Send a screenshot, fill photo, or CSV from a trade that is already closed. RiskDrive reviews the historical decision path, evidence confidence, limitations, and your submitted rule context.
Below is a sample replay. It is not a trading signal, forecast, or live app endpoint.
Red and green belong to the market, and the market is closed. A review engine doesn't ask what you made. It asks what was knowable at each moment you acted — and what you did with it.
The pair had extended +1.8% — inside your own playbook's no-chase zone. At this moment the engine estimates low evidence confidence for the rule context against the risk you took. The ink is light here: a drift from the plan, not yet a clear breach.
Maximum adverse excursion during the gap: −2.1R against an unprotected position. Nothing on the P&L records this — the trade still ended green. The review records it in the deepest ink it has.
Price rolled over. You closed manually for +24.5 pips — a reasonable exit, honestly taken. The only decision of the three that earns light. One good move doesn't acquit a sequence.
Repeat these three decisions often enough and the rule breach becomes expensive. That's what makes a lucky win dangerous: it pays you to repeat it. Your platform just rewarded the behavior. This page is the only place that didn't.
GBPUSD short, stopped out on a surprise CPI print. The stop executed as placed — not pulled, not widened. That loss was a system cost, and the review clears it. Correct losses should stay visible as process costs.
No broker login · screenshot, photo, or CSV · no signals, review only
Judgment must never be confusable with price. So correct decisions earn light — candlelight in a closed room — and flawed ones are marked in ink, deeper with severity. Neither can be mistaken for a candle on the chart.
Independent of how the trade ended.
Even when the trade made money.
Photo of a fill confirmation, platform screenshot, or CSV. Vision models extract instrument, direction, size, prices, exact timestamps. No broker integration means no broker we don't support.
Around each timestamp: realized volatility, the price path, and scheduled economic events — the CPI print your journal never knew about. Decisions are judged against the market they were made in.
Entry, sizing, every stop move, every exit — with historical evidence labels at that timestamp: submitted stop distance, target progress, rule exposure, evidence confidence, limitations, and a historical rulebook comparison.
Process and outcome agree. The only quadrant that compounds.
Correct play, losing result. Counted, never punished.
Broken rules, paid anyway. The review marks the repeated rule breach.
Avoidable. Attribution shows how much loss came from violations vs original risk.
TradeZella, TraderSync and Edgewonk are good products — broker auto-sync and pattern analytics across hundreds of trades. RiskDrive answers a different question: what did the submitted historical record support at that decision point?
| AI trading journals TradeZella · TraderSync · Edgewonk | RiskDrive AI | |
|---|---|---|
| Unit of analysis | The trade — patterns across 50–100 trades | The decision — scored separately, from trade one |
| Market context | Your trades and tags | Volatility, price path & economic events rebuilt at your timestamps |
| Review checks | Historical summary stats | Submitted stop distance, target progress, rule exposure, evidence confidence, and limitations |
| Outcome bias | Inherited — winners look like skill | Decoupled by design — lucky wins flagged, correct losses cleared |
| Broker auto-sync | ✓ Up to 500+ brokers | None by design — screenshot / photo / CSV, zero credentials held |
| Agent access | Human dashboards only | ✓ founder-reviewed review packet |
Category comparison is abstract and RiskDrive-owned; it is not a live competitor-status claim.
RiskDrive can package a closed-trade review as structured JSON for internal and founder-reviewed workflows. This release does not expose a live MCP endpoint, does not settle payments, and does not let any agent touch broker accounts.
# local review packet, founder-reviewed pilot tool = packet.load("closed_trade_review") resp = await tool.call( trade = closed_trades[-1], mode = "manual_review_only" ) resp.conformance # "1 of 3 decisions followed the plan" resp.verdict # "LUCKY_WIN" resp.moves[1].label # "rule breach: stop missing 40 min" resp.boundary # "historical_review_only"
The current release is a founder-reviewed pilot, not a mature self-serve SaaS checkout. The public offer is narrow on purpose: one closed-trade verdict first, then decide whether a larger review packet is worth paying for.
No. Upload a screenshot, a photo of a fill confirmation, or a CSV of closed trades. RiskDrive never holds credentials and never places orders — which also means it works with every broker, including the MT4/MT5 brokers journal apps don't sync with.
Those are journals: they sync trades and find patterns across your history. RiskDrive is a review engine: each individual decision is scored against historical evidence labels at that timestamp, with the market and economic events reconstructed around your timestamps — and the verdict is decoupled from the outcome. Many traders run both.
Because a profitable violation can hide a repeated rule breach. P&L can hide it; the review records it as historical process evidence.
No. Historical review only. It evaluates decisions already made and never recommends a new trade. Review is governance, not advice.
Yes. Founder-reviewed workflows can export structured review JSON with sample size, confidence, limitations, and boundary fields. No autonomous payment, broker access, or order execution is enabled.
Early access opens in cohorts. First review packet is manually reviewed — screenshot, photo, or CSV.