Level one: exploratory simulation
Fast simulations are used to eliminate weak hypotheses and compare broad rule sets. They are labelled exploratory and do not support a customer release or performance claim.
Evidence before release
JPTC research methodology
This policy defines which test levels JPTC uses and what each level can support. A result is reported with its assumptions and cannot be promoted merely because its equity curve looks attractive.
Published by JPTC Research Team under the JPTC editorial policy.
View current researchLevel two: broker-terminal testing
Candidates are rerun in the intended platform using broker data and recorded spread, commission, swap and order assumptions. Terminal results take precedence over a research-engine result.
Level three: unseen-period checks
Settings are checked on periods that did not choose them, including separate years and recent data. A candidate that depends on one favourable period fails the release gate.
Execution realism
The report identifies timeframe, symbol specification, session rules, spread, commission, swap, slippage assumptions and fill model where available. Missing execution information is a limitation, not a detail to hide.
Frequency and concentration
Trade count, stagnation, worst day, losing streak and exposure concentration are reviewed alongside return statistics. A result with too few trades cannot be rescued by a high headline metric.
Release and rejection
A customer file receives a version, intended use, risk notes and evidence record. Failed, unstable or over-fitted candidates remain in research and are not shipped.
Common questions
What is the difference between exploratory and validated research?
Exploratory research screens ideas quickly, while validation uses the intended terminal, execution assumptions and unseen periods. Only the second can support a customer release decision.
Does JPTC optimise only for the best result?
No. Selection considers stability, drawdown, trade count, stagnation and behaviour across windows. A single best result is normally a reason for further checking, not automatic selection.
Will JPTC publish proprietary strategy rules?
No. JPTC can publish methodology, aggregate evidence and failure reasons without exposing entry code, private setfiles or customer-only implementation details.