# REVENQOR · SMA Strategy Audit

**SAMPLE / DEMONSTRATION — not a client project.** SAMPLE-SMA-001, v1.0, 30 September 2026. Self-authored educational example. No real market data or trading performance.

## 1. Strategy specification

Starting brief: “Buy when the fast moving average crosses above the slow one. Sell on the reverse crossover. Use SMA 20 and 50.” This deliberately incomplete brief is the only strategy specification reviewed. No EA, Pine source or tester report was supplied.

Proposed clarification: long-only; SMAs of fully closed bars. Entry at t requires Fast(t−1) ≤ Slow(t−1) and Fast(t) > Slow(t). Exit on the reverse crossover. At most one open position, no repeated entry from an already processed signal, at least 51 closes for the initial comparison. Earliest execution is after the signal close, not retrospectively at that price. Market, timeframe, broker, sizing, protective stop and failure rules remain undefined.

## 2. Assumptions

SMA 20/50 are demonstration parameters, not optimized settings. Closed-bar evaluation and long-only operation are audit proposals, not approved client requirements. No suitable market is asserted. Position sizing, protective orders and loss limits must be specified. Costs and initial capital remain unknown.

## 3. Backtest design

Freeze rules and version; define data source, time zone, execution and costs; reserve development and untouched out-of-sample periods; log every attempted variant. Validate data gaps, duplicates and session alignment. Archive a reproducible baseline before sensitivity studies and a separate demo forward test. Revising the system after inspecting OOS results consumes that period's independence.

An executed **synthetic signal-logic check** uses `[6,5,4,3,4,5,6,5,4,3]` and shortened SMA windows 2/3. Expected upward crossover: index 5; downward: index 8; earliest execution indices: 6 and 9. Five checks pass: known crossovers, insufficient warm-up, flat series, future-data independence for earlier signals and execution scheduled after the signal. Reproduce with `node scripts/audit-sample.cjs`. These are artificial inputs, not a market backtest or order simulation.

## 4. Metrics

Profit factor, win rate, average trade, average R, equity, balance, drawdown, trade count and sessions: **unavailable / not calculated**. Two synthetic crossovers do not represent two executed trades. A later report must define net/gross basis, break-even classification, cashflow-adjusted equity and initial risk for each R multiple.

## 5. Issues found

| ID  | Priority | Evidence in the brief                      | Action / retest                                                             |
| --- | -------- | ------------------------------------------ | --------------------------------------------------------------------------- |
| F01 | P0       | No bar timing                              | Define closed bars; test prefix independence and platform realtime behavior |
| F02 | P0       | No sizing, stop, exposure or restart rules | Specify risk and broker constraints before order testing                    |
| F03 | P1       | No source, period or costs                 | Complete a versioned manifest; missing evidence blocks publication          |
| F04 | P1       | No OOS or variant policy                   | Reserve untouched data; log every trial                                     |
| F05 | P2       | No failure or duplicate handling           | Specify position reconciliation, signal IDs and failure tests               |

## 6. Bias risks

Treating an open bar as closed, tuning parameters after viewing results, selecting the strongest period, omitting costs and hiding failed trials. These are risks inferred from missing rules, not proven defects in existing trading software. The prefix check covers a small signal function, not data feeds or TradingView repainting.

## 7. Suggested improvements

Resolve P0 requirements; build a minimal prototype separating signal, execution and risk; complete the data manifest; archive a baseline with every trade; evaluate OOS, sensitivity and demo forward evidence separately. Explicitly record unavailable evidence.

## 8. Automation readiness

**Blocked for trading execution.** The proposed signal core can be specified and passes limited synthetic checks. Risk, market/data contract, position state, protective controls and realistic execution tests are missing. The next useful step is specification work, not a promise of more profitable parameters.

## Limitations and disclosure

Self-authored demonstration with no client, live funds, real market or performance record. Not investment advice or a promise of profits. An audit improves traceability and testability and may recommend discontinuing an idea.