Skip to short-horizon research
Short Term Trading StrategiesCondition / Request / Position

Common short-term research mistakes

Check cost units, within-bar ambiguity, late instructions and unfinished order states before trusting short-term results.

Original editorial revision: 10 October 2026. Expanded edition: 14 October 2026.

Educational execution planning. Invented costs and clocks are assumptions, not forecasts, broker instructions or a measured strategy record. Check the exact venue, instrument and account rules.

Mistaking a small gross movement for a net edge

Short horizons often make the price difference being measured comparable with the assumed trading friction. Name whether the prices are midpoints, reference closes or executable quotes before subtracting costs. A bid-to-ask convention already contains a spread component; subtracting that spread again can overstate friction. Conversely, midpoint returns without an execution adjustment can overstate a hypothetical gain. Both errors concern the accounting contract, not an opinion about trading frequency.

Use invented notional 10,000 on twenty round trips, gross movement six basis points per trip, variable cost eight basis points per completed round trip and two additional currency units per trip. Gross cash is 120, variable cost 160 and extra fixed cost 40, giving net -80. The sign changes even though every entered gross movement is positive. These figures are fictional arithmetic, not a slippage sample or a profitable strategy test.

Mixing per-side and completed-round-trip charges

A cost labelled four basis points may mean one side or the full entry-and-exit journey. Save the definition before multiplying by trade count. In the fictional twenty-trip case, four basis points per side becomes eight per completed round trip. Entering four as the round-trip figure halves that assumed variable charge. A fee per order also differs from a fee per filled unit, particularly when an instruction is partially executed.

Record minimum charges, financing and currency conversion only where the exact contract supports them. Do not copy a current fee page into a historical study without a version and effective date. If costs are already inside a net cash ledger, applying another generic deduction changes the meaning of the record. Keep the raw figure, cost convention and adjusted result together rather than overwriting the original column with whichever convention produces the desired headline.

Choosing a favourable sequence inside a price bar

An invented bar opens at 100, reaches a high of 103 and a low of 97, then closes at 101. A proposed long has target 102 and stop 98. The bar range includes both levels but does not establish which was reached first after the entry. Target-first and stop-first paths are both compatible with the same four bar values. No screenshot of that bar supplies the missing sequence.

Use more detailed verified observations when the claim requires them. Otherwise retain ambiguity and a predeclared hypothetical convention rather than granting the winning exit to every ambiguous bar. A limit touch also does not establish account queue priority or executed quantity. Keep correctness fixtures for both possible paths. Their purpose is to expose the model's assumption, not to manufacture tick history or prove that an exit would have executed on a particular venue.

Treating a session label as an actual deadline

A rule saying close before the session ends needs the named venue, trading phase, timezone and calendar version. Different products can have different auctions and permitted order windows. Use the official NYSE schedule only for the market it describes; do not extend it automatically to futures, FX or crypto. A shortened session requires its own dated boundary rather than a remembered regular closing time.

Separate the deadline for a decision, submission, acknowledgement and completed fill. A cancellation submitted before a boundary may leave an execution race or a still-open position. Record the final state instead of labelling every timely request flat. Preserve daylight-saving mapping and source receipt times in historical tests. No universal number of seconds before close guarantees an exit; an operational buffer is an assumption requiring its own policy and evidence.

Deleting missed, partial or late actions from the sample

A short-term population includes opportunities the system could not act on as well as convenient completed trades. Save eligibility, rejection reason, late arrival, accepted order quantity and remaining exposure. A chart condition that occurred before receipt cannot become an earlier subscriber entry merely because the historical file contains it. Keep research reference fills under their declared convention and actual executions under broker reports.

An instruction cancelled after thirty of one hundred requested units fill leaves thirty units of exposure unless a separate closing execution is documented. Cancellation of the remainder does not erase those fills. Do not turn a repeated message into another independent winning trade. Preserve order identifiers and amendment lineage across the entire period, including outages. A closed-only summary needs its scope visible beside the metric, not a claim that it measures every original opportunity.

Repairing the specification after inspecting the result

Changing a threshold, market filter, holding clock or cost convention after seeing an adverse interval creates a changed experiment. Save the revision and the already-inspected sample. A reserved region is not untouched after it has helped choose the new rule. Keep entry selection on its declared causal objective and apply the predeclared exit separately; a later exit improvement must not redefine the original entry ranking.

The blank mistake-audit record captures the specific error, affected population and evidence needed to resolve it. Reproduce the friction and ambiguous-bar fixtures, then inspect real data only under a frozen specification. No empirical latency distribution, account return or optimal trading frequency is supplied here. A useful audit can conclude that bar-only data cannot support a claimed execution sequence, while still preserving the valid reference calculations that the same dataset can support.

Review checkpoints

Evidence requests, not completed study results
ObjectPreserveDo not infer
CostsSide, round-trip and price conventionsA gross positive bar is a net gain
SequenceEvent order and ambiguity policyOHLC proves target was reached first
PopulationLate, partial and unresolved actionsOnly completed winners define the strategy

Download blank review record (CSV)

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

Source documents

Documentation establishes the stated mechanism, not tested trading performance. Checked 10 October 2026; a later version can differ.

Detailed research

A closed-leg winner can hide unfinished exposure

The new default has five invented closed price-change units and negative six on the unclosed remainder, before 1.7 incurred fees. Reporting only the positive fragment would omit the snapshot's adverse total. Its R denominator remains the entered original twenty rather than a later smaller risk. No actual position or broker statement is implied. Keep the original mistake catalogue, complete quantity record and source clocks when using the packet; it does not provide a market path, probability or independently validated strategy result.