How Pinbar AI Detects Revenge Trading & FOMO
Most trading-risk tools rely on crude rule-based thresholds — "more than 3 consecutive losses triggers a warning," or "any trade above $50K gets flagged." These rules produce two outcomes: constant false positives that train operators to ignore alerts, and genuine risk behaviour that slips through because it doesn't match a hardcoded pattern.
Pinbar AI takes a different approach. We look at the shape of trading behaviour over time — the relationships between consecutive trades, position sizing trends, hold-time decay, and time-of-day clustering. Each trader's baseline is their own recent history, not a global average. This means a scalper taking 60 trades a day and a swing trader taking 3 trades a week are evaluated against their own norms.
This page walks through the exact signals we use, so Brokers and Prop Firms can evaluate the methodology before integrating. No buzzwords — just the data pipeline, the trigger logic, and the scoring model.
The Raw Signals We Pull From Trade Data
Every analysis starts with per-trade data points ingested from the broker connector or platform sync: entry timestamp, exit timestamp, instrument, side (long/short), entry price, exit price, position size (contracts, lots, or notional value), realised P&L, and — when available — account equity at entry.
From these raw fields we derive a set of computed signals for each trade:
- Hold duration (exit − entry)
- P&L per minute of hold time
- Position size as % of account equity
- Time-of-day bucket (session open, mid-session, close)
- Time-since-previous-exit (gap between trades)
- Rolling win/loss streak (last N trades)
No external data is required — no market-data feeds, no news APIs, no social sentiment. Everything is derived from the trader's own execution history, which means the system works with any broker, any instrument, and any market globally.
How We Detect Revenge Trading
Revenge trading is re-entering the market quickly after a loss with elevated position size and reduced patience. The trader is trying to "win it back" — and the trade structure is measurably different from their normal entries.
The Three-Signal Trigger
We flag a candidate revenge trade when all three of these conditions fire on the same trade:
- 1Time-since-previous-exit is below the trader's own 25th-percentile hold gap. This isn't a hardcoded "5 minutes" threshold — it's calibrated from the trader's last 100 trades.
- 2Position size is at least 1.4× the trader's rolling 20-trade median size. The trader is sizing up.
- 3The previous trade was a loss, or 2 of the previous 3 trades were losses. The emotional trigger is present.
Why Per-Trader Calibration Matters
A hardcoded rule like "trade within 5 minutes of a loss = revenge" produces enormous false positives for scalpers who naturally re-enter within seconds, and completely misses the same behaviour in swing traders who normally wait hours between entries. By calibrating against each trader's own rolling percentiles, the system adapts automatically. A scalper's 25th-percentile gap might be 45 seconds; a swing trader's might be 3 hours. Both get meaningful alerts — without manual configuration by the broker.
Confidence Scoring
The revenge-trade flag isn't binary. We score each candidate trade from 0 to 100 based on how extreme each of the three signals is relative to the trader's baseline. A position size at 2.5× the median scores higher than 1.4×. A hold gap in the 5th percentile scores higher than the 20th percentile. Brokers and Prop Firms can configure the threshold at which a flag becomes a visible alert — some firms want to see every candidate above 40; others only care about scores above 75.
How We Detect FOMO Entries
FOMO — fear of missing out — manifests as chasing a move after it has already extended. The trader enters late in a candle's range, usually with elevated size, at a price that's objectively unfavourable relative to the move's origin.
The FOMO Signature
The FOMO signal looks for a convergence of four conditions:
- 1Entry price is in the top 20% of the candle's range for long entries (or bottom 20% for shorts). The trader entered late in the move — not early, not mid-range.
- 2Position size is at least 1.3× the trader's rolling median. FOMO traders tend to size up because they feel the urgency of the move.
- 3Time-of-day clustering: FOMO entries skew heavily toward the first 30 minutes after a major session open or news release. We tag entries in these windows.
- 4No prior position in this instrument that day. It's a fresh chase, not a planned scale-in or add to a winning position.
Why This Is Hard With Rule-Based Systems
Simple threshold rules miss FOMO because they don't have the price-within-bar context. Knowing that a trader entered at $152.80 is meaningless without knowing the 1-minute candle ranged from $151.20 to $153.00. We compute the entry's percentile position within the execution-time candle range, which gives us the late-entry signal cheaply and without requiring tick-by-tick market data from the broker.
Drawdown Breach Detection
We track peak-to-trough equity over rolling windows — intraday, daily, and weekly — against each trader's configured maximum drawdown limit. This is especially critical for Prop Firms with hard daily loss limits where breaching the threshold means account termination.
Alerts fire at three escalation levels: when drawdown crosses 50%, 75%, and 100% of the configured limit. The point is to alert before the limit is hit, not after — giving the trader (and the firm's risk desk) time to intervene. For firms using hard cut-offs, the 75% alert is typically the one that triggers a human review.
Over-Leverage Detection
We compute notional position size as a percentage of account equity — both per individual trade and across all open positions simultaneously. Two distinct flags are evaluated:
Single-Position Over-Leverage
Any individual trade where position size exceeds the trader's configured maximum as a percentage of equity. A trader risking 15% of equity on a single position when the limit is 5% gets flagged immediately.
Aggregate Over-Leverage
Total open notional across all positions exceeds a configured multiple of equity. This catches the scenario where each individual trade is within limits, but the trader has 12 positions open simultaneously.
For multi-account Prop Firms, over-leverage detection rolls up across all of a trader's accounts — preventing the common workaround of splitting oversized positions across sub-accounts.
Why This Approach Beats Rule-Based Systems
Rule-based systems apply the same thresholds to every trader. A "max 3 losses before alert" rule fires constantly for active day-traders and never for swing traders who rarely string 3 losses together. The result is alert fatigue — operators stop reading the warnings, and the system becomes decoration.
Per-trader calibration means the system learns each trader's normal baseline and only flags statistical deviations from that baseline. This is the same principle behind anomaly detection in fraud monitoring, infrastructure alerting, and credit risk scoring — it works because the reference point is personal, not global. A 1.5× position size increase matters more for a consistent trader than for one whose sizing is naturally volatile.
What Brokers and Prop Firms Get
Real-Time Alerts
Pushed via webhook, Slack, or email the moment a flag crosses your configured threshold.
Flagged Trader Dashboard
A ranked view of flagged traders by severity score, with drill-down into individual signal values.
Full Audit Log
Every flag is logged with the underlying signal values, timestamp, and confidence score — ready for compliance review.
Configurable Thresholds
Set thresholds per trader, per cohort, or firm-wide. Adjust sensitivity without engineering work.
Automated Risk Actions
Escalate flags into automated responses via API: notifications, position size limits, or trading halts.
Cohort Analytics
Compare behavioral patterns across trader cohorts — by experience level, strategy type, or account size.
Want to see this running on your own trade data?
Schedule a partnership call and we'll walk through the methodology with your specific platform and trader profiles.