Research series
The desk generates a 5-minute continuous series per market (Jan 2024–Jul 2026) with the right session: CME 09:30–16:00 ET for NQ and ES, COMEX 08:20–13:30 ET for gold, US cash overlap 09:30–16:00 for CME Bitcoin (Globex is nearly 24h — that overlap is liquidity, not an auction). Walk-forward uses six-month in-sample / three-month out-of-sample folds. Mini and micro share a tape; $ PnL and Pine qty follow the contract you pick. It is not a CME tick archive.
Real tape · Yahoo NQ=F / ES=F / GC=F / BTC=F
After discovery, every candidate is replayed on its own root tape. 1h / 4h / 6h cover ~2 years. Daily and 2D cover ~10 years. 5m / 15m / 30m are Yahoo’s last 60 days. Gold uses COMEX hours, not the NYSE open. Bitcoin core is US cash overlap on a nearly 24h Globex tape.
Sign in to refresh the Yahoo tape. Research cycles fetch it automatically.
NQ · MNQ · Yahoo NQ=F
Highest vol. Opening drive, ETH handoff. Mini for history, MNQ to size.
ES · MES · Yahoo ES=F
Same 09:30–16:00 ET session. More mean-reversion literature. Use MES to size.
GC · MGC · Yahoo GC=F
COMEX core 08:20–13:30 ET — not the NYSE open. Asia and London matter.
BTC · MBT · Yahoo BTC=F
CME futures, not a 24/7 perp. US 09:30–16:00 is liquidity overlap, not a cash open. Use MBT to size.
What is modeled
Sunday 18:00 ET through Friday 17:00 ET, US holiday calendar, early closes, opening gaps, lunch contraction, power-hour expansion, and GARCH volatility clustering. RTH is CME 09:30–16:00 ET on NQ/ES, COMEX 08:20–13:30 ET on gold, and US cash overlap on CME Bitcoin. Higher timeframes are clock-aligned ET resamples of the 5-minute tape (calendar epoch-day buckets for 1D and 2D). A bar is flagged RTH if any constituent 5-minute bar is RTH. Execution matches TradingView’s OHLC broker emulator (Bar Magnifier off): signal on close, fill next open + 1 tick, stops fill at the stop or at the open if the bar gaps through then +1 tick, targets fill at the limit (or the open on a gap through) with no slip, same-bar stop-before-target. Commission and qty follow the contract (NQ $4.20/side, MNQ $0.62, same split on ES/MES, GC $2.50 / MGC $0.70, BTC $5 / MBT $1.20). Intraday flatten is a market order on the next bar after 15:55 ET on equity index and Bitcoin, 13:25 ET on gold; daily and 2D do not flatten.
TradingView transfer
Generated Pine is the TradingView engine — there is no second backtester to install. Stay on TradingView. Use the continuous on the script header (NQ1! / MNQ1!, ES1! / MES1!, GC1! / MGC1!, or BTC1! / MBT1!), matching timeframe, order size = contracts. Margin is 10% in the script so 1 mini opens inside $100k; do not switch leverage by hand. Brackets are tick distances from the fill (strategy.exit loss/profit), not frozen prices from the signal close. Flatten is a market close_all after the contract’s flatten time on intraday charts. Daily and 2D Pine skip flatten. Leave Bar Magnifier off.
TradingView PFs used to print systematically worse than Meridian. That was not “tick vs bar” — both testers are OHLC. The leak was fill rules: Meridian only slipped the entry and filled gap-through stops at the stop; Pine also placed absolute stop/limit from the signal close, so a gapped fill could close a winner as a loser. Those are aligned now. There is no multiply-by-X formula. What remains is data: TV continuous back-adjust and TV’s bar builder are not Yahoo NQ=F / ES=F / GC=F / BTC=F and not the synthetic GARCH path. A Real tape badge is permission to take the script to the matching 1! symbol, not a forecast of the TV equity curve.
Literature
The desk harvests arXiv q-fin, SSRN, NBER, Journal of Finance, JFE, RFS, JFQA, and central-bank notes for NQ, ES, COMEX gold, and CME Bitcoin, plus X / Substack from a curated quant roster. A paper becomes a biased family and a walk-forward on the market you pick — never a validated edge.
Promotion gates
Profit factor, expectancy, drawdown, degradation, and fold-pass are the same on every chart. Trade-count floors scale with bar size so a daily strategy is not punished for having fewer prints than 5m. 4h/6h/1D/2D use a lower trade floor (8 / 8 / 6 / 6 OOS trades).
- OOS PF ≥ 1.12 · expectancy positive · DD ≤ 28%
- IS→OOS degradation ≤ 60% · fold pass ≥ 50%
- 5m: OOS trades ≥ 18, ≤ 6 trades/day
- 15m: OOS trades ≥ 10, ≤ 3 trades/day
- 30m: OOS trades ≥ 10, ≤ 1.5 trades/day
- 1h: OOS trades ≥ 10, ≤ 1.2 trades/day
- 4h: OOS trades ≥ 8, ≤ 2 trades/day
- 6h: OOS trades ≥ 8, ≤ 2 trades/day
- 1D: OOS trades ≥ 6, ≤ 1.05 trades/day
- 2D: OOS trades ≥ 6, ≤ 0.6 trades/day