Enterprise Platform

Research, Execution and Review — One Loop

Harold researches and originates. Mack executes, one dedicated agent per trade, inside the boundaries you set. Iris grades the whole book four times a day and corrects the drift. Every decision is logged with its reasoning, its evidence, and its order receipts.

3 Agents Strategist · Executor · Reviewer
3 Brokers Equities · Options · Crypto · Perps
Full Audit Every Query, Tool Call, Order
Trade Canvas — Fleet 4 agents active
AUM$2.41M
Open P&L+$18,420
Deployed64%
Adherence92%
AAPL wheel Auto +$12,800
12345
Alpaca · $400k
BTC-USD scale_and_ladder Auto +$4,220
12345
Kraken · $800k
NVDA event_surfer Awaiting catalyst +$1,400
12345
Hyperliquid · $200k
Iris · pre-close review: 1 idea drifting from its grammar — NVDA time-stop is 3 days past the style's proven hold. Correction proposed.
Illustrative fleet. Not a performance record.

Interface rendering. Illustrative data — not live positions.

The loop

From Data to Execution to Correction

Six layers. The last one feeds back into the second, which is what makes it a system rather than a pile of tools.

Market Data

50+ sources across six timeframes. Price, options chains with greeks, 80+ indicators, alternative data, news, fundamentals. Automated gap detection and backfill, plus your own proprietary feeds.

Research (Harold)

Natural language in, structured analysis out. 45+ tools chained automatically, every number grounded against the tool output, every options contract validated against the live chain.

Origination

/hunt and the setups pipelines turn the universe into a ranked, priced slate. Promotion pushes a curated idea straight from chat into the execution platform as a live trade idea.

Evidence (Style Matrix)

17 playbooks graded across 10 years and 4 market eras. The matrix picks the playbook and the risk tier; the grammar hands the executor its stop distance and time limit.

Execution (Mack)

One agent thread per trade idea. Structured decisions, three-layer pre-trade validation, multi-broker routing, virtual stops where the venue has none, and full order receipts.

Review (Iris)

Four daily checkpoints grade the live book on adherence, expectancy and drift, then dispatch corrections — bounded by the same matrix that made the original call.

Three agents, three jobs

Separation of Duties, Enforced in Code

The strategist cannot execute. The executor cannot pick its own playbook. The reviewer cannot migrate a trade into a refuted strategy. These aren't prompt instructions — they're tool surfaces, and each agent is bound to its own.

H
Harold
Strategist

Reads the question, chains the tools, reads the scorecard, and originates the trade idea with the playbook and risk tier the evidence supports.

  • Research surface: data, discovery, calendar, backtest, web, output
  • Read-only view of the live book plus trade-idea promotion
  • Promotion is chat-only — excluded from every autonomous cycle
  • Never places an order directly
M
Mack
Executor

One dedicated thread per trade idea, rebuilt fresh each cycle. Assesses the mission, decides, and submits — through exactly one order tool.

  • ~12 tools bound per cycle, no per-surface subsetting
  • assess_mission is the single per-cycle read
  • submit_decision is the only order or exit path
  • Stop distance and time limit come from the grammar, not from judgment
I
Iris
Reviewer

The performance manager. Grades the fleet against the evidence, dispatches tuning, and escalates what it can't fix on its own.

  • Four daily ET checkpoints, deduplicated across workers
  • Adherence graded against the grammar in force at trade time
  • Blocked in code from targeting avoid, refuted or cost-death cells
  • Shadow mode by default — observe before it acts
Read + flagAutonomous monitor cycle
Assist / AutoPer-idea agent modes
1–5Risk dial tiers
Kill switchStop one agent or the fleet
Harold

Research That Ends in a Routable Order

Harold chains analytical tools on every query — deciding which to call, in what order, and how to synthesize the results. On Enterprise the output doesn't stop at a document: one conversation can produce a backtest, a validated multi-leg structure, and a trade idea promoted into the execution platform.

