Skip to content

Simulator and evidence

Simulator assumptions and data completeness

Paper deployments and backtest jobs use the Venue Simulator as their execution destination. Simulator-backed results are useful only when the assumptions and data coverage are visible.

These results are evidence under explicit assumptions. They do not predict live fills, guarantee returns, guarantee risk reduction, or guarantee live execution quality.

Simulator assumption flow Simulator-backed results are produced by combining selected strategy context, data, account state, and explicit assumptions.
Strategy and run context
Selected revision Revision-owned venue and instrument
Strategy Variable values Run-provided values
Historical or live data Backtest job range or paper deployment stream
Assumptions
Starting balances or simulated account state Initial simulated context
Fees, slippage, latency Execution cost and timing assumptions
Fill model and liquidity assumptions How order activity becomes simulated fills
Data completeness Complete segments and gaps for backtest jobs
Simulated execution
Target Position Executor Produces order activity
Venue Simulator Does not model the full exchange order book
Outputs
Simulated fills and account state Positions, balances, fills, and order state
Result metrics Evidence under assumptions, not guarantees
Paper deployment Uses live market data and simulated execution for a long-running evaluation.
Backtest job Uses historical market data and data completeness for a finite historical evaluation.
Legend
  • Market data Live or historical market data sources.
  • Target Position Executor Target-position intent converted into order activity.
  • Venue Simulator Simulated execution destination and assumptions.
  • Account state Live or simulated positions, fills, balances, and order state.
  • Results Recorded metrics, telemetry, and review context.
  • Paper path Long-running paper deployment boundary.
  • Backtest path Finite backtest job boundary.
  • Revision-owned Immutable strategy revision logic and settings.
  • Run settings Run-provided values, balances, ranges, or assumptions.

Simulator-backed evaluation depends on several groups of assumptions:

Assumption groupWhy it matters
Starting balances or simulated account stateDefines the initial capital and exposure context.
FeesAffects PnL, drawdown, and cost sensitivity.
SlippageAdjusts simulated execution away from visible reference prices.
LatencyChanges the relationship between market data, target-position changes, and simulated fills.
Fill modelDetermines how order activity becomes simulated fills.
Liquidity assumptionsDescribes how much simulated order activity can be evaluated against available market context.
Data completenessShows which requested historical periods were available for evaluation.

Review assumptions before comparing jobs or promoting a revision toward live operation.

Starting balances and simulated account state

Section titled “Starting balances and simulated account state”

Starting balances define the simulated account state at the beginning of a backtest job. For paper deployments, simulated account state may be selected or initialized when the deployment starts.

This context affects position sizing, margin behavior where modeled, PnL, and drawdown. A result from one starting balance should not be compared casually with a result from a different starting balance.

Fee assumptions apply costs to simulated trading activity.

Fees can materially affect strategies with frequent order activity. When comparing results, keep fee settings visible and avoid comparing a low-fee result against a high-fee result unless fee sensitivity is the purpose of the comparison.

Slippage assumptions make simulated execution less or more favorable relative to the observed market data context.

Higher slippage can test whether a strategy remains acceptable under less favorable execution. Lower slippage can show behavior under less conservative assumptions. Neither setting guarantees what a live venue will do.

Latency assumptions represent delay during simulated evaluation.

Latency can affect when target-position changes turn into simulated order activity and fills. This is especially important for strategies that react frequently or operate in fast-moving markets.

The fill model determines how simulated order activity becomes simulated fills. Liquidity assumptions describe how much simulated order activity can be evaluated against available market context.

These assumptions are central to result interpretation. They can affect:

  • Whether simulated order activity fills.
  • How position changes over time.
  • PnL and drawdown.
  • Trade count.
  • Sensitivity to larger target positions.

The current Venue Simulator does not model the full exchange order book.

That limitation matters because live venue behavior can include depth, spread changes, queue position, cancellations, partial fills, rejected orders, and other microstructure effects. Simulator-backed results should not be read as predictions of live fills.

Historical data coverage affects every backtest job.

When the requested range contains gaps, Structure surfaces data completeness so users can see which parts of the requested period were evaluated. A metric summary without data coverage can be misleading.

Use data completeness to answer:

  • Which requested periods had available historical data?
  • Which periods were excluded because data was unavailable?
  • Which results use the same coverage?
  • Whether a comparison is based on equivalent data.

Complete segments are the portions of the requested range that had enough available data for evaluation. Gaps are portions where required data was unavailable.

If two jobs have different complete segments, their metrics may not be directly comparable even if the selected revision and Strategy Variable values are similar.

Use these rules when comparing simulator-backed results:

  • Keep the selected revision attached to each result.
  • Keep Strategy Variable values attached.
  • Compare over the same historical range when possible.
  • Compare with the same starting balances when possible.
  • Keep fees, slippage, latency, fill model, and liquidity assumptions visible.
  • Check data completeness before ranking metrics.
  • Treat a single changed assumption as a sensitivity test, not as a guaranteed future outcome.