A meaningful date range
A range that covers several market regimes instead of a single trend.
Backtest / Optimization
Backtest and optimization
Historical backtest, optimization, forward test and statistical analysis.
- Test conditions statedDate range, spread, commission, tick model
- Overfitting controlForward test after optimisation
- Suggested parametersWith a sensible range
Scope
What this service actually covers
The scope is settled before work starts; the items below are part of this service.
Controlled optimisation
A search through parameter space with recorded results, not a pile of output rows.
Overfitting review
In-sample and out-of-sample results are compared, then forward tested.
A transparent report
Assumptions, limitations and data quality are all stated in the report.
Every report states its measurement conditions: data window, commission, spread, tick model and limitations.
If tick data of the required quality is not available, that limitation is stated explicitly — better than publishing a number without its conditions.
The final handover includes recommended parameters with their stability band, showing which settings are sensitive to small changes.
Delivery
How this service is delivered
Four stages with defined output; nothing moves on without your approval.
-
Scenario definition
Date range, tick model, spread and commission are fixed.
Stage output Test scenario document -
Baseline backtest
The current build is run and its trade paths and errors reviewed.
Stage output Backtest report -
Optimisation
Parameters are searched with overfitting control and recorded results.
Stage output Suggested parameters -
Forward test
The final build runs on demo and the differences are recorded.
Stage output Demo test report
Handover
What you receive
What you receive at the end of the project.
3 items
- Backtest report with measurement conditions
- Overfitting and walk-forward analysis
- Recommended parameter set
This list is part of the project scope; each item is handed over at the end of its stage, before final delivery.
Who it is for
Who this service is for
If one of these describes your situation, this is the right service.
A finished EA that was never tested
The system runs, but you have no defensible numbers for its behaviour.
A trader who wants decisions from data
You want to know which parameters actually affect the outcome.
Questions
Common questions about this service
Short answers to what clients ask about this specific service.
How is optimisation different from a backtest?
Answer
A backtest measures the current build; optimisation searches parameter space and, left uncontrolled, ends in overfitting.
How is data quality assured?
Answer
The data source is named in the report along with its limitations; no number is published without its measurement conditions.
What is the final output of this service?
Answer
A backtest report with stated conditions, an overfitting review and a list of suggested parameters within a sensible range.