  • Tool chaining is visible. Every call, its arguments and its result are shown and logged — the audit trail is the interface.
  • Grounded numbers. Deterministic classification of every numeric claim: sourced, derived, analytical or unverifiable. No second model grading the first.
  • Contract validation. Options structures are checked against the live chain before they are ever shown, let alone routed.
  • Promotion with a wizard. prepare builds the draft with no database write; promote validates and pushes the confirmed idea to your instance.
  • Model choice. Configure the model slot per stage; bring your own provider keys, stored encrypted.
Harold — Research Chat Online
You
Backtest a covered call on AAPL over six months, then build a hedged structure with SPY puts sized to $50k max risk.
H
get_indicators AAPL · 1d · 6mo
backtest_strategy buy_write · 4 folds
get_options_data AAPL · monthly chain
get_options_data SPY · hedge puts
verify_option_contracts 4 legs · all resolve
tc_promote prepare · draft only

Backtest: fold-mean R/DD +1.38, positive in 4/4 eras, 74% win rate, -6.4% worst-era drawdown. Tier solid, confidence confirmed for mega-cap daily.

Structure: sell 3× AAPL 245C monthly, buy 2× SPY 580P. Net credit $4,260, max risk $48,740 — inside the $50k bound.

14 sourced 6 derived 3 analytical 100% grounded
Draft trade idea ready — review the wizard payload, then promote to route.
Backtested results are historical and do not predict future performance.

Rendering of the research chat. Illustrative data.

The style matrix

The Playbook Is Chosen by Evidence, Not by Vibe

Seventeen playbooks, each with a machine-readable grammar — entry expression, stop distance in ATR units, time-stop, trailing and scale-out behaviour, regime gate. Each one graded per asset segment, timeframe and security type.

The heavy statistics run offline in the research harness. What reaches the live path is a table: for this situation, these playbooks have proven edge, ranked, with a confidence label and a risk tier attached.

  • Walk-forward across four eras — bull, crash, bear, recovery. We report the mean and the worst era.
  • Confirm-grids and plateau checks promote a result from provisional to confirmed. Only confirmed cells are recommended.
  • Cost gating. Cells whose edge is smaller than realistic spreads are marked cost_dies and structurally excluded.
  • Per-style indicator packs feed the executor only the gauges its grammar reads — about 72% smaller than the old universal bag, with off-pack gauges available on demand.
  • Allowlists and advisories. Index-wheel underliers are allowlisted, inverse products hard-blocked, levered products flagged through the promote advisory channel.
Style Matrix — cell detail income_harvest
Income Harvest Solid Confirmed mid-cap · 1d · options
Fold-mean+1.42
Fold-min+0.19
Win %71%
Dial tier2
Bull+1.9
Crash+0.2
Bear+1.4
Recovery+2.2
gate iv_rank ≥ 70  ·  dte 30–45  ·  delta -0.20 to -0.30
exit 50% of max premium or expiry  ·  stop none (collateralized)  ·  roll chain-resolved
SegmentTierFold-meanStatus
mid-capSolid+1.42Locked
small-capSolid+1.16Locked
mega-capAvoid-0.08Gated off
cryptoAvoidNo chain edge
Hypothetical backtested performance. Not a projection of future results.

Rendering of a style-matrix cell. Illustrative data.

Mack

What Happens on Every Agent Cycle

Each active trade idea gets its own thread. The agent is rebuilt fresh each cycle so state can't quietly accumulate between decisions.

Assess

A single call returns the whole mission picture: live market state with the style's indicator pack, the regime stack across daily, 4-hour and 1-hour, open positions with P&L, working orders, remaining budget and buying-power coverage.

Analyze

The agent evaluates that context against the resolved grammar — entry expression, stop distance, time-stop, exit rules — plus the thesis, any strategy intel Harold has distilled, and the setup-alignment cross-check that grounds its confidence.

