Seven live reads from your own record โ two more on the roadmap
Sample: 270 trades, ~$50K account, 8 months. Each demo runs on this log and shows what RGI measures โ in P&L, win rates, equity curves, and setup names you recognize. Post-trade, observation-only. It reads your completed trades. You decide. Open-source, post-trade, runs on your machine.
Snapshot ยท Minute Three
"Point it at 8 months of fills โ here's what resolved"
Load a sample log. Watch the dashboard fill โ signal trust, bias flags, regime band, certified claims. Three minutes.
You have 5 setups. Your journal says you're best at RSI+MACD โ you trade it 3ร more than anything. The engine reads your verified outcomes and ranks them by what actually makes money. The ranking inverts.
The blow-up is the story you tell. The damage is the drip โ revenge re-entries, post-streak size creep, Friday afternoon weakness. Each invisible in monthly P&L. Each priced by the engine.
• Equity curve with bias-caused drawdowns labelled
• What-if โ toggle each bias off, watch the curve reshape upward
• All fixed: +$16,600/yr (modeled). $16–26K/yr in aggregate.
• That gap is the value of knowing.
Demo 3 ยท Volatility ยท Short-Vol Cliff ยท The Shape Every Vol-ETF Trader Knows
"The $17K Cliff Your Backtest Didn't See"
Context โ the short-vol equity curve
Steady small gains. Rare deep losses. You're making a directional bet on volatility itself โ long SVXY, short VXX, positions that profit when vol stays low. The equity curve climbs $500/week for 4 months. Then VIX spikes: −$18,000 in one day. Four months of work, gone. The health band fires AMBER two weeks before the cliff. That's the signal no backtest gives you โ and 60–70% of funded-trader blow-ups end here (estimated from prop-firm drawdown data).
Run: two equity curves โ same premium book, with vs without health band
What RGI shows that a backtest can't
• Health band fires AMBER before the cliff โ that's the signal
• Without it: 4 months of $500/week, then −$18K in a day. Net: −$10K
• With it: AMBER at week 14 → step back → spike hits at −$2K, not −$18K → resume when vol is rich
The trade-off: The health band fires conservatively. In this sample, it fired AMBER on 3 weeks that would have been profitable โ missed gains of ~$1,500 (modeled). The cliff it caught: −$18,000. Net: +$16,500. The gate would rather cost you a small win than let you hold through a regime change.
Investor note: 60–70% of funded-trader blow-ups are drawdown breaches from short vol (estimated from prop-firm drawdown data). This is the shape that kills accounts — and the flag that fires before it.
Demo 4 ยท Rolling Profit Factor ยท Strategy Health Check
"Can This Strategy Take More Size?"
Context
Three strategies in your book. A journal says they're all running. The rolling profit factor โ a chart every serious trader reads โ tells a different story. One is dying. One isn't proven. One is safe to scale.
Run: rolling 20-trade profit factor for each strategy
Small-cap momentumโ
New momentumโ
Volume-breakoutโ
Under the hood: The conservation gate checks whether a strategy has enough verified evidence to size up. Strategy A: 68% accuracy over 90 decisions, but rolling quality has decayed below its threshold → AMBER (don't add size). Strategy B: only 50 trades → insufficient evidence for a verdict. Strategy C: 82% accuracy, rolling quality well above threshold → GREEN (safe to scale). Math: ฮธ_min = 23.53/(ฮฑ×V). A: 0.31 < 0.384. C: 0.52 > 0.319.
What RGI shows that a backtest can't
Rolling profit factor from YOUR execution, not a simulation. A backtest says "this strategy returns 15%/yr." Your profit factor line shows it crossing below 1.0 six weeks ago. The gate names the cause: "regime shifted trending โ ranging." Caught at the floor, not at month-end.
Demo 5 ยท Time ร Regime Heatmaps ยท Where Your Edge Lives
"The Edge You Only Have in a Storm"
Context
Two heatmaps you'd recognize from any journal app โ time-of-day ร day-of-week, and regime ร strategy. But these are colored by P&L, not just win rate. And the regime heatmap is something no journal can build.
Run: fill the heatmaps from 500 trades
P&L by Time of Day ร Day of Week
P&L by VIX Regime ร Strategy
Your journal shows 3 strategies. Your risk shows 1 bet.
