Passive · Local · Validated

See the decision,
not just the trade.

Kynexis reads how a professional actually behaves and surfaces the behavioral risk before it compounds. It runs on the trader's own machine, and nothing sensitive ever leaves it.

META$492.70▼ -5.1%
NASDAQ · 1H
KYNEXIS · MONITORING
Kynexisnow
Loss aversion detected
4 closed positions held past their stop-loss thesis. Losers held 24 days on average versus 2.6 for winners.
BRS 52 · ELEVATEDView report →

Passive by design

No forms, no journaling, no change to the workflow. It reads the screen, detects the trade, and scores the behavior on its own.

Private by construction

Screens, trades, positions and profit and loss stay on the device. In a firm, only a sealed alert envelope crosses a machine.

Validated, not asserted

Eighteen signals, each defined from the academic literature and tested blind against 490 synthetic traders on 30 years of real market data.

How it works

From a trade on screen to an insight, without the data leaving the desk.

A walkthrough of the passive pipeline, using the readings from a real recorded session.
01 A trade is placed, on any platform
TradingView · paper
BX 130.00 +4.42%
SideBUY
SymbolNYSE:BX
TypeMarket
Qty1
ORDER FILLED · 130.00

Kynexis asks nothing of the trader. It watches the screen the way a risk officer would look over a shoulder, only without seeing anything leave the room.

02 The screen is read, on-device
> capturing frame 1280x800
> ocr: NYSE:BX   MARKET   1   FILLED
> resolve ticker: exchange-prefix -> BX
> side: BUY   platform: TradingView
BXBUYTradingView0.82

A local vision service reads the order off the pixels. The ticker resolver prefers the exchange-tagged symbol, so a busy screen full of menu words never gets mistaken for a trade.

03 Behavior is scored against the history
Overconfidence
100
Loss aversion
100
Revenge trading
94
Emotional sizing
80
Attribution error
77
Overtrading
63

Each signal is a construct from the literature, disposition, calibration, extrapolation, scored on the trader's own closed history. These are the readings from the recorded session.

04 An insight surfaces, with its evidence
OvertradingHIGH
Your recent trading pace is 2.9x your usual daily rate.
Overtrading is a sustained rise in trade frequency above your own baseline, often a sign of impulsive or emotionally driven activity.
0.93
Baseline
2.71
Recent
Trades per day, recent against baseline

The alert is written in plain language and always carries the numbers behind it. The trader can open the full report and respond to it, all on the device.

05 In a firm, only a footprint crosses the line
On the trader's machine
• Ticker, thesis, position
• Profit and loss, conviction
• Every score and its evidence
SEALED
ENVELOPE
What the supervisor sees
• Pattern: overtrading
• Severity: high
× No ticker, thesis, or figure

The envelope has a fixed shape. By construction it cannot carry a ticker, a position, or a number, and a supervisor only ever sees their own branch of the tree. The boundary is enforced in code and proven by the test suite.

Auto-playing
The engine

Eighteen signals, each grounded in the literature, none merely asserted.

Validated non-circularly: the ground truth is the academic construct, not the detector rule.
Loss aversion
Losers held far longer than winners, on the trader's own hold times
Strong
Revenge trading
Rapid re-entry and upsizing in the minutes after a realized loss
Strong
Overconfidence
The gap between stated conviction and how often the top calls land
Strong
Attribution error
Losses blamed on external factors in the post-trade review
Strong
Emotional sizing
Position size rising after a loss rather than with edge
Strong
Trend following
Trading with the recent tape, the footprint of extrapolation, honestly labeled
Footprint
Consensus clustering
A desk converging on the same names in the same window
Footprint
Stop discipline
Harder to separate from luck on the blind test, so kept and down-weighted
Low confidence
Strong: clean separation on the blind backtest Footprint: a real pattern, not proof of intent Low confidence: kept, scored low, never oversold
18+2
Individual signals, plus two desk-level aggregate detectors
490
Synthetic traders carrying literature-injected biases
30yr
Of real market data across regimes, as the test bed
0
Circularity: detectors run blind against the injected truth
The moat

Nothing sensitive leaves the machine.

A supervisor never receives trader data. They receive a sealed envelope with a fixed shape that, by construction, cannot carry a ticker, a position, or a number.

Captured and scored locally
• Screens and the text read from them
• Orders, fills, positions, profit and loss
• Stated conviction and written thesis
• Every behavioral score and its evidence
SEALED
ENVELOPE
Crosses in a firm, only this
• A pattern and a severity
• Scoped to one branch of the tree
× Never a ticker, position, or figure

Enforced in the code and proven by the correctness suite. That is the difference between a claim a bank can accept and one it cannot.

The asset

A clean, finished asset. A transfer, not a project.

Finished, documented, licence-clean. No team, no infrastructure, no customer data to inherit.
01

Full source

Behavioral engine, local vision service, the desktop app, and the firm and desk web view.

02

Packaged installers

Reproducible builds for macOS, Windows and Linux, signing-ready for the acquirer's own identity.

03

Validation harness

The synthetic generator and blind backtest that produced the scorecard, runnable from zero.

04

Complete data room

Architecture, security white paper, signal reference, deployment runbook, and the validation report.

05

Licence and IP file

Bill of materials, a clean licence audit with no strong copyleft, and chain of title prepared.

06

Privacy architecture

The envelope boundary and scoped hierarchy, enforced in code and proven by the suites.

For acquirers

Acquirable without entanglement.

The hardest thing to build and the hardest thing to prove, a privacy architecture a bank will accept, is already done. Sign it and ship it under your own identity.

Best fit
Trade surveillance, behavioral analytics, and RegTech acquirers
Structure
Clean asset purchase, optional short transition
Enquire