Report 2026-W42 · 2026-10-16

What changed

There is no earlier published weekly report to compare with, so “changed” means the most recent update to this site's data (the newest entry in the validation log, which also lists earlier same-day entries):

• All-Instrument Stack: mixed NQ + ES + GC set computed (replaces the earlier NQ + ES reference set)
• Source: TV4REAL's all-instrument file, run on the saved TradingView trade lists (one position at a time across all three instruments; same costs and window as everything else; year split at exits 2025-09-23). Site numbers are recomputed from the trade lists and match the file exactly.
• Tested set (priority order): ES B12_01 1h > NQ Supertrend Flip 1h > ES B8_03 30m > GC Coral 30m > NQ DPO 15m > GC Elder Force 30m. Net +$210,412, PF 1.94, win 48.7%, 433 trades taken of 611 signals (178 blocked), closed-trade max DD $7,138, net ÷ DD 29.5×, average entry-day overlap 21%.
• Year 1 +$78,825 (PF 1.80); year 2 +$131,588 (PF 2.05) — 63% of the net in year 2. Last 3 months: 53 trades, +$13,672, PF 1.43. By instrument (taken trades): NQ 150 trades +$82,745 PF 1.92; ES 70 trades +$51,312 PF 2.33; GC 213 trades +$76,355 PF 1.81. The five best trades are 45% of net. Weakest full quarter: 2025 Q3 (+$3,852, PF 1.16).
• Compared with the single-instrument stacks (same engine and window): higher net and lower drawdown than all of them (net ÷ DD 29.5× vs 12.4× NQ best set, 12.6× GC best pair, 6.6× ES best-PF pair), but a LOWER profit factor than the best NQ set (2.02), the ES pair (2.06), ES B12_01 + B12_03 (2.56) and GC Coral + G4 (2.59). Higher than the GC best set (1.82).
• FLAGGED — selection bias: the set is the product of a search over 16,455 mixed subsets of 17 strategies, all scored on the same two years, so net ÷ DD (29.5×) is optimistic; neighbouring sets are nearly as good, so the exact membership is not special. The window is a strong uptrend.
• FLAGGED — three of the six members are Tier B: DPO failed long-history validation (2019–2026 PF 1.29, DD $21,275, 2020-crash PF 0.28, walk-forward positive in 48% of folds) yet is the biggest contributor (119 trades, +$56,005); Coral lost $4,710 in its own year 1 (PF 0.73, 107% of its net in year 2); Elder Force lost $4,145 over the last 3 months (PF 0.78). Coral and Elder overlap heavily (73% of Coral's entry days).
• Reconciling earlier wording: the NQ Stack still leaves DPO (and Kalman) out because inside the NQ-only stack DPO's taken trades net −$785 and DD rises to $14,655; in the mixed set DPO fills slots the other markets leave open. Both are in-sample statements. DPO stays flagged (Tier B) and the earlier 'Tier B stays out of headline numbers' wording now applies to the NQ-only numbers.
• Leave-one-out (net / PF / DD): without DPO $145,955 / 1.86 / $9,562; without Coral $170,528 / 1.81 / $8,758; without ES B12_01 $181,888 / 1.80 / $14,018; without Supertrend $191,580 / 1.89 / $13,075; without Elder $203,625 / 2.18 / $9,260; without ES B8_03 $211,512 / 1.97 / $9,740 (B8_03 nets only $4,668 here).
• Priority order matters for net: all 720 orders keep closed DD at or under $15k (DD $7,138–$10,328) but net ranges $195,535–$247,328. The stated rarest-first order was kept rather than the best-net order to avoid another layer of selection.
• Not done: neighbour timeframes, ±15% parameter nudges, per-year PF ≥ 1.15 for every member, next-bar-open fills. Backtest only; no forward results. The All-Instrument Stack stays 'building' / waitlist (billing not open).
• The earlier NQ + ES reference set (Connors + ES B12_01 + half-size VZO: +$102,042, PF 2.02, DD $12,175) is kept in the combined-sets table, marked superseded.

Status by strategy

Status legend: Pass Meets the pass bar on the most recent run. · Watch Still passes on the main window, but recent trades or an extended window show weakness. Subscribers are told; position-size caution advised. · Flagged Passes only under conditions (e.g. half size) or carries a known fragility. Kept out of headline numbers and marked low priority. · Removed Fails the bar or a robustness check. Taken out of the stack and announced in the weekly report. (None removed yet.)

Not yet run for any strategy: neighbour-timeframe checks, ±15% parameter nudges, per-year PF ≥ 1.15 across the board, next-bar-open fills. They will be listed here when completed.

Status by stack

In-sample backtest figures on one 2-year window; year splits use each set's own cut-off (see the performance page). Hypothetical results; expect live results to be worse. Forward results: none yet.

Past reports

New reports are added at the top and past reports are never edited; corrections are listed in the next report. The running changelog is on the validation page.

What every weekly report will contain

Subscribers get the full report by email each week. This is the structure — the same sections every time, including the unflattering ones.

# Stacked Edge — Weekly validation report
Report ID: YYYY-Www   ·   Run date: YYYY-MM-DD   ·   Data through: YYYY-MM-DD
Window: rolling 2 years   ·   Costs: 4-tick slippage per side + $2.50/side commission   ·   Fills at bar close   ·   Backtest = hypothetical

> A weekly report is a process, not a guarantee. It records whether each strategy still meets the pass bar on the latest data.
> It cannot predict future results. Futures trading involves substantial risk of loss.

## 1. Summary
Pass: n   ·   Watch: n   ·   Flagged: n   ·   Removed: n      (change vs last report: +/- n)
One paragraph in plain English: what the data said this week, including anything unflattering.

## 2. What changed since the last report
- Status changes:  <strategy>: <old> -> <new> — <reason, with the number that triggered it>
- Stack changes:   members added / removed / priority order changed — <reason>
- Method or data changes: (anything that makes numbers not comparable with last week)
- Corrections to earlier reports (never edit past reports; list corrections here)

## 3. Strategy table
| Strategy | Mkt / TF | Status | Net (2y) | PF | Max DD | Trades | Last 3 months | Year 1 / Year 2 PF | Note |

## 4. Stack table
| Stack | Set (priority order) | Net | PF | Closed-trade DD | Trades taken / entry setups | Year 1 | Year 2 | Last 3 months |

## 5. Follow-along posts this week (subscribers)
Posts published, taken vs skipped (one position at a time), and outcomes of closed trades. Losers included.

## 6. Checks run / not run
Neighbour timeframes, +/-15% parameter nudges, per-year PF >= 1.15, next-bar-open fills: run (with result) or not run.

## 7. Known issues & caveats
In-sample selection, single 2-year window, overlap between strategies, small samples, anything flagged.

Risk disclosure: hypothetical/backtested results have inherent limitations; no guarantee of profit; education only; not financial advice.