Backtesting Engine
Test a strategy against years of market history.
A backtest is only useful if you can configure it honestly and interrogate it afterwards. Pulsar lets you set the capital, the window, the universe and the execution assumptions — then hands back every trade, not just a headline return.
A configurable engine
You control the assumptions, not a preset.
Default backtests flatter strategies. Pulsar exposes the settings that decide whether a result is real — so the run you keep is the one you actually believe.
01 — SCOPE
Capital, window and universe
Set the starting capital, the historical range and which instruments take part. Run a single symbol to understand behaviour, or a whole universe to see whether the edge generalises.
- Starting capital and position sizing basis
- Custom date ranges across years of data
- Single symbol or multi-symbol universes
02 — EXECUTION
Costs and fill assumptions
Frictionless backtests are fiction. Configure the execution model so the simulated fills resemble what a broker would actually have given you.
- Commission and fee treatment
- Slippage assumptions
- Order handling and fill behaviour
03 — EVIDENCE
Every trade, on the chart
Results are plotted onto the price series with entry and exit markers, so you can see exactly where the strategy acted — and where it sat on its hands through a move you expected it to take.
- Entry, exit and signal markers on the series
- Per-trade inspection
- Timing analysis across the run
04 — DIAGNOSIS
Where the result came from
The engine breaks the run into signal generation, entry funnel and exit quality, so a disappointing number points at a stage you can fix rather than a strategy you must abandon.
- Signal counts and filter attrition
- Entry accuracy and conversion rate
- Exit quality against the ideal exit
05 — DISTRIBUTION
Outcomes instead of averages
Trades are bucketed by return band so you can see the actual shape of the result. A strategy carried by two outliers looks very different here than it does in a mean.
- PnL distribution by outcome band
- Win and loss magnitude profiles
- Hold-time and drawdown behaviour
06 — ITERATION
Compare, tune, run again
Backtesting is a loop, not a verdict. Adjust a parameter in the Strategy Creator, run it again, and hold the two results next to each other before you decide anything.
- Re-run against an unchanged window
- Compare variants side by side
- Promote the version that survives
A word on backtests
Simulated results are evidence, not a promise.
A backtest describes how a rule set would have behaved on data that has already happened. It cannot account for every fill, every gap or every regime change ahead of it. Pulsar gives you the tools to test a strategy rigorously and to see its weaknesses — it does not tell you the future, and neither does any number it produces.
Next in the workflow
From a result to a decision.
Analytics Workspace
Take the run apart with risk-adjusted ratios, distributions and attribution.
Explore analyticsStrategy Creator
Change the rules that produced the result and run the whole thing again.
Explore the builderPortfolio & Live Trading
Move a strategy that earned its place into paper or live execution.
Explore the portfolioPut your strategy under pressure.
Pulsar is in closed beta. Create an account, or tell us what you are trying to build.
Backtested and simulated results are hypothetical and are not indicative of future results.