Swing Trade Methodology
A selective multi-day signal process built to identify unusual price dislocations, qualify the ones with durable recovery potential and manage every published card through a defined lifecycle.
Price moves first. Context decides whether it matters.
Swing Trade begins with statistically unusual price behaviour. It then asks whether the move has the quality, context and historical character required for a multi-day position. The model is not a generic trend screen and it does not publish every breach.
Observe
Causal market data is evaluated as it becomes available. No future bars or later revisions are used to create a live signal.
Qualify
Proprietary quality and context checks separate ordinary noise from dislocations with a historically useful multi-day profile.
Grade
Validated model lanes contribute to an A-D conviction grade. The grade is evidence metadata, not a guarantee of outcome.
Publish
The subscriber card records instrument, direction, entry basis, grade, manager state and publication time.
The entry model and trade manager have separate jobs.
Entry research decides whether a card should exist. The live manager then watches how price behaves after publication. This separation keeps a promising setup from being confused with a successful trade.
At risk
The card is active but has not yet demonstrated enough favourable movement to earn protection.
Confirmed
Price has met the model's confirmation condition. The card transitions to a protected management state.
Managed
Favourable movement can raise the protected level. That level never moves backwards.
Closed
A risk event, protected-level event or the model clock ends the card and records the published result.
Observable outputs are public. Implementation remains proprietary.
Transparency means letting users verify what was published and what happened next. It does not require publishing enough source detail to reconstruct the model.
Published and verifiable
Card timestamps, direction, entry basis, grade, active state, displayed risk level, close state and realized result. Model-level performance is reported from the live ledger with its scope stated.
Retained as research IP
Feature weights, numerical gates, calibration constants, routing logic, internal model identifiers, training artifacts and the complete implementation code.
The model can be wrong, late or early.
A defined process controls how a signal is managed. It does not remove market risk. These are the conditions in which a multi-day recovery model is most vulnerable.
Event gaps
Earnings, policy decisions and unscheduled news can move price through a displayed level before a follower can act.
Regime change
A dislocation can become the start of a new trend rather than a temporary departure from normal behaviour.
Execution difference
The published research record and an individual account can differ because of timing, liquidity, spread, slippage and position sizing.
Judge the record, not the story.
Historical research, counterfactual replays and internal shadow models are not presented as post-publication live results. The current performance page states the evidence window and lists realized subscriber cards.
Before the outcome
A card is recorded when published, with its identifying fields and timestamp.
Through the lifecycle
State changes are tied to the card manager rather than rewritten after the result is known.
After closure
Closed cards remain in the ledger and flow into the scoped live-performance record.
Read the methodology, then inspect the tape.
The strongest evidence is not a formula on a marketing page. It is a consistent pre-outcome publication record that remains visible after good and bad trades.