# IFVG Desk — original v1.0.0 measured cases

Recorded on 6 September 2026 from the actual TradingView Strategy Tester interface. **These eleven original v1.0.0 cells are historical broker-emulator observations, not paper-ledger totals or estimates.** Every recorded case appears below, including losses. The short samples do not establish a transferable advantage.

This table preserves the original source/scenario records without rewriting them as v1.0.1 results. The corrected release's regression observations are recorded separately in [review2-report.md](review2-report.md), with raw data in `docs/review2-broker-measurements.json` when captured. No v1.0.1 result or equivalence is claimed by the eleven cells below.

Raw evidence: [broker-measurements.json](broker-measurements.json). Each case contains the tool response; parse result.content[0].text as JSON and read its result field for the tester text. Some records also contain a separate paper snapshot. The case/scenario and applied input deltas identify the experiment; a reused study title can still say baseline after its inputs change.

## Method and conditions

| Setting | ETF comparison | Futures check |
|---|---|---|
| Chart | QQQ and SPY, standard 1m candles, regular hours | MNQ continuous micro Nasdaq futures, standard 1m candles, available delayed feed |
| Requested symbol/feed | NASDAQ:QQQ and AMEX:SPY; chart data loaded through BATS substitution | MNQ1! delayed futures feed |
| Tester date label | Aug 17, 2026 — Sep 5, 2026 | Aug 31, 2026 — Sep 5, 2026 |
| Date interpretation | Exact UI label; chart display used UTC+7. This is not a separately verified New York start/end-date statement. | Same UI-label limitation. |
| Starting capital | 100,000 USD | 100,000 USD |
| Commission | 0.005 USD per share per fill | 2.25 USD per contract per fill |
| Slippage | One tick | One tick |
| Margin model | 50% each direction | 5% each direction |
| Quantity | Risk-based explicit quantity: planned 100 USD, quantity step 1 | Same sizing rule, using futures point value |
| Entry/session baseline | Inversion close; 09:30–11:00 New York gate on; other confluence gates off | Same |
| Plan baseline | Prior 10-bar swing, one-tick stop buffer, 1R target with outward tick rounding, one resolved paper trade/day, 120-bar timeout, optional session close off | Same |

The cost and margin settings are stated experimental assumptions, not quotations of a particular broker's current charges or requirements. The ETF sources use a suitable per-share cost override; the distributed factory companion remains futures-oriented. The script submits explicit quantity, so Strategy Properties' default order size does not replace its risk/step inputs.

Before each original comparison, baseline rule settings were restored and only the named factor was changed. The raw overrides include resets from the preceding case. The v1.0.0 mechanical engine remained unchanged across this original grid; a later checks-header direction label was a display correction. This statement does not cover the subsequent v1.0.1 fixes. Ten ETF cells compare five configurations on each chart. One futures baseline is a separate portability/measurement check, not a competing optimized candidate.

## QQQ — all five recorded ETF cells

| Scenario | Change from baseline | Trades | Positive | Positive share | Net PnL USD | Profit factor | Max drawdown USD |
|---|---|---:|---:|---:|---:|---:|---:|
| etf-baseline | None | 15 | 5 | 33.33% | −525.57 | 0.299 | 739.62 |
| etf-htf | Require aligned HTF gap overlap | 14 | 8 | 57.14% | +99.41 | 1.162 | 368.40 |
| etf-retest | Retest rejection instead of inversion close | 13 | 2 | 15.38% | −909.36 | 0.177 | 909.36 |
| etf-sweep | Require recent liquidity sweep | 15 | 7 | 46.67% | +65.73 | 1.120 | 621.68 |
| etf-smt | Require direct SMT; requested comparator AMEX:SPY | 8 | 5 | 62.50% | +182.11 | 1.596 | 255.80 |

## SPY — all five recorded ETF cells

| Scenario | Change from baseline | Trades | Positive | Positive share | Net PnL USD | Profit factor | Max drawdown USD |
|---|---|---:|---:|---:|---:|---:|---:|
| etf-baseline | None | 15 | 9 | 60.00% | +210.81 | 1.571 | 278.69 |
| etf-htf | Require aligned HTF gap overlap | 13 | 5 | 38.46% | −387.37 | 0.537 | 696.60 |
| etf-retest | Retest rejection instead of inversion close | 14 | 5 | 35.71% | −432.07 | 0.510 | 638.44 |
| etf-sweep | Require recent liquidity sweep | 15 | 6 | 40.00% | −581.92 | 0.332 | 604.11 |
| etf-smt-reverse | Require direct SMT; requested comparator NASDAQ:QQQ | 8 | 4 | 50.00% | −197.59 | 0.532 | 417.93 |

The word reverse in the final scenario name means the reciprocal symbol pairing, **SPY compared with QQQ**. It does not turn on the engine's Inverse relationship setting.

