Skip to main content

Validation

HyperionX Validator runs a compiled strategy across one or more instruments and data-series configurations. Use it to compare components, inspect portfolio totals, reopen saved presets, export a summary, and send completed results to performance analysis.

Validation is different from optimization. Optimization selects parameter combinations from historical data. Validator tests the strategy and parameters you assign.

Verify the installed research components

The project changelog records product-owned research defaults in signed release 1.1.11 and a signed 1.1.12 MSI containing the validated Python Research Runtime. This documentation audit did not independently test the distributed MSI. Verify its signature, confirm Help > About, and make sure Validator's required modules appear after first-launch compilation. If they do not, collect diagnostics and contact support; do not import untrusted modules as a workaround. See Install, Update, And Uninstall.

Create A Validator Run​

  1. Open Validator.
  2. Select Add instrument for a single configuration, or Add test set to load a saved group.
  3. Choose the connection, instrument, data series, bar type, interval, and date range for each item.
  4. Load historical candles for every item you intend to run.
  5. Select a compiled strategy.
  6. Assign the strategy to each item, or use Apply all when the same strategy and preset should be shared.
  7. Review strategy parameters, commission, leverage, historical fill model, and continuous-futures merge behavior where applicable.
  8. Select Run.

Run is enabled only when the required strategy and historical candles are available for the relevant items. Use Cancel to stop an active run.

Unreleased 1.1.14 multiple-timeframe cases

The signed 1.1.12 workflow above uses one data-series configuration for each Validator item. Reviewed 1.1.14 development source adds multiple timeframe cases within a Validator item and preserves that expanded configuration in the unified strategy-workspace snapshot. This is not part of the signed 1.1.12 baseline; wait for and verify a matching public release before depending on it.

Work With Test Sets And Presets​

Validator can:

  • Add individual instruments.
  • Load and save test sets.
  • Load and save strategy presets.
  • Apply one strategy configuration to multiple items.
  • Preserve bar type and custom-bar settings in a preset.
  • Preserve backtest leverage, fill model, and continuous-futures merge settings.
  • Re-run changed configurations.

Give each test set one clear purpose. For example, compare one futures strategy across related contracts or one crypto strategy across a defined group of markets. Combining unrelated instruments can make an aggregate curve look smoother without explaining why.

Review Results​

Start with the aggregate result, then inspect every component:

  • Net profit and max drawdown.
  • Trade count and average trade.
  • Profit factor.
  • Commission.
  • Long and short contribution.
  • Instrument contribution.
  • Ambiguous historical fills.
  • Performance concentration in one market or date range.

Completed Validator results are saved to the performance-results store. Validator also provides a summary export and a Find Improvements research action.

Find Improvements is an analysis aid. It does not replace out-of-sample testing, playback, simulation, or a review of the underlying trades.

Historical Assumptions​

Validator uses candle-based historical fill modeling. See Backtesting for the two current fill models and their ambiguity behavior.

Treat the signed public 1.1.12 contract as not exposing a documented configurable slippage model for ordinary Validator or Optimizer runs; the project changelog classifies the newer setting as Unreleased. Commission is modeled separately when a commission module is available. Record spread, slippage, liquidity, and market-impact limitations outside the saved result.

Unreleased source version 1.1.14 adds a Validator-only adverse tick-slippage setting for historical Market and StopMarket fills; historical Limit fills remain unchanged. That development behavior is not part of signed release 1.1.12 and does not add an Optimizer slippage control. Wait for and verify a matching signed public release before treating it as available.

Strategy Readiness Checklist​

Before assigning a strategy to Validator:

  1. Compile it in Code Lab with no errors.
  2. Confirm it uses the HyperionX.Custom.Strategies namespace and inherits from Strategy.
  3. Keep managed order mode enabled for normal strategies.
  4. Use EnterLong(...), EnterShort(...), ExitLong(...), or ExitShort(...) with managed stop-loss and profit-target helpers where appropriate.
  5. If the strategy deliberately uses direct SubmitOrder(...), verify order action, order type, prices, OCO values, signal names, working-order tracking, cancellation, rejection, partial-fill, and recovery paths.
  6. Guard against duplicate entries.
  7. Verify quantity, exits, stops, targets, commission, leverage, and margin assumptions.

After historical validation, run the strategy in playback or simulation with an explicitly selected LocalPaper account before considering live routing. Playback does not select it automatically. Do not rely on build-specific external-account guards as a substitute for verifying the account.

Indicator And Workspace Checks​

Validator is a strategy-research tool. Use direct application checks for other areas:

  • Load an indicator on a chart with enough bars and watch Logs for runtime errors.
  • Save and reopen a workspace to verify charts, instruments, timeframes, indicators, and window layout.
  • Confirm a connection can load history, receive live updates, and reconnect.
  • Verify manual orders only in the intended paper/test account before live use.

AI-generated scripts require the same compilation, historical review, playback, and simulation checks as hand-written scripts.

Saved-Result Limitation​

The results database preserves useful identifying information, headline metrics, and trade artifacts, but it does not currently preserve every setting required to reproduce a run. Keep strategy parameters, commission configuration, starting capital, application version, and research notes separately.

See Performance Database.