Decide

A structured decision with reasoning and a confidence score. One tool submits orders and exits; everything else is read-only. Assist mode proposes and waits for approval, auto mode acts and then reports.

Execute & reconcile

Approved orders pass three validation layers before reaching the broker. A post-decision reconciler then runs winner management and the deterministic options-exit grammar against what actually filled.

Mack — AAPL wheel · cycle 214 2 min ago
Decision
TAKE_PROFIT
Confidence
87%
Grounding
Setup alignment 3/4 · regime stack aligned long · 12 hunting cycles since entry
Reasoning
Short call at 82% of max premium with 3 DTE. IV contracting post-earnings and the grammar's exit rule fires at 50%. Close to bank the premium and re-enter the cycle at the next monthly.
Order
BUY_TO_CLOSE 3× AAPL 260320C245 @ $1.20 LMT · Alpaca
Validation
Account consistency ✓   Strategy budget ✓   Broker balance ✓
Every cycle is persisted with its prompt, tools, decision and receipts.

Rendering of an agent decision record. Illustrative data.

Bounded by design

Autonomy without boundaries is just an unsupervised bot. Every dimension of what an agent may do is a configured limit, enforced below the model.

  • Risk dial, 1–5. Capital Preservation caps a single order at 10% of the idea's capital and daily loss at 1%; Degen lifts that to 100% and 15%. Compliance runs on every idea, unconditionally — there is no master switch to turn it off.
  • Per-idea budget. Allocated, used, reserved and available capital tracked separately, with a batch pre-flight before any multi-leg submission.
  • Mission lifecycle. One-shot ideas retire when the position closes; ongoing ideas keep cycling. The style's archetype sets the default.
  • Cadence floor. The playbook sets the minimum time between cycles, so a patient strategy can't be run like a scalper.
  • Cost ceiling. A daily spend limit per agent, with auto-pause after consecutive non-transient errors.
  • Kill switch. Stop one agent, emergency-exit one idea, or halt the fleet.
Iris

Someone Has to Grade the Tape

Most automated systems have no feedback loop — strategies drift, parameters go stale, and nobody notices until the drawdown. Iris closes that loop on a schedule.

Four times a day, in defined ET windows, Iris builds a scorecard for the live book: is each idea actually following its playbook, is expectancy tracking the evidence, are the monitor's flags being resolved. Then it dispatches corrections.

  • Adherence, honestly measured. Trades are graded against the grammar in force at the time of the trade, inside the grade window — not re-graded against today's settings.
  • Rich evidence table. Tier, fold-mean, regime split, reconfirmation status, dial tier and evidence grade — not a hand-maintained list of style names.
  • Matrix-bound migrations. Iris cannot move a trade into an avoid, refuted, cost-death or gated-off cell. That bound is enforced in the dispatcher, not in a prompt.
  • Flag intake. Concerns raised by the autonomous monitor during the session are surfaced into the review context so they get resolved rather than repeated.
  • Shadow mode by default. Run it observing for as long as you want before you let it act.
Iris — Fleet Review · pre-close 15:30 ET Shadow
08:30Pre-market
12:30Midday
15:30Pre-close
16:30Post-close
Fleet scorecard
AAPL · wheelmid-cap · solid · confirmed 94
BTC-USD · scale_and_laddercrypto · flagship · confirmed 88
NVDA · event_surfermega-cap · solid · confirmed 61
XLE · income_harvestmid-cap · solid · confirmed 38
Proposed — NVDA: time-stop is 3 days past the style's proven hold and the catalyst has passed. Re-seed the grammar's time_stop to the confirmed value, or exit per the playbook.
Escalated — XLE: adherence below threshold for a third consecutive window. Short-premium exits are firing early against the grammar. Flagged for human review; no automatic change dispatched.
Grades are internal process metrics, not returns.

Rendering of the fleet review scorecard. Illustrative data.

