Decision supervision for discretionary desks

Every control fires at order entry or after the fact. The decision happens before both.

Trading plans evaluated against the firm's actual mandate before capital moves. A behavioral engine that grades its own predictions against fills. A sealed evidence chain on every decision. Deployed single tenant inside the firm's own environment, with no data egress of any kind.

THE SYSTEM, ENTIRE
Mandate evaluated before capital moves
Falsifiable forecasts, graded against fills
Sealed evidence chain on every decision
Your Azure tenant · zero data egress
Firm sees process · trader keeps the dialogue
Observes everything · blocks nothing
0DATA EGRESS · NO EXTERNAL CALLS
2PLANES · OFFICER AND TRADER
94BPS PER YEAR · THE DOCUMENTED COST OF UNMANAGED BIAS¹
90DAY PILOT · ONE DESK · FIRM TENANT
DEPLOYS ON Microsoft Azure · Entra ID
¹ Essentia Analytics client research, 2020. Full sourcing in Exhibit A below.
FIRM TENANT · MICROSOFT AZUREPRIVATE NETWORK · FIRM KEYS · ENTRA ID
TRADER PLANETHE COACHING DIALOGUE
FIRM PLANEOBSERVATION REGISTER · RISK & COMPLIANCE
EVERYTHING ABOVE RUNS INSIDE THIS BOUNDARY. MODELS, DATA, RECORDS. ALL OF IT.
The dialogue on the left is never summarized, scored, or forwarded. Committed plans, mandate evaluations, and observations cross to the officer register. Nothing else does, and nothing at all crosses the tenant boundary.
I.

The supervisory record stops where the decision begins.

Pre trade compliance evaluates coded limits at the moment an order exists. Surveillance examines executions and communications after the fact. Retrospective analytics reconstruct decision quality from holdings data, months later. Each is necessary. Each is silent about the same thing: the plan, the reasoning, and the sizing logic that formed before any order reached the OMS. Three exhibits from the research behind this product.

FROM THE RESEARCH FILEANNUAL COST OF UNOBSERVED BEHAVIOR
94BPSPER YEAR. The average excess return Essentia Analytics reports uncovering for its clients, previously lost to behavioral biases, across roughly $150B of assets under daily analysis.
$500M book
$4.7M / yr
$1B book
$9.4M / yr
$5B book
$47M / yr

Dollar figures are the 94 bps applied to book size, indicative only; the precise number differs by desk. The structural point does not: these biases operate at the intent stage, precisely the stage the current control stack cannot see. Every control the firm runs today fires after this cost is already incurred.

SOURCE · Essentia Analytics client research, stated publicly by CEO Clare Flynn Levy (Markets Media interview and company publications, 2020). Essentia's decision analysis methodology is published in the peer reviewed Journal of Investing.

A / C
II.

Position in the control stack.

DojiPad Institutional replaces nothing and competes with nothing. Every incumbent control fires somewhere on the decision timeline. We took the one position on it that was empty, and it happens to be the position where the decision is actually made.

INTENTDOJIPAD INSTITUTIONALORDER ENTRYPRE TRADE COMPLIANCEEXECUTIONSURVEILLANCE, AFTERREVIEWRETROSPECTIVE ANALYTICSTHE ONLY CONTROL BEFORE THE ORDER EXISTS
THE DECISION TIMELINE · READ DOWNWARD
INTENT
DOJIPAD INSTITUTIONAL
The plan can still change. Evidence and influence coexist here, and nowhere else.
THE ONLY CONTROL BEFORE THE ORDER EXISTS
ORDER ENTRY
PRE TRADE COMPLIANCE
The order already exists. The thinking is done.
EXECUTION
SURVEILLANCE, AFTER
The money has already moved.
REVIEW
RETROSPECTIVE ANALYTICS
The quarter is over.

Read the timeline left to righttop to bottom and the pattern is plain: the further rightfurther down a control sits, the more it can prove and the less it can change. By order entry the thinking is done. By execution the money has moved. By review the quarter is over. The intent stage is the only point where evidence and influence coexist, and until now nothing has operated there.

III.

The nearest neighbor.

One firm operates closest to this space, and deserves to be named rather than gestured at.

