Your whole book in one live view — research, your own models, and human-approved, audit-logged execution.
It connects straight to your brokerages through Plaid — live positions, never a stale export.
secure link
Schwab$1.4M
Fidelity$0.7M
IBKR$0.3MLive portfolioreal-time
SEC EDGARReuters · earningsBloomberg · filingsInteractive BrokersOrder ticket
Approved by Priya M., Portfolio Manager · 09:14 ESTThis morning, before the open
Your models ran. You approved. Nothing drifted.
Signed off by Priya M., Portfolio Manager
We connect your brokerage and run the models. You approve every trade.Chronexa doesn’t sell an AI or a trading strategy. We orchestrate proven models — XGBoost, LSTM, Claude — with your brokerage and your sign-off. We build the system; you own every decision.
Click a step above to jump · the run loops on its own
What it is
What is the AI Investment Research Engine?
The Investment Research Engine is an operations layer for a research and portfolio team. It pulls your whole book across every custodian into one live view, runs your research and your own models on that live data automatically, and turns what the PM approves into logged orders — every execution human-authorised, every position private to you. It removes the hours between a signal and an action. It does not decide what to buy.
Be clear on what it is not: it is not an “AI that beats the market,” and it makes no promise about returns. It surfaces candidates and organises the evidence — cited back to the filing, the transcript, or the model rule that triggered — and the portfolio manager exercises discretion on every one. The value is speed and rigour on the operational work, not a claim to alpha. The models it runs are your models; where we name techniques like gradient boosting or fractional-Kelly sizing, those are implementation details, not the pitch.
That design is what makes it usable at a regulated firm. Every trade is human-authorised before it routes, and who approved what — and when — is logged for your books and records. Your holdings, signals and client data stay inside your own environment and are never sent to a public AI service or used to train anyone’s model — the control an RIA’s compliance function needs before any model touches the book.
How it works
How the Investment Research Engine works, step by step
Six steps take the desk from a scattered morning to a decision-ready view — data always from live sources, never a cached spreadsheet. Each step is specialised, and a person stays in control of every trade. Here is exactly what happens, and where the judgment stays yours.
- 01
Unify the book
Connects directly to your custodians and portfolio systems — Schwab, Fidelity, IBKR, and aggregators like Addepar or Orion — not consumer account-aggregation rails. It pulls holdings, lots and cost basis, and cash across every account into one live view, with no morning export and no CSV. Everything downstream runs against what you actually hold right now.
What you get A single live view of the whole book — every account, before any research begins.
- Schwab API
- Fidelity API
- IBKR API
- Addepar
- Orion
- 02
Systematic research
A research agent reads news, earnings-call transcripts, SEC filings and analyst commentary across every holding and watchlist name, and surfaces what is material — each item linked back to the filing or transcript it came from, so the analyst reads the evidence rather than a summary they cannot check. It does the morning read; the analyst forms the view.
What you get A prioritised, cited research feed for every position — the reading done, the judgment still yours.
- News APIs
- SEC EDGAR
- Earnings transcripts
- Analyst feeds
- Your research notes
- 03
Your models & rules
The engine runs the models and rules your team already uses — a factor screen, a mean-reversion rule, or your own ML — on live data automatically, instead of only when someone remembers to refresh a spreadsheet. Each candidate is shown with the rule that triggered it and sized to your written limits (fractional Kelly, or your own sizing method). These are decision inputs the PM reviews — never instructions, and never a claim about what a security will do.
What you get Your own models, run continuously on live data — each candidate shown with its trigger and sized to your policy.
- Your signal models
- Rules engine
- Position-sizing
- Backtest harness
- 04
Human-approved execution
Every candidate the PM accepts becomes a draft order — ticker, direction, size, order type. Nothing routes to the broker until a named person approves it. Once approved, it executes via your broker or OMS, and fills, partials and rejections are reconciled back to the book. Who approved what, and when, is recorded for every single order.
What you get Approved orders placed in seconds — with a complete, timestamped record of who authorised each one.
- Broker / OMS API
- Human approval gate
- Audit log
- Fill reconciliation
- 05
Risk vs your limits
Portfolio-level risk — exposure, beta, drawdown, sector concentration, correlation — is watched continuously against the limits written in your investment policy, not spot-checked when someone logs in. A breach triggers an immediate alert, and the current state feeds back into the next research cycle so it always runs on live data.
What you get Continuous risk visibility against your own limits — a breach reaches you the moment it happens.
- Live P&L
- Exposure limits
- Drawdown alerts
- Concentration
- Correlation
- 06
Tax-aware rebalance & report
When allocation drifts past your threshold, the engine computes the minimum set of trades to return to target and flags tax-lot harvesting opportunities lot by lot, not as a rough estimate. The plan is presented for approval, never executed on its own, and the result is formatted into a client-ready report.
What you get A tax-aware, minimal-turnover rebalance plan and a client-ready report — without the spreadsheet.
- Drift detection
- Tax-lot analysis
- Rebalance scheduler
- Client report
The problem
The research problem it solves
Portfolio managers and research analysts at mid-size investment firms face a structural problem: the tools exist to do quantitative research, but the data pipeline between market sources and model inputs is entirely manual.