The execution platform

Where Strategies Become Live Positions

Each trade idea is a self-contained execution environment with its own capital, broker connections, lifecycle rules and dedicated agent. Run dozens simultaneously, isolated from each other, on a per-user instance.

Portfolio of Portfolios

Every idea carries its own capital allocation, multi-leg positions across multiple symbols, per-leg broker routing, and lifecycle rules. Budget is tracked as allocated, used, reserved and available — and a batch pre-flight prices the whole basket before anything is submitted.

Isolation — one instance and one database per user
Attribution — broker positions mapped back to the idea that opened them

Multi-Broker Routing

Route different legs of the same structure to different venues. Equities and options on one account, crypto spot on another, perpetuals on a third — each leg independently managed and reconciled.

Alpaca — equities and options
Kraken — crypto spot
Hyperliquid — perpetual futures
Paper alongside live to validate before committing capital

Virtual Order Monitor

Crypto and options venues often can't hold a native stop. The platform holds it instead — monitoring price on a tight loop and firing the exit itself when the level breaks.

Protective floor — a breached option stop fires a marketable limit at a floor, then escalates to market if it stays unfilled
OCO & trailing handled virtually where the venue lacks support, including fractional equity

Collateral Sequencing

Unwinding a covered position in the wrong order can leave you momentarily naked. The wave planner sequences it — coverage-consuming orders are staged behind the order that provides the coverage, and released only when it fills.

Pending-dependency status for staged children, cancelled if the parent dies
Pledged collateral tracked distinctly from order-reserved quantity
Refusals alert — a half-completed pair never fails silently

Lifecycle Automation

Wheels run themselves: cash-secured put, assignment, covered call, repeat — with a deterministic protective floor underneath the assigned-stock phase for when the thesis breaks.

Options exit grammar — close, scale, roll or bail resolved deterministically from a greeks snapshot with a staleness fail-safe
Winner management — bank part of a runner at the proven level, flip the rest to trailing

Execution Analytics

The decision-time quote is stamped onto every order at submit, so fill quality is measured against what the agent actually saw — not against a price reconstructed after the fact.

Fill-vs-mid capture computed on every fill
Event-sourced P&L — one ledger behind the header, each idea and every position
Broker-truth snapshots reconciled continuously
Pre-trade controls

Three Layers Before Anything Reaches a Broker

Every order — whether a human placed it or an agent did — passes the same three independent checks. Options carry their own collateral math for cash-secured puts, covered calls and naked positions.

  • Account consistency. Broker connection live, account in good standing, asset type supported on that venue.
  • Strategy budget. Allocated capital, reserved amounts and available balance within the trade idea — including the risk dial's per-order and daily-loss ceilings.
  • Broker balance. Real-time balance and collateral verification at the venue itself, with buying-power drift alerting below a warn threshold.

Ideas whose account lacks the buying power to run them are flagged capacity-blocked rather than left to fail at submission.

Risk dial — per trade idea Always on
TierNameMax / orderMax daily loss
1Capital Preservation10%1%
2Conservative20%2%
3Moderate33%3%
4Aggressive50%5%
5Degen100%15%
12345
20%/order · 2% daily
How it behaves
The dial constrains in-trade behaviour — size, daily loss, stops, trailing. It does not gate entries; that stays the agent's judgment inside the directive.
Per-idea overrides are supported, and mistyped fields come back as explicit warnings rather than being silently ignored.
An idea with no tier set falls through to Moderate. There is no configuration in which dial compliance is skipped.
Limits are enforced in the funds layer, below the model.

Rendering of the risk dial. Values match the shipped tiers.

Monitor feed

One Journal, Everything That Happened

A permanent, per-user event journal. Agent decisions, order fills, stops and targets, assignments and exercises, pattern signals, outsized market moves, macro briefings and system alerts — typed, deduplicated and timestamped.

