$11 Trillion Behind the Curtain
In 2019, I walked into BlackRock's risk analytics floor for the first time. I was joining the team behind Aladdin — the operating system that manages $11 trillion in assets. If you've never heard of it, that's by design. Aladdin doesn't have a marketing page for retail investors. It wasn't built for you.
Aladdin is what happens when you give 17,000 engineers unlimited budget and tell them to build every financial model ever published. Discounted cash flows. Residual income. Regime-switching models. Monte Carlo simulations. Bayesian inference. Topological data analysis. All of it, running 24/7, across every asset class on earth.
My job was to sit inside that machine and help it think.
The Moment That Changed Everything
There's a screen in Aladdin — I can't tell you which one — that shows you a company's fair value from multiple independent models side by side. Not one estimate. Not two. Thirteen. Each model approaches the same question from a completely different angle, and you see them all at once.
When they agree, you have conviction. When they diverge, you dig deeper.
I remember staring at that screen one evening and thinking: this is the most important tool in finance, and 99.97% of investors will never see it.
Not because they're not smart enough. Because Bloomberg charges $25,000 a year for their terminal. Refinitiv charges $22,000. FactSet charges $12,000. And Aladdin? Aladdin isn't even for sale. You have to be a BlackRock client managing billions.
The entire financial information industry is built on a simple premise: charge professionals what they can afford, and lock everyone else out.
I started losing sleep over that.
What Bloomberg Doesn't Want You to Know
Here's the thing about Bloomberg Terminal — it's incredible software. I respect it. I've used it. But let me tell you what it actually is beneath the surface.
Bloomberg gives you data. Mountains of it. Real-time quotes, news, chat, analytics. But when it comes to valuation — the single most important question an investor asks: "What is this company actually worth?" — Bloomberg gives you a basic DCF template and says "good luck."
You still have to build the model yourself. You still have to pick your own assumptions. You still have to pray your Excel formula doesn't have a circular reference in row 847.
$25,000 a year. And you're still doing the math by hand.
The institutions don't use Bloomberg for valuation. They use Bloomberg for data, and then they feed that data into proprietary systems like Aladdin. Systems you'll never touch.
That's the gap. Not data. Not news. Not real-time quotes. The gap is the valuation engine itself.
Quitting
I left BlackRock in 2023. My colleagues thought I was insane.
"Where are you going?"
"I'm going to build the valuation layer that doesn't exist for retail investors."
"...You mean a DCF calculator?"
"No. I mean all thirteen models. Running on every US-listed stock. Updated every market day. For a fraction of what Bloomberg charges."
Silence.
Then: "That's a $200 million engineering project."
"Not anymore," I said.
Building the Engine
They were right that it's a massive engineering challenge. They were wrong about the cost.
I chose Rust. Not because it's trendy — because when you're computing 13 independent valuations across 6,000+ companies every single day, you can't afford a language that wastes cycles. Python would take 14 hours. Rust does it in under 3.
I rebuilt every model I'd seen behind the curtain:
Bayesian DCF — Not your spreadsheet DCF. This one treats every assumption as a probability distribution, runs thousands of scenarios, and gives you a confidence-weighted fair value.
Earnings Power Value — Strips out growth entirely. Asks: "If this company never grew again, what would it be worth?" Brutal honesty.
Markov DDM — A dividend discount model that understands companies go through phases. High growth. Stable maturity. Decline. It models the transitions.
Dynamic NAV — Breaks the company into its assets, applies sector-specific recovery rates, and asks what a liquidator would pay. The model Wall Street uses when they smell blood.
EROIC Spread — Compares what a company earns on its capital (ROIC) versus what that capital costs (WACC). The spread tells you if the company is creating or destroying value. Simple idea, devastating accuracy.
ML Residual Income — Takes the classic residual income model and replaces human assumptions with machine learning predictions trained on 10 years of financial history.
First Chicago — Models three futures: bull case, base case, bear case. Assigns probabilities. Weighted average. This is how private equity values companies before writing nine-figure checks.
PWERM — Similar philosophy, but with Monte Carlo simulation. Thousands of scenarios, not three.
Regime Cross-Sectional — Detects the current macroeconomic regime (expansion, contraction, crisis) and values companies relative to their sector peers within that regime.
Sentiment SOTP — Sum-of-the-parts valuation adjusted by market sentiment. Because sometimes the market's mood matters more than the math.
FTNN Topology — Uses topological data analysis to find hidden patterns in financial time series that traditional models miss entirely. This is the one my former colleagues would raise an eyebrow at.
RCMH-DCF — A regime-conditioned DCF that adjusts its discount rate and growth assumptions based on where we are in the economic cycle.
CUCE Ensemble — The meta-model. It takes all 12 models above, weights them by historical accuracy, penalizes correlated errors, and produces a single unified estimate. The model of models.
Every stock. Every model. Every market day.
The Quality Layer
Valuation alone isn't enough. A stock can look cheap and still be a terrible investment. Enron looked cheap at $40. Lehman looked cheap at $20. WeWork looked cheap at... well, it always looked expensive, but you get the point.
So I built two additional scoring systems:
Quality of Company (QOC) — A 1-to-10 score across 8 pillars: profitability, growth trajectory, balance sheet strength, cash flow quality, capital efficiency, earnings stability, shareholder returns, and governance signals. A score of 8.5+ means the company is genuinely excellent by every measurable dimension.
Value Trap Detector — A dedicated system that flags stocks where cheap valuation is a symptom of structural decline, not an opportunity. Declining revenue, rising debt, negative free cash flow, insider selling, accounting red flags — if the pattern matches, the trap score goes red.
Together, they answer the two questions every investor needs: "What's it worth?" and "Is it worth owning?"
Why I Named It CirclFi
Because the circle closes.
The same models that BlackRock uses to manage $11 trillion. The same analytical rigor that Bloomberg charges $25,000 to approximate. The same confidence framework that hedge funds guard behind NDAs.
All of it. Running on every US-listed stock. Updated every market day. Accessible to anyone with an internet connection.
The circle of information asymmetry — where institutions see everything and retail investors see nothing — finally closes.
What's Live Today
CirclFi covers the entire US equity market:
- 6,000+ stocks valued by 13 independent models
- Quality Score on every company (1–10)
- Value Trap detection on every company
- Updated daily after market close
- Price, upside %, and confidence % for every model on every stock
No Excel. No Bloomberg. No $25,000 subscription. No Aladdin access required.
The Bloomberg Comparison
I get asked this constantly: "Is CirclFi a Bloomberg competitor?"
No. Bloomberg is a data terminal with chat, news, real-time quotes, and thousands of functions. It's a Swiss Army knife for finance professionals.
CirclFi does one thing: it tells you what a company is worth.
Bloomberg shows you 50,000 data points and says "figure it out." CirclFi runs the math for you — across 13 models — and shows you the answer.
Bloomberg is the library. CirclFi is the analyst who read every book in it.
What's Next
I'm building in public. Every improvement to the engine, every new model, every fix — it ships to all subscribers automatically.
The institutions have had a 40-year head start. I'm not trying to catch them in a day. I'm trying to make sure the gap stops growing.
If you've ever opened a 10-K filing and wished someone would just tell you what the company is worth — not what some analyst with a conflict of interest thinks, but what the math actually says — that's what CirclFi is for.
I spent years helping the largest asset manager on earth value companies. Now I'm helping everyone else do the same thing. If you have questions about quantitative valuation, financial engineering, or building in Rust — I'm here. Drop a comment.
Top comments (0)