Backtesting
What did your strategy assume about the past?
Use historical tests to examine behaviour and limitations, not to predict what a market will do next.
A test you can inspect
We configure a historical test from your written rules and an agreed data source, then record the period, instruments, transaction-cost assumptions and known gaps. A focused setup commonly takes one to three weeks once the rules and usable data are available. You receive a test configuration, summary tables and notes on interpretation. Historical results are sensitive to data quality and assumptions; they do not establish how a strategy will behave in future markets.
Four questions to ask of any backtest
Which data was used?
Check its source, coverage, adjustments and missing periods. A chart can look complete while its underlying records omit difficult trading conditions.
What did execution cost?
Review fees, spreads and slippage assumptions. A historical signal is not the same thing as an order that could have filled at that price.
Was the rule changed later?
Keep a record of revisions made after seeing results. Repeatedly tuning a rule to one historical period can make the test less informative.
Which periods were difficult?
Inspect weak periods and changes in market conditions rather than relying on one summary figure. The exceptions often explain the operating burden.
What a historical test cannot settle
A backtest cannot reproduce every queue position, rejected order, data interruption or change in market structure. It also cannot repair a vague strategy: unclear instructions create unclear results. We document the assumptions we can test and flag those that require judgement. The useful outcome is a more informed review of the rule set and its operational demands, not a claim about future performance or a recommendation to trade.
Inputs needed for a focused test
Clear source material helps distinguish a strategy issue from a data issue.
- A fixed version of the trading rules
- The instruments and historical period
- A usable historical data source
- Fee, spread and slippage assumptions
- Rules for missing or invalid observations
The most useful test report often shows where a strategy is hard to explain. Those findings deserve attention before another parameter is adjusted.
Intersparc product team
Ask about testing your rule set
Tell us what data and documentation you have, and we will outline a suitable test scope.
Ask about backtesting