On top of it, a periodic analysis cycle: Harold reads the recent activity on a cadence that adapts to what you actually hold — every 15 minutes during regular hours or for round-the-clock crypto ideas, stretching to four hours overnight, and skipped entirely when nothing is live.

  • Critical fast path. Fills, rejections, stop and target breaches and assignments bypass the poll and arrive immediately.
  • Repeat collapsing. Identical events are collapsed with a count, so a recurring problem reads as one issue to fix rather than forty lines of noise.
  • Escalation rules. Three consecutive agent errors, a position above a quarter of equity, a breached daily loss limit, or confidence trending down across cycles all surface for attention.
  • Read and flag only. The autonomous cycle cannot brief an agent, mutate a directive or place an order — it raises a flag that Iris picks up. Interactive monitor chat is where control tools live.
  • Compliance trail. Every event typed and logged for audit and regulatory review.
Monitor Feed Live
15:42
agent_decision
AAPL wheel — TAKE_PROFIT at 87% confidence. Short call closed at 82% of max premium.
Auto mode
15:42
order_filled
BUY_TO_CLOSE 3× AAPL 260320C245 @ $1.18 — 2bp better than the decision-time mid.
15:18
stop_loss
SOL-USD virtual stop breached at $182.40. Exit routed to Kraken, filled inside the protective band.
×1
14:55
option_event
XLE 260220P86 assigned. Wheel phase advanced to covered call; collateral reclassified, budget net-neutral.
14:30
monitor_flag
Harold flagged NVDA event_surfer — maintenance directive references a catalyst that has already passed. Raised for fleet review.
×3
14:00
macro_report
Afternoon briefing — breadth narrowing into the close, energy leadership intact.
Events are deduplicated at append and preserved permanently when critical.

Rendering of the monitor feed. Illustrative data.

Always on

Eight Briefings a Day, Fact-Checked Before They Are Saved

The market doesn't wait for your morning routine. Harold writes a briefing at every window that matters — from the Asia overnight session to the crypto evening wrap — and each one lands in your feed with the numbers already grounded.

  • Regime stack. Daily, 4-hour and 1-hour regime states for the market spine, so you can see when the timeframes disagree.
  • Day thesis. The pre-market briefing on trading days carries a directional call with a conviction level — the one deliberate exception to the observation-only house style.
  • Screaming signals. Outsized movers pulled out of the briefing and offered as trade ideas you can build on.
  • Rotation and breadth. Asset and sector rotation, daily movers, and a weekly rotation through macro series.
  • Feeds the agents. The regime stack the briefing renders is the same one the execution agents read in their per-cycle mission assessment.
Briefing schedule — ET 8 windows
05:00Asia overnight
08:00Pre-market
09:45Post-open
12:00Midday
14:00Afternoon
15:001h to close
16:15Post-market
20:00Crypto evening
◆ Neutral · Low Day thesis: Timeframes split — daily still constructive, 1-hour has rolled. No directional edge into the open; wait for the first hour to resolve.
Delivered into your feed
Four windows run on weekends and holidays too — crypto doesn't take days off.
Each briefing is fact-checked before it is saved, not after.
Market commentary, not a recommendation to buy or sell anything.

Rendering of the briefing schedule and day-thesis pill.

The toolset

45+ Research Tools, 12 Categories

Every tool returns structured data that can be chained, compared and built on across queries. Enterprise unlocks the execution and promotion categories on top.

Data & Discovery

get_dataget_price_dataget_indicatorsget_options_dataget_price_structuresearch_assetsresolve_symbolsget_market_stats

Six timeframes, 80+ wide-format indicators, full options chains with greeks, volume-profile price structure, and symbol discovery across thousands of instruments. Proprietary feeds join the same interface.

Backtesting & Portfolio

backtest_strategybacktest_optionsparameter_sweepregime_analysisrisk_diagnosticsanalyze_portfolio

Deterministic sims for equity grammar, modelled options premium and pairs, with realistic slippage. Parameter sweeps, regime analysis and portfolio risk diagnostics. Compute runs in an isolated sidecar, off the request path.