Your 3 short-vol positions are 1 bet. SVXY, UVXY, VXX: ρ = 0.89 correlated. When VIX moves, they all move together — −$11,100 in a single spike (modeled).
What RGI shows that your journal can't
• Regime-conditioned P&L โ a journal shows Friday PM is bad. Can't show WHY.
• Your edge concentrates in VIX 25–35. Only trade VIX 20+: +$23,000 (modeled)
• Actual: +$21,600. Giving back $1,400/yr in calm markets where you have no edge.
Demo 6 ยท R-Multiple Histogram + Claim Gate ยท Which Insights Are Real
"The Insight That Didn't Survive Correction"
Context
Your R-multiple distribution โ the chart every prop trader reads โ shows your risk discipline. But your journal flags 12 "insights" from it. The claim gate tests each one. 8 don't survive correction.
Run: R-multiple distribution + insight gate
What RGI shows that your journal can't
• R-multiple distribution โ shows risk discipline shape
• Claim gate tests each insight: "earnings plays" → 23% chance it's noise. "Volume breakout in trending" → 71%, CERTIFIED.
• Gray badge ("insufficient data, N=4") โ worth more than a dozen greens from a journal
• It would rather say nothing than sell you noise.
Demo 7 ยท Judgment Memory ยท The Compounding Moat
"One Trade, Three Improvements"
Context
Fork the code tonight โ it's Apache 2.0. But watch what happens over 6 months. Your enriched record pulls away from a cold start. The gap widens every month. That's the moat.
Run: two equity curves, same engine, different starting points
AlpacaIBKRCSVโ one graph
Judgment Memory โ the moat
• Code is Apache 2.0 โ fork it tonight. But a fork starts cold.
• Judgment Memory: ~200 learned values from every trade โ not in the code. In the graph.
• Switch brokers? The graph carries. Step away for a quarter? It remembers.
• That's the compounding a static tool can't match.
v1.1 · Roadmap · For volatility traders and investors
If your strategy is a bet on market volatility — selling premium, buying protection, timing VIX — this is the timing layer you don't have. Same strategy, split three ways: by IV/RV regime, by DTE, and by VRP state. The difference isn't the strategy — it's when you enter.
IV/RV < 1.2 (thin premium)
IV/RV > 1.5 (rich premium)
Same strategy. $600/trade difference (modeled). Your edge peaks at 7DTE when IV/RV > 1.5. The 0DTE volume is noise. GREEN only when premium is rich. v1.1 — on the roadmap.
The 0DTE trap: You already know 0DTE is noise — it's > 75% of retail SPX volume and retail lost ~$125M there. The VRP term structure shows where the real premium edge is: 7DTE, IV/RV > 1.5. That's a $600/trade difference just from timing. Most traders enter by feel. This is the math.
Demo 9 ยท Desk View ยท v1.1 Roadmap ยท For Teams & Shared Books
"Which of Your Traders Is Running a Decayed Strategy Right Now?"
v1.1 ยท For teams, desks, or anyone who shares signals
Five traders on a shared book โ or five strategies you share with a buddy. Per-strategy conservation status. One person leaves โ their Judgment Memory โ the evidence trail โ transfers to the next seat. The governance gap that Demo 4 shows for one trader, applied across the whole book.
For desk pilots
Which trader is running a decayed strategy right now? Judgment Memory in action: a departed trader's ~200 learned values transfer to the new seat โ 8,000 trades of judgment, transferred in minutes. Portfolio-collapse signal: the $30โ80K tail, across the whole book. $120โ200K retune cost (modeled) โ $0 with evidence trail.
Recap
Seven live reads. All from your own record. Two on the roadmap.
48%โ71%
favorite vs best setup โ
$16โ26K
leakage surfaced /yr โ
$17,200
the cliff, one flag โ
+0.18R
expectancy (R-multiple) โ
+$26K
enriched vs cold start (6mo) โ
Charts you already read โ equity curves, P&L calendars, R-multiples, profit factor lines, heatmaps. What's new: the layer on top that shows which of your signals have verified edge, where your drawdowns come from, and when your strategy left its competence zone. Post-trade, observation-only. The material earns the next conversation.
Open-source ยท post-trade ยท runs on your machine ยท Your record, your data, your machine. RGI never sees your fills.
RGI Trading ยท Demo v23 ยท October 2026 ยท Post-trade, observation-only (no trade directives) ยท Figures modeled / illustrative from sample ยท No production customers ยท Apache 2.0
