The Execution Truth Update: 20 New Research & Simulation Features

Most trading platforms compete on how easy it is to produce a buy or sell signal. We think the harder — and more valuable — problem is the integrity of the chain from observation to decision to execution. Today we're shipping 20 features built around one question:

"Does this help you know what is true, understand what remains uncertain, and retain meaningful control over the consequences?"

Here's what's new, in two families.

Family 1: Execution Truth

A simulator that trades often but models reality poorly teaches the wrong lesson. These ten features make the simulation's assumptions explicit, selectable, and honest.

1. Market Session & Stale-Candle Gate

Every price now has an age and a session context. Bots no longer evaluate signals on Saturday's frozen candle: FX and US-equity calendars (DST-aware, holidays included) plus candle/quote freshness checks gate every scan, with your choice of reject, defer, or evaluate-without-fill. Every blocked evaluation is explained in the run timeline.

2. Simulator Reality Profiles

Choose the world your simulation lives in: Idealized, Retail Typical, Thin Market, News Shock, or User Calibrated. Each controls slippage, spread widening, and deterministic order rejection — and the run page shows your net P&L under all profiles side by side. If your strategy is only profitable in the idealized world, you'll see it before the market shows you.

3. Partial Fills

A 500,000-unit order against a thin bar doesn't fill — and now the simulator says so. Bar-volume participation caps fills, cancels remainders with explicit reasons, and records requested vs. filled quantity on every position.

4. The Order-Type Laboratory

Market, limit, stop, stop-limit, bracket, OCO, Day/GTC — tested on deterministic price scenarios with full event timelines. Our favorite lesson: watch a gap open fill your plain stop 30 pips past its trigger while the stop-limit refuses the bad price and misses the move. Both are correct; they're different promises.

5. FX Financing & Rollover

Holding EUR/USD long overnight costs money right now — and your simulation finally knows it. Interest-rate differentials, Wednesday triple swaps, and per-broker markup schedules accrue financing on every held position, reported separately from trading P&L.

6. Margin & Liquidation Sandbox

Inject a price shock against your whole book and watch the margin level curve cross the margin-call line, then the stop-out line — and see exactly which positions get force-closed first. Leverage, understood before it destroys choice.

7. Corporate Actions Ledger

A 4:1 stock split looks like a 75% crash to an indicator computed on raw data. The ledger shows every split and dividend and exactly how it rewrites price history.

8. Execution Quality (TCA) Grades

A profitable decision can still be executed badly. Every run gets a grade from arrival-price slippage and fill latency — how much value leaked between deciding and executing.

9. Broker Capability Contracts

An honest matrix of what each broker can actually express — order types, attached stops, minimum sizes, netting vs. hedging — plus a one-click check of whether your strategy fits that broker, with blockers surfaced before deployment.

10. Market Data Quality Passports

Every conclusion inherits the weaknesses of the data beneath it. Grade any dataset A–F on completeness, gaps, duplicates, outliers, and freshness — downloadable as evidence.

Family 2: Research Science

A strategy result should be an inspectable scientific claim, not a screenshot.

11. Dimensionally Safe Strategies

Comparing RSI to a price is like comparing kilograms to kilometers. Every indicator output now carries a semantic unit, and meaningless comparisons are flagged before they silently break your logic.

12. Strategy Contracts

Declare what must never happen — stop tighter than the spread, risk above your ceiling, orders on stale data — and get pass/fail verdicts with concrete counterexamples.

13. Parameter Sensitivity Surfaces

Is your strategy on a plateau or a knife edge? A 5×5 parameter sweep re-backtests every combination on identical candles and scores fragility. Isolated bright cells are usually curve-fitting, not edge.

14. Ablation Lab

Remove each condition, filter, and exit one at a time and re-test. Components rank as load-bearing, decorative, or actively harmful. Complexity must earn its keep.

15. Counterfactual Trade Lab

For any closed trade: what if you'd skipped it, entered a bar later, halved the size, or moved the stop? Replayed on the same candles with the same information — labeled as analysis, never rewritten history.

16. Signal Decay & Half-Life

Edges are temporary relationships. Rolling hit rate and expectancy show whether yours is fading, an estimated half-life says how fast, and a latency test reveals whether two bars of delay kills it.

17. Capacity & Market Impact

Returns are partly a function of how little money is asking for them. Find the position size where impact costs consume your per-trade edge — before scaling teaches you the expensive way.

18. Benchmark Attribution

Your bot made 3%? Buy-and-hold made 4% over the same window. Every run now answers "was this skill or market drift" with buy-and-hold comparison, correlation, beta, and a plain-language verdict.

19. Experiment Registry

Research becomes hindsight when the hypothesis is written after the result. Preregister what you expect and your failure threshold, lock the record, attach runs as evidence, and keep an honest accepted/rejected ledger — permanently.

20. Reproducible Run Manifests

Every run can emit a deterministic manifest — engine version, strategy content hash, settings, seed — with its own SHA-256, plus a one-click Reproduce This Run that pins the exact strategy snapshot. A result you can't reproduce is a story, not evidence.

Where to start

• Open any run page — the Execution Reality, Counterfactual, and Run Evidence cards are already there.
• Open any strategy's Map — contracts, sensitivity, and the Research Lab are one click away.
• Try the Order Lab and Margin Lab — five minutes in each will change how you think about execution.

Full documentation lives in Execution Truth & Research Labs.

Not financial advice. Trading involves risk of loss. Simulated results do not guarantee live performance — that's rather the point of these tools.