THE NEAREST NEIGHBORSIX DIMENSIONS
DOJIPAD INSTITUTIONAL
ESSENTIA ANALYTICS
Where it acts
The intent stage, while the position is still a plan
After the fact. Decision patterns reconstructed from holdings and trade history
What it needs
Committed plans, inside the firm's own tenant
Portfolio and trading data shared with a vendor hosted platform
Forecasts
Falsifiable predictions issued before a session, graded against actual fills
Nudges and insights derived from historical pattern analysis
When behavior changes
In session, while the decision is open
In review, after the cost is incurred
Privacy model
Two planes. The coaching dialogue never reaches the firm
Analytics visible to the firm by design
Mandate awareness
Every plan evaluated against the firm's actual limits at planning time
Not a mandate control

Essentia established the category: behavioral analytics for professional investors, and the published evidence that decision level behavior carries a measurable cost. We consider that work foundational. Characterizations of Essentia are based on publicly available product descriptions as of mid 2026. Retail journals and coaching apps are not compared here; none operate inside a firm tenant, carry a mandate register, or produce supervisory records.

THE READ

The difference is architectural, not incremental. Essentia analyzes decisions after they are made, from data that leaves the firm. DojiPad Institutional supervises decisions while they are being made, from inside the firm's own walls. One is a mirror held up quarterly. The other is an instrument in the cockpit. A desk can reasonably run both; only one of them exists at the moment the plan can still change.

IV.

Architecture: what moves, and what never moves.

One diagram carries the whole design. Market data and fills flow in. The dialogue circulates between trader and coach and never leaves that plane. Committed plans cross one boundary, to the officer register. Orders travel the firm's own execution path, which runs around this system entirely. And nothing, of any kind, leaves the tenant.

FIRM TENANT · MICROSOFT AZUREMARKETDATATRADERAT THE DESKDOJIPAD INSTITUTIONALTRADER PLANEDIALOGUE · PRIVATEFIRM PLANEPLANS · EVALUATIONS · CHAINRISK &COMPLIANCEOMS / BROKERTHE FIRM'S OWN EXECUTIONIN ONLYDIALOGUECOMMITOBSERVATIONS · RECORDSORDERSFILLS, FOR GRADINGDOJIPAD HOLDS NO ORDER ROUTING OR BLOCKING CAPABILITY. THE EXECUTION PATH RUNS AROUND IT.×EGRESSNONE
Never leaves the trader plane Firm records, in tenant Inbound data only No route out of the tenant
01 / 06MARKET DATA FLOWS IN. NOTHING FLOWS BACK OUT.
SINGLE TENANT

The entire system deploys inside the firm's own cloud subscription, under the firm's identity provider and the firm's keys. There are no external calls, no telemetry, and no vendor infrastructure the firm's data could reach. The vendor cannot see the firm's trades. That is a property of the architecture, not a contractual promise.

TWO PLANES

The firm plane carries mandate evaluations, observations, and evidence chains, and is built for risk and compliance officers. The trader plane carries the coaching dialogue itself and is structurally separated: never summarized, scored, or flagged across. Coaching runs on candor, and candor requires a boundary that cannot be reconfigured away.

BEHAVIORAL ENGINE

Each trader's specific failure patterns are modeled from their own history. The engine issues falsifiable predictions before a session and grades itself against actual fills. Every claim the system makes about a trader is scored against execution data, and the scoreboard is part of the product.

DEPLOYMENT
Single tenant, firm cloud subscription · Microsoft Azure, firm region, firm keys
IDENTITY
Microsoft Entra ID · firm directory, role separation between officer and trader surfaces
DATA EGRESS
None · no external calls, no telemetry, no vendor endpoint
EXECUTION PATH
Not in it · the system holds no order routing or blocking capability
RECORDS
Append only evidence chains · plan, evaluation, observation, outcome
GOVERNANCE
Regulatory alignment memo · architecture review with the firm's security function, under NDA
V.

Doctrine: the asset is judgment.

A discretionary desk has exactly one asset: the judgment of its traders. Every control the firm buys should be measured against one question. Does it make that judgment better, or does it merely constrain it?