Pattern Detection

detect_chart_patternsscan_candlestick_patternspattern_scanscan_historical_patternsget_pattern_alertsanalyze_detector_performance

24 detectors with per-timeframe cooldowns, each signal carrying its backtested win rate and sample size. Detector performance itself is measurable, so a decaying signal can be identified rather than trusted.

Calendar & Events

get_upcoming_eventsget_earnings_calendarget_fomc_scheduleget_options_expirationget_current_time

NYSE trading calendar, earnings, options expiries and the macro schedule. Catalyst scans are batched into a single request so a wide hunt costs one call, not fifteen.

Fundamentals & Filings

get_fundamentalsget_sec_filingsget_filing_contentget_crypto_fundamentals

Financial statements, SEC filing search with full document retrieval, historical earnings, and on-chain crypto fundamentals.

Macro, Sentiment & Web

get_macro_dashboardget_fred_seriesget_sentiment_analysisnews_searchweb_searchsearch_prediction_marketsget_asset_prediction_insights

Macro dashboards and FRED series, cross-source sentiment, live web search for breaking context, and prediction-market implied probabilities on political and economic events.

Verification & Output

verify_response_accuracyverify_option_contractssummarize_researchrequest_prediction_reportget_user_risk_profile

Number-level grounding on every response, contract validation on every options structure, research summaries and saved reports. Risk-profile awareness is enterprise-only.

Execution & Control

tc_monitormarket_scantc_executetc_managetc_promotetc_flag

Enterprise only. Read the book, promote a curated idea from chat, start and stop agents, approve or reject proposals, brief an agent with time-boxed intelligence, and raise flags from the autonomous cycle. Each surface binds a different subset — promotion never appears in an autonomous cycle.

Your data, first class

Bring Your Own Signals

The platform ships with market data out of the box. Enterprise clients connect proprietary feeds that become first-class citizens in every workflow — not a bolted-on side table.

Your data flows through the same pipeline as everything else: available in research, in backtesting, on charts, and inside the agent's per-cycle context. No separate system, no export and re-import.

REST & WebSocket ingest

Push data through authenticated REST endpoints or stream it over WebSocket for real-time ingestion.

Custom data types

Define schemas for proprietary signals, alternative datasets or internal research output.

Full integration

Your feeds reach research, backtests, the toolset and the agents — exactly like built-in sources.

Market Data API

Read-only programmatic access to the whole data layer for your own models and systems.

Who it's for

Built for Institutional Teams

Hedge Funds & Family Offices

Evidence-backed research workflows, walk-forward backtesting, and bounded autonomous execution for systematic and discretionary books alike.

Broker-Dealers

Research infrastructure and order management that integrates with existing execution systems, with a complete typed audit trail for review.

RIAs

Structured research, portfolio monitoring and compliance-ready analytical workflows, with per-team roles and separated permissions.

Prop Trading

High-throughput data, real-time intelligence, and multi-strategy execution across equities, options, crypto and perps from one surface.

Infrastructure & Governance

Choose your AI Model slots configured per stage, with your own provider keys stored encrypted. Your prompts stay yours.
Isolated instances Each user runs a dedicated execution instance with its own database. No shared trading state.
Team roles Multi-user RBAC with research, execution and admin permissions kept separate.
Audit trail Every query, tool call, agent cycle, decision, order and event logged with timestamps and reasoning.
Custom SLA 99.5%+ uptime, defined response times, and service credits.
DPA GDPR and CCPA compliant, 72-hour breach notification, audit rights.
Onboarding classification FINRA-aligned questionnaire on signup, with the record retained for audit.
Dedicated support A named account manager and direct engineering access.

See the Full Loop Live

We'll walk you through research, origination, the evidence engine, agent management and live execution — on your symbols, with your scenarios.

Enterprise plans require verification of regulatory status or institutional affiliation.