Skip to swing research
Swing Trading CoursesThesis / Sessions / Cash

Build a Swing Trading Strategy Specification

Write a causal swing specification with session-aware features, corporate-action handling, native exits and a complete exposure ledger.

Original source review: 8 October 2026. Examples are fictional; no empirical swing backtest or paid trading account was tested. Expanded edition: 13 October 2026.

Guide revision and linked-source check: 10 October 2026

Specify a multi-session question that can fail

Swing describes a holding pattern, not a verified strategy. Choose the instrument universe, decision clock and observable objective before assembling indicators. Define whether the question concerns contraction toward an earlier reference, continuation or another declared relationship. Each needs a population and future horizon. An attractive chart selected after a move is not the specification of a reproducible entry.

Write a simple compatible baseline and a rejection condition before testing. The baseline needs the same eligible observations and access assumptions. Save missing labels, abstentions and universe changes instead of choosing only securities that still trade today. A causal boundary-recovery entry objective must remain distinct from later exit-adjusted economics. The first deliverable is a written contract another researcher could reject, not a claim that several days of holding automatically improve the strategy.

Declare the observation window and session mapping

Name the data source, event and receipt clocks, adjustment convention and warmup. A completed daily bar is available only under the declared source timing; a corrected historical close may be a different input from the originally received value. Identify which venue session a date belongs to and how holidays or shortened sessions alter eligibility. NYSE's reference applies to its named markets, not every instrument a swing service might mention.

Do not include the decision observation in an earlier reference accidentally. In a fictional prior-close window of 100, 102 and 98, the reference mean is 100. A decision close of 94 is six away from that earlier reference. Keep that calculation separate from an actual entry price and economic exit. Irregular or missing observations need an explicit policy rather than silently shortening the window to keep a trade in the sample.

Handle corporate actions without manufacturing a price shock

Separate source adjustment, share quantity and actual broker action. An invented four-for-one split turns eight shares at 100 into thirty-two at 25, preserving 800 of value at those theoretical before/after prices. Treating 100 to 25 as an unadjusted seventy-five-percent market loss ignores the quantity change. The actual market price after an event can differ; the example isolates the unit transformation only.

Record announcement, effective date, source availability and the rule for existing orders. A broker may adjust or cancel orders under its own terms, so historical price adjustment cannot establish the account's order state. Dividend, merger and delisting cases need their own contracts. Save both the raw source and justified transformed research series. A current adjusted file should not be presented as the exact information a reader saw before an earlier decision.

Freeze the action and holding contract

Specify activation, entry reference or order mechanism, supported quantity increment and intended whole-position exit. Preserve the native stop, target where applicable and maximum-hold policy. Observation-count, elapsed-time and completed-session clocks have different meanings across weekends and missing bars. A completed-session update must not silently gain authority to introduce daily profit floors or daily stop exits into a minute-managed system.

Keep entry identity frozen on the declared causal objective before applying one predeclared exit for economic validation. An exit can reject or constrain implementation without selecting the original entry retrospectively. Save amendments, discretionary intervention and previously inspected regions as separate versions. Do not choose a more favourable holding horizon after seeing the reserved path. A chart's eventual recovery does not authorise extending a losing position beyond its original contract.

Join sizing, funding and overnight stress to the same position

Use a named account lane rather than a generic ratio. The existing worksheet supports cash-funded whole long shares, not shorts, options or margined contracts. With fictional entry 100, stop 97, additional per-share cost 0.10 and chosen budget 150, downward rounding gives forty-eight shares and planned loss 148.80. An assumed adverse exit at 94 gives 292.80 on the same quantity.

These figures do not identify a suitable personal budget or a gap probability. Check funding and outstanding exposure independently. Prices, costs and quantities must share one currency convention, and costs already inside executed prices must not be deducted again. A pending exit or cancelled remainder can leave exposure open. Keep the actual account lane and order reports beside a hypothetical stress instead of calling the planned stop a maximum possible loss.

Reserve evaluation and release the complete replay

A useful package contains source identity, session and corporate-action maps, candidate history, frozen entry, native exit and all eligible opportunities. Check label spans at evaluation boundaries, not just chronological starting dates. Warmup context can be allowed under a declared rule without becoming extra evaluated outcomes. Software fixtures for splits and missing sessions establish implementation behaviour, not independent market validation.

The blank strategy record captures the decisions needed before a real backtest. Reproduce the reference, split and stress examples separately from empirical results. Preserve rejected or no-model conclusions rather than promoting a candidate because code ran successfully. A deployment claim additionally needs current cloud revision, active route and durable execution evidence. This guide reports a construction method, not a completed strategy study, paid-account test or promise of profitable swing trading.

Review checkpoints

Evidence requests, not observed strategy results
ObjectPreserveDo not infer
InputsEarlier observations and action-aware unitsAn adjusted chart is original subscriber information
PolicyFrozen entry and native holding/exit ownershipA later recovery permits an unlimited hold
ImplementationFunding, quantity and labelled gap stressA planned stop proves maximum loss

Download blank research record (CSV)

Headers only. No trades, account outcomes or operating receipts are supplied.

Primary Sources

Swing Gap Stress Worksheet

Continue the Research

One accounting example does not replace the factory

The filled cash cases make original quantity, action factor, gross distribution and one declared cost explicit. They do not define a causal entry objective, real eligibility or independent economic sample. Preserve the original long strategy-building protocol and its blank download. A real research task uses the authorised route and complete independent master rather than renaming a share fixture as a new asset class. The canonical best hypothetical observation-window fill and fixed-entry/fixed-exit boundaries remain unchanged. No source code entitlement, broker permission or production promotion follows from the tutorial.