We build instruments, not autopilots

A wall produces compliance at the moment of the wall and nothing afterward. An observation, surfaced while the decision is still open, with the trader's own record and the mandate's own words attached, produces something a wall never can: a trader who catches the pattern themselves next time. One of these compounds across a career. The other has to fire again tomorrow. We chose the control that compounds, and the system's design follows from that choice completely.

The pilot flies the plane

Aviation did not become the safest form of travel by automating pilots out of the cockpit. It did it with instrumentation, checklists, and a culture where every decision leaves a record and every record teaches. That is the model here. The trader decides. The firm supervises. The system makes both happen with full information at the intent stage, and seals the evidence that they did. Judgment stays where the returns come from, and so does the accountability.

VI.

Built against the regulatory horizon.

We do not ask desks to take a young vendor on faith, and we do not hand out our technology to prove ourselves. What we show instead is where the rules are going and what the system already implements for each. The full mapping is available as a memo, and the architecture is reviewed live with the firm's security function under NDA.

EU AI ACTRegulation (EU) 2024/1689 · high risk obligations phasing in from August 2026
What the framework asks: risk management, logging, transparency to deployers, human oversight, and record keeping for AI systems that touch consequential decisions. What the system implements: append only logs of every model interaction that produces a supervisory artifact, a human decision maker on every trade by design, full in tenant auditability of the engine's predictions and their grades, and documentation written for the deployer's own conformity work.
SEC RECORD KEEPINGExchange Act Rule 17a-4 as modernized October 2022 · Advisers Act Rule 204-2(a)(7)
What the rules ask: electronic records preserved in systems that permit viewing, download, and audit, and preservation of communications relating to recommendations and advice, a perimeter enforcement has been widening since 2022. What the system implements: sealed, exportable evidence chains for every decision, retained inside the firm's own storage under the firm's retention policy, in a format built for examination rather than reconstructed for it.
ALLOCATOR ODDAIMA DDQ standard · AI governance now on most institutional checklists (industry ODD analyses, 2025)
What allocators ask: whether the manager holds a formal policy on AI usage within the investment process, and how that usage is governed and evidenced. What the system provides: the answer as an artifact. A documented governance posture, role separated oversight surfaces, and a contemporaneous record of supervised decision making that no other layer of the stack can produce.

Bloomberg and Charles River do not publish their internals, and neither do we. What a serious vendor owes a serious counterparty is not source code. It is evidence of understanding the obligations the counterparty actually carries, and a system already shaped to them.

VII.

Pilot terms.

Ninety day evaluation

ONE DESK · FIRM TENANT
SCOPE
A single desk, with portfolio managers participating on an opt in basis.
ENVIRONMENT
Deployed in the firm's own tenant from day one; nothing is hosted by the vendor at any point.
MANDATE REGISTER
Configured with the firm's actual limits, not demonstration values.
PRIVACY
The trader plane is structurally private from the first session, which is stated to participating traders in writing.
DECISION
Exit criteria are agreed before the pilot begins, and the decision meeting is scheduled at kickoff, not after.
VIII.

Provenance.

DojiPad Institutional is built by ARQNXS, the firm behind DojiPad, a behavioral coaching platform for discretionary traders. Founder Nigel van der Laan is a discretionary order flow trader with a background in biopsychology and neuroscience, building trading infrastructure since 2017.

The behavioral engine was proven on the retail platform: model a trader's specific failure patterns, evaluate plans before execution, grade every prediction against fills. The institutional system wraps that engine in the deployment architecture professional desks require, and in the design decisions their traders will accept.

Paper series

Four short papers on the thinking behind the system, publishing weekly. Subscribers receive each one on release.

IThe supervisory record stops where the decision beginsFIRST
IISoftware that cannot see your tradesFOLLOWS
IIIThe firm sees process. The trader keeps the conversation.FOLLOWS
IVWe built a system that could block trades. It never will.FOLLOWS

Request a briefing

The system running live, on a seeded demonstration firm, viewed from the officer register and the trader plane in turn. Attendance from the risk function is welcome and usual.

A briefing is the system running live, on a seeded demonstration firm, viewed from both planes. A regulatory alignment memo is available in advance on request.