Subject
Every alert names what it's actually about — a position, a sector tilt, a correlation pair — not a generic risk warning.
Portfolio Intelligence
A cockpit that tells you what's actually happening in a portfolio — and why, not a wall of alerts.
Verdicts on the book you already run — backed by real computed evidence across correlation, asset flow, stress, and bias — not a black-box score or a flat list of warnings.
Decision Spineverdict feed · live
Correlation clusterevidence
Asset-flow shiftflagged
Stress exposureelevated
Bias driftwatch
Thesis gate 3 of 4on track
vs BIST100 TRahead
01 / Cockpit
Every alert has a subject, an evidence drawer, and a real number behind it.
This cockpit was rebuilt after an earlier version was described internally as veneer — it looked complete, but the numbers behind the verdicts weren't real. Every verdict reason here is backed by actual share-of-signal math.
Evidence drawer
Open any alert and see the real data behind the verdict — the computed numbers themselves, not a description of them.
Filtering
Narrow the feed by subject, severity, or app — a busy morning shouldn't mean forty alerts read one at a time.
Clustering
Related alerts group together, so three symptoms of the same underlying shift read as one story, not three unrelated lines.
Deep-links
Every alert links straight into the underlying data — the position, the correlation matrix, the flow series — no separate lookup required.
Share-of-signal math
Verdict reasons are computed as an actual share of the signal driving them, not assembled from a canned sentence.
Evidence, not assertion
Every verdict reason is a real share-of-signal number you can open and check, not a sentence that sounds plausible.
Clustered, not scattered
Related alerts surface as one cluster, so the cockpit reads like a story instead of a stack of unrelated lines.
One click to the data
Deep-links take you straight to the position, the matrix, or the series behind the alert — no separate lookup.
A verdict without an evidence drawer is just an opinion with a timestamp.
02 / Essence
Four cards, not one black-box score.
Correlation structure, asset-flow patterns, stress exposure, and behavioral bias — each distilled into its own card, each backed by real computed evidence, not a single number standing in for all four.
Correlation structure
How the book's positions actually move together right now — not an assumed diversification, a computed one.
Asset-flow patterns
Where money is actually moving inside the book and around it, read from real flow data.
Stress exposure
What happens to the book under a real stress scenario, not a generic risk label.
Behavioral bias
The same computed bias signals the platform's behavior engine already tracks, surfaced at the portfolio level.
No single risk score
Concentration, flow, stress, and behavior each get their own card because they call for different responses.
Same engines, portfolio level
The bias card draws on the same behavior engine used elsewhere on the platform — not a separate, simplified model.
Read in one glance, checked in one click
Each card opens into the same evidence-drawer detail as the cockpit's alerts.
Four honest numbers beat one confident-sounding one.
03 / Workbench
The PM workbench: four gates a thesis has to clear, and a straight line to action.
A thesis-tracking view structured around four gates, with bridges into the platform's strategy-building tools the moment a verdict says something needs to change.
PM Workbench
From verdict to thesis to action, without leaving the cockpit.
Every position or portfolio thesis is tracked against four gates, not a single conviction rating. When a verdict from the cockpit suggests something has changed, the workbench bridges straight into the platform's strategy-building tools — and the propose-from-cockpit loop turns "here's a problem" into a concrete next step, instead of a dead end.
Four gates per thesis, tracked not assumed
Bridges straight into the platform's strategy tools
Propose from the alert — not a dead end
Thesis workbenchgates · propose loop
1 A verdict surfaces a concern cockpit alert
2 Thesis checked against four gates PM workbench
3 Bridge to the strategy tools if action is needed built in Quant Lab
4 Propose straight from the cockpit not a dead end
This is the consumption half of the platform's decision spine — Quant Lab is where a strategy gets built and validated; the workbench is where you track whether the thesis behind a position in your book is still holding, and where a verdict turns into an actual next step.
A problem surfaced without a next step is just a notification.
Benchmark
Health measured against a benchmark that actually pays dividends.
Performance compares against the total-return BIST100 — dividends included, not just the price index — alongside a Sharpe ratio and TLREF as the reference funding rate.
The price index alone understates a long-horizon comparison — dividends compound. Switching the benchmark to total return closes that gap and gives an honest answer to "has this portfolio actually beaten the market."
A price index doesn't pay dividends. Ours does.
Total-return BIST100, not price
Dividends are included in the benchmark, which materially changes any long-horizon comparison.
Sharpe ratio, computed
Risk-adjusted return sits alongside raw performance, not instead of it.
TLREF as the funding-rate reference
Excess return is measured against the actual reference rate, not an arbitrary benchmark.
Built with Quant Lab, monitored here
Strategy construction and validation happen in Quant Lab; Portfolio Intelligence is where you watch how the resulting book is actually doing.
04 / FAQ
The questions an advisor asks before trusting a verdict.
Short, specific answers — the same understated register as the rest of this platform.
Is a verdict actually a black box?
No — every verdict reason is a real share-of-signal number, and the evidence drawer shows you the exact computation behind it. You can open it and check the number yourself; nothing is asserted without something to point at.
How does this relate to Quant Lab?
Quant Lab is where a strategy gets built and validated — the construction side. Portfolio Intelligence is the monitoring side: once a strategy or a position is actually in the book, this is where you watch how it's doing and get told when something changes. They're two halves of the same decision spine, not competing tools.
What happens when an alert turns out to be a false positive?
The evidence drawer is exactly where you check that — the real number is right there, so you can judge for yourself whether the alert holds up. Filtering lets you deprioritize a class of alert once you've decided it isn't useful for your book.
Was this cockpit always this rigorous?
No, and that's worth saying plainly: an earlier version had the structure of a cockpit but not the computation behind it — verdict reasons weren't backed by real numbers. It was rebuilt specifically to fix that, the same way other parts of this platform have been rebuilt when they were found to be underbuilt.
Why total return and not the price index for the benchmark?
Because dividends compound, and a price-only index quietly understates any long-horizon comparison. Switching to total-return BIST100 gives an honest answer to whether a portfolio actually beat the market, not a flattering one.
Give your desk a cockpit it doesn't have to take on faith.
Request a walkthrough focused on your book, your alert history, and what a first verdict would actually show in week one.