or filled;
  • hard invalidation or a documented no-stop policy with a portfolio-level loss limit; and
  • the same fees, spread, and fill assumptions used for the one-entry version.

Report the maximum exposure, worst excursion, time in trade, and fraction of capital tied up as prominently as the final return. A smoother sequence of small recoveries is not evidence that the concentrated failure path is acceptable.

Design an experiment that can reject the base

Keep every eligible base, including the ones that break and never recover. Partition candidates by the earlier bounce size, crack depth, volume state, market trend, and time since the base formed. Then compare entry policies that use the same set of candidates:

  • immediate next-event entry versus a completed-bar reclaim;
  • fixed percentage crack versus a volatility-scaled crack;
  • one entry versus a capped layered schedule; and
  • a base target versus a nearer partial-reversion target.

For each variant, preserve the point-in-time base and the decision time. The research output should show candidate count, fill rate, realized costs, maximum adverse excursion, capital usage, and the distribution of unresolved or timed-out positions. A result that depends on one exchange, one quiet period, or one hand-selected definition of a "strong" bounce has not yet identified a robust QFL rule.

Where QFL breaks

  • Lookahead base selection: a visually obvious low may not have been a confirmed base until several bars later.
  • Trend and repricing risk: a support break can be new information, not temporary liquidation.
  • Concentration risk: layers accumulate the most size in the deepest part of a decline.
  • Venue and instrument variation: fees, liquidity, funding, minimum order size, gaps, and delistings can dominate a small rebound target.
  • Reuse bias: multiple entries around one failed base can make one event look like many independent opportunities.

QFL is therefore best treated as a family of base-defined rebound experiments, not as a claim that support will always be recovered.

Try it in Arizmic

Strategy composition

Build the premise with shipped signals

Treat a QFL base as a causally confirmed support reference. A break below it creates a rebound candidate, not a promise of recovery, so the strategy must state the point-in-time base, decision delay, exposure cap, and invalidation before it can be compared with a layered alternative.

These are starting structures, not presets or evidence of an edge. Choose one, replace the bracketed decisions, and keep the signal roles separate as you test it.

Composition 01

Confirmed pivot base with retest entry

Structure-and-retest recipe
  • Swing High/Low And Pivot StructurePoint-in-time base reference
  • Break Of Structure / RetestCrack and reclaim state
  • Relative VolumeParticipation context
  • Wilder ATRVolatility-scaled risk

Data: OHLCV Bars

View configuration and complete rule

Swing High/Low And Pivot Structure

Point-in-time base reference

Pivot Structure makes the support reference and the lookahead boundary inspectable.

Configure

left_lookback · [bars before pivot]
State how much prior structure must surround a possible base.
right_lookback · [bars to confirmation]
Make the delay before a base becomes knowable explicit.

Use the output

last_pivot_low
Use the most recently confirmed pivot low as the candidate base; do not backdate it to the original pivot bar.
pivot_low
Audit when the pivot flag is emitted so the study preserves its confirmation delay.

Break Of Structure / Retest

Crack and reclaim state

Break Of Structure / Retest separates the support crack from the later recovery evidence.

Configure

retest_window · [eligible retest bars]
Limit how long a base break can remain an active candidate.
tolerance · [base-distance tolerance]
Define what counts as a meaningful return to the broken level.

Use the output

bos_down
Identify a completed downward break using the explicit point-in-time pivot level.
retest_down
Require the separately declared retest or reclaim state before a confirmation-delayed entry.

Relative Volume

Participation context

Relative Volume keeps a low-participation break distinct from a more active liquidation or reclaim event.

Configure

window · [same-time volume window]
Compare the crack or reclaim with a declared participation baseline.

Use the output

relative_volume
Filter or partition candidates by [minimum relative volume] without treating volume as direction by itself.

Wilder ATR

Volatility-scaled risk

Wilder ATR makes a base break comparable across changing price levels and volatility states.

Configure

window · [ATR window]
Scale crack, invalidation, and size to recent range rather than raw price points.

Use the output

atr
Express the maximum adverse move and size budget in ATR units.

Assemble the rule

Entry
After a confirmed downward structure break below the last known pivot low, enter long only at the next eligible event after the declared retest or reclaim condition is complete.
Exit
Exit at [partial return toward the base], [ATR-based invalidation], or [candidate expiry], whichever occurs first.
Decision time
A pivot low is usable only on its confirmation row; the crack and any retest are separate completed-event decisions.
Sizing
Use one initial entry sized to a fixed risk budget before comparing it with any capped layered schedule.

Useful variations

  • Compare next-event crack entry with retest-confirmed entry while preserving the same causally confirmed bases.
  • Compare fixed percentage crack distance with ATR-normalized crack distance.
  • Partition candidates by relative volume and time since the base was confirmed.