## MNQ — factory futures baseline

| Scenario | Trades | Positive | Negative | Net PnL USD | Profit factor | Max drawdown USD |
|---|---:|---:|---:|---:|---|---:|
| futures-baseline | 5 | 5 | 0 | +322.00 | Undefined; tester displayed — | 64.50 |

The tester reported 100.00% positive trades in this five-trade sample and zero gross loss. Profit factor has no defined finite denominator here; it is **not** reported as infinity or evidence of a perfect method. Five trades on a short delayed-feed history cannot settle whether the rule is useful on futures. Do not promote this cell while suppressing the ETF losses.

## Paper snapshots are separate evidence

The raw log contains these supplemental QQQ snapshots. They are before costs and are not the broker results above.

| Case | Paper closed plans | Paper total R | Broker closed trades |
|---|---:|---:|---:|
| Retest rejection | 15 | −7.00 | 13 |
| Recent sweep required | 15 | +0.8054 | 15 |
| SMT versus SPY required | 8 | +2.00 | 8 |

The retest count difference is direct evidence that the records are not interchangeable. This log does not provide a trade-by-trade reconciliation explaining every difference. The engine's stop-first paper convention, broker fills/costs, sizing or margin constraints, and skipping a new paper signal while broker exposure remains open can affect the broker sequence. Those are possible mechanisms, not a retrospective attribution of each missing trade.

The initial indicator runtime record separately showed QQQ baseline n=15 and −5.16R in the Data Window; the dashboard rounded this to −5.2R. Display/theme/depth changes preserved that paper record. The QQQ 5m runtime observation showed n=45 and approximately −6.1R, but it used a different timeframe/history and is **not** an additional broker-comparison cell. Source: [runtime-evidence.json](runtime-evidence.json).

Final v1.0.0-source OANDA:XAUUSD 5m also ran with Auto 15/30/60m context. Its indicator displayed n=20, eight positive and −4.0R before costs, with a last short ticket of 4413.285 / stop 4434.931 / target 4391.639 and quantity estimate 4. This is a non-index runtime observation, **not a twelfth broker result**. No gold Strategy Tester cell was recorded in this grid. Source: [runtime-final-evidence.json](runtime-final-evidence.json).

For that gold chart, switching 5m → 3m → 5m produced identical extracted full tables and 17 label text/price pairs, including three prior paper plans. This tests historical recalculation on unchanged closed data, not app restart or live ticks. Source: [recalculation-evidence.json](recalculation-evidence.json). An attempted MNQ intraday replay was unavailable behind the Basic plan upgrade UI; it contributes no replay result.

## What the grid says

- Baseline has opposite signs on the two ETFs: QQQ −525.57 USD; SPY +210.81 USD. There is no common baseline result to advertise.
- HTF overlap also changes sign by chart: it improves the observed QQQ cell but turns SPY's baseline into a loss. The sweep and SMT gates show the same cross-chart disagreement.
- Retest rejection is lower than inversion-close baseline on both observed ETF samples. That is an observation about these cases, not a general law about retests.
- The SMT cells contain only eight trades each. The MNQ cell contains five. Their small samples are not stronger evidence because one printed attractive numbers.
- No gate was promoted to a required default by picking a winner from this grid. Structural defaults remain a simple baseline; factor switches remain available to study.

## Reproduce the configuration, then repeat the observation

Use tools/make-scenario.mjs with etf-baseline, etf-htf, etf-retest, etf-sweep, etf-smt, etf-smt-reverse or futures-baseline. The current tool writes a new scenario copy and manifest under build/scenarios-generated without overwriting distributed factory source or the original v1.0.0 records under build/scenarios. The manifest records the current source hash, overrides and cost assumptions. It recreates a configuration using the current implementation; it does not reproduce the old source byte-for-byte or freeze future feed revisions and loaded history.

Add a fresh study or explicitly reset inputs before each cell, verify the applied settings, allow the tester to finish and save its actual output. A name change or source update alone does not prove that saved inputs changed. Preserve the feed, sessions and UI timezone alongside the date label. A separate held-out period, fuller futures history and trade-by-trade reconciliation remain unmeasured here.