What it is
This article explains the settings and market conditions you can use when building simulator policies - entry filters and stop/target replay. It helps you interpret results and build meaningful rules without needing to know how Tessera computes every label behind the scenes.
What you can do
- Scope simulations to the right account, dates, symbol, and side
- Build entry filter trees with AND/OR groups
- Filter on sessions, regimes, volatility, structure, and alignment at entry
- Replay stop-loss and take-profit distance changes on past trades
- Understand coverage, baseline vs filtered metrics, and when labels show as unknown
How it works
Scope parameters
| Setting | What it controls |
|---|---|
| Account | Which broker account’s closed trades are included |
| Date range / timeframe | Which trades by exit date (e.g. last 30 days, custom range, all-time) |
| Symbol | Optional - limit to one instrument |
| Side | Optional - long only, short only, or both |
The cohort uses the same date and timezone rules as the Trade Log. See Timezones and dates.
Entry filter actions
| Action | Effect |
|---|---|
| Remove | Trades that match the condition tree are removed from the filtered set |
| Keep only | Only trades that match the tree are kept |
You can nest conditions in AND / OR groups. Each leaf compares a market field at entry to a value (equals, in list, greater/less than for numbers).
No lookahead: every field is evaluated from candles at or before your entry timestamp.
Market conditions you can filter on
Trading sessions
Whether the trade opened during a named session window, such as:
- London, New York, Tokyo, Sydney
- Overlaps (e.g. London-Tokyo, London-New York)
Useful if your edge is session-specific.
Market condition (regime)
A broad label for the state of the market at entry:
| Label | Plain meaning |
|---|---|
| Accumulation | Quiet, compressed conditions |
| Expansion | Volatility picking up |
| Trend | Directional movement is dominant |
| Range | Sideways, mean-reverting conditions |
| Unknown | Not enough data at entry to classify |
Direction - longer horizon
Overall price drift leading into your entry: up, down, or flat. Flat means the move was too small to treat as clearly directional - not “no data.”
Direction - shorter horizon
Recent drift just before entry: again up, down, or flat with the same flat rule. Use both horizons to distinguish a long-term trend from a short-term pullback.
Structure (trend strength)
Buckets such as weak, medium, or strong trend - based on trend-strength style measures (e.g. ADX-style) at entry.
Volatility
Several related fields, including:
- ATR regime - whether volatility is low, normal, or elevated relative to recent history
- Bollinger squeeze - whether bands are tight (compression) or not
- Bollinger zone / position - where price sits relative to the bands
Higher-timeframe alignment
A score from 0 to 1 indicating how well longer-timeframe direction aligns with your trade direction (long or short). Higher = more agreement.
Entry window statistics
Numbers describing the price path up to entry, for example:
- Net percentage change over the window before entry
- Range (high − low) as a percentage of starting price
- Recent change over a shorter slice of that window
Useful for momentum or mean-reversion style filters.
Trade direction
Filter on whether the trade was long or short.
Exit replay parameters (stop-loss and take-profit)
When simulating management rules (not entry-only filters):
| Parameter | Meaning |
|---|---|
| Stop distance override | Counterfactual stop placed at a specified distance from entry |
| Take-profit distance override | Counterfactual target at a specified distance from entry |
| Replay legs | Test stop only, target only, or both |
Tessera scans historical candles from entry through your actual exit (plus a forward window when needed) to see whether stop or target would have been hit first on each bar.
Simulated PnL adjusts proportionally from your real trade - it does not re-size positions or re-simulate commissions from scratch.
Exit Placement reports use the same engine with recommendations per market context slice.
Reading results
Baseline - metrics for all closed trades in scope.
Filtered / simulated - metrics after your policy.
Delta - simple difference between the two.
Coverage
- Trades total - how many closed trades in scope
- Trades with entry bar - how many had a usable candle at entry for context
- Trades missing entry bar - gaps in market data; those trades may get unknown labels or limited condition evaluation
If coverage is weak, treat filter results with extra caution.
Workspace policy inputs (quick connection)
Workspace nodes use the same policy inputs you build in the simulator:
- Entry filter trees (market conditions evaluated at entry) when your node removes or keeps trades
- Exit replay parameters (stop-loss and take-profit distance overrides) when your node changes exit management
If you are unsure how a specific session/regime/condition label works, or how coverage and “unknown” labels should be interpreted, use the same simulator interpretation rules for Workspace inputs.
For the node + sequence view of how these inputs connect, see Workspace - nodes, sequence, and policy inputs (and the overview at Workspace overview).
Limitations
- Only stop-loss and take-profit distance replay is supported for exit management today - not breakeven or trailing stops.
- Entry filters do not change per-trade PnL unless you are in exit-replay mode.
- Labels depend on market data quality; missing candles → unknown or reduced coverage.
- Condition trees are capped at 16 levels of nesting and 256 rules and groups total per policy - split across policy versions if you need more.
- The simulator still cannot model new entries or trade-sequence cascades - see Simulator overview and Understanding when results change.