- Pulling portfolio data from multiple brokerages into a single view takes 1–2 hours per morning before any analysis begins.
- News and sentiment monitoring is ad hoc — the analyst reads what they happen to see, not a systematic signal scan calibrated to their holdings.
- ML models exist but sit idle because re-running them requires a manual data refresh and export cycle.
- Rebalance calculations live in Excel — not tax-aware, not version-controlled, and one formula error away from a costly mistake.
- Trade execution is disconnected from the model output: the signal lives in one tool, the order is placed in another by hand.
- Portfolio-level risk metrics — beta, Sharpe, drawdown — are checked periodically, not continuously.
The engine does not replace the portfolio manager — it removes the hours between signal and action, so the manager spends time on judgment, not data plumbing.
Time to value
How fast you go live
Most teams are live in 2–3 weeks.
- Week 1Connect custodians & systemsAuthenticate your custodians and portfolio systems — Schwab, Fidelity, IBKR, and Addepar or Orion. Map holdings and lots. Validate every position against your own records before anything runs.
- Week 1–2Load your models & policyBring in your watchlist, your signal models or rules, and your written investment policy — sector limits, position caps, drawdown tolerance. Where you want a backtest, we run your rules against your own history so you see how they would have behaved before anything is live.
- Week 2Approval gate + executionConnect your broker or OMS and wire the human approval gate. Run draft orders through the approval flow before any live order is placed.
- Week 2–3Monitor and calibrateRun live with daily review. Tune thresholds and limits against what you actually observe before full rollout.
What you need to start
- Portfolios held at your custodians (Schwab, Fidelity, IBKR, Alpaca) or aggregated in Addepar or Orion.
- A written investment policy — sector limits, position caps, drawdown tolerance.
- A designated approver for orders — PM, CIO, or compliance officer.
- Your models or rules, and historical data if you want them backtested — even if they live in a spreadsheet today.
Your holdings, signals and client data never leave your environment and are never used to train anyone’s model. The engine runs inside a tenant you control — the requirement any RIA’s compliance and client agreements impose before an AI system touches the book.
ROI
The return on an Investment Research Engine
The return here is operational, not a performance claim. The hours a PM and analysts lose every morning to pulling positions, stitching spreadsheets and re-running models by hand come back — redirected to research, client conversations and judgment. It also removes a class of quiet risk: a broken Excel formula in a rebalance, a limit breach noticed late, a trade decision with no clean audit trail. We do not promise returns and we never will. What we can show you, before you commit, is the engine running your own models on your own book in a read-only pilot.
Proof
How we prove it — before you commit
We connect to a read-only copy of your book and run the engine on your real positions and your own models — you watch it work before anything is ever wired to a broker.
Where you have a signal model or rule, we backtest it against your own history and show you how it would have behaved — your numbers, not ours. We publish no performance claims of our own.
Every draft order, approval and limit check is logged from day one, so your compliance team sees the exact audit trail they will keep — before you commit to anything.
FAQ
Investment Research Engine FAQ
Does the engine place trades automatically?
No. Every candidate the PM accepts becomes a draft order, and nothing routes to the broker until a named person approves it. The approval can be one click in a dashboard or a message in Slack — your choice — and the audit log records who approved, when, and what the order was. This is a hard design constraint, not an option.
Which custodians and systems does it connect to?
It connects directly to your custodians and portfolio systems — Schwab, Fidelity, IBKR, Alpaca, and aggregators like Addepar or Orion — for holdings, lots and cash, and executes through your broker or OMS. It is built for how a firm actually custodies assets, not consumer account-aggregation.
Does the engine promise returns or "alpha"?
No, and it never will. The engine runs your models and rules — not a black box we claim beats the market — and organises research cited to its source. Where you want a backtest, we run your rules against your own history so you see how they would have behaved; we publish no performance figures of our own. It is an operations and risk layer, not an alpha claim.
Is this suitable for a registered investment adviser?
Yes — it is built for SEC-registered advisers. Every trade is human-authorised and logged, your data stays in your own tenant, and the engine provides decision inputs, not investment advice; the adviser retains full discretion and accountability. The human-approval audit trail is designed to support your books-and-records obligations, and your compliance team reviews the setup before go-live.
Where does our portfolio data go? Is anything sent to public AI?
Nothing is sent to a public AI service. Your holdings, signals, models and client data are processed inside your own environment or a dedicated tenant you control, and are never used to train anyone’s model. For a book of business, that data boundary is the entire point.
What happens if the market moves against a position?
Stop levels come from your own policy, not ours. If a position breaches a stop you have set, the monitor alerts you and queues an exit order for approval the moment it triggers. The engine surfaces the decision — it does not override your risk rules or act on its own.
Bring us the workflow that keeps eating your team's week.
Let's find the first one to fix.
The audit is free. If we can't find automation worth more than it costs to build, you owe us nothing — and you keep the roadmap.
Prefer email? info@chronexa.io
Or tell us what's slow
We'll review your workflows and come back with where AI saves the most time and cost.