Keep in view

Break Of Structure / Retest consumes explicit point-in-time pivot levels; it does not calculate the base internally. Pivot flags are emitted on the confirmation row, and Relative Volume needs a declared same-time-of-session baseline policy.

Composition 02

Base crack with stabilization gate

Recovery-confirmed recipe
  • Swing High/Low And Pivot StructureBase reference
  • DrawdownSelloff severity
  • SMA RSIStabilization confirmation
  • Wilder ATRRisk scale

Data: OHLCV Bars

View configuration and complete rule

Swing High/Low And Pivot Structure

Base reference

Pivot Structure supplies a stable, inspectable reference for the base-defined study.

Configure

left_lookback · [bars before pivot]
Define the structural context of an eligible base.
right_lookback · [bars to confirmation]
Preserve the point at which the base becomes observable.

Use the output

last_pivot_low
Measure the base crack against the last confirmed low rather than a low identified after the rebound.

Drawdown

Selloff severity

Drawdown distinguishes a marginal break from a larger displacement that may have a different rebound path.

Configure

No configurable parameter is required for this role.

Use the output

drawdown
Require a declared minimum drawdown state before treating a base crack as a rebound candidate.
running_high
Review the reference peak used by the cumulative drawdown calculation.

SMA RSI

Stabilization confirmation

SMA RSI delays entry until the chosen short-horizon momentum measure has stabilized.

Configure

window · [recovery window]
Set the short horizon used to require that downside momentum has stopped extending.

Use the output

rsi
Require SMA RSI to recover through [threshold] after the crack before entering at the next eligible event.

Wilder ATR

Risk scale

Wilder ATR turns the risk boundary into a comparable distance across instruments and regimes.

Configure

window · [ATR window]
Make the stop and capital budget responsive to the current range.

Use the output

atr
Set [invalidation distance] and report adverse excursion in ATR units.

Assemble the rule

Entry
After price has cracked the last known pivot-low base and the drawdown condition is active, enter long only after the completed SMA RSI recovery condition is met.
Exit
Exit at [partial reversion target], [ATR-based hard invalidation], or [timeout] rather than adding exposure automatically as price falls.
Decision time
The crack, drawdown state, and RSI recovery are evaluated on completed observations; the fill is modeled only after the final condition is known.
Sizing
Cap the initial position by the ATR-based invalidation distance and reserve any layered comparison for a separate experiment.

Useful variations

  • Compare a no-confirmation entry with the completed-bar RSI recovery gate.
  • Test drawdown severity buckets while holding the base rule and risk budget fixed.
  • Compare a single entry with a fully capped two-layer schedule and report total exposure at every layer.

Keep in view

Drawdown is cumulative from the first bar rather than session-anchored, and the shipped RSI is SMA RSI rather than Wilder RMA RSI. Neither output establishes that the prior support base will recover.

Ask the AI Companion

Draft this strategy

Strategy draft

Turn the guide's base-break idea into a risk-capped draft that keeps the support reference, confirmation delay, and failure path explicit.

I want to research a base-break rebound strategy for [instrument] on [timeframe]. Help me define a support base that is knowable before the later decline, a crack condition, and a cautious confirmation policy for a long entry. Recommend a starting risk cap, exit, and timeout, and make clear how I should compare a single entry with a separately capped layered version. Then build the strategy and explain where hindsight could enter the base definition.

Extend it in Marimo

Study review

Begin from a retained Study so candidate bases, decision times, fills, and unresolved positions remain tied to their evidence.

Trace each confirmed base from qualification through crack, entry decision, fill, recovery, invalidation, timeout, or any capped added exposure.

Bring in
engine-reported Pivot Structure last_pivot_low and pivot_low, Break Of Structure / Retest bos_down and retest_down, Relative Volume relative_volume, Wilder ATR atr, decisions, and fills, declared base-qualification, crack, retest, target, expiry, and capital-cap policies, retained candidate identifiers, costs, order attempts, fills, and unresolved-position records
Build
base-to-crack and crack-to-decision timeline with confirmation delay, candidate, fill, recovery, invalidation, and timeout table by crack depth and participation state, single-entry versus capped-layer exposure, adverse-excursion, and capital-usage comparison

Interpretation: Check whether apparent improvement comes from the base rule, delayed recovery information, or increased capital committed while the thesis is failing.

Value origin: Signal values, decisions, order attempts, fills, costs, and retained outcomes are engine-reported. Base-age labels, event-aligned paths, and exposure decompositions are notebook-derived.

With Companion: Ask Companion to draft a base-timeline and exposure-reconciliation cell set, inspect confirmation delay and capital accounting, then apply the diff explicitly.

Further reading