Stage 01 the sketch

Lattice Chiti

structured intelligence

A quantitative firm turning market noise into structure, with AI, predictive models and our own data engine. One market today. Built to weave them all.

A company at the very start. You are looking at the drawing; scroll and it gets built.

Darshan at his desk behind a laptop, holding a mug of coffee and glancing over at you.

Hello.

my prediction

1089

the room is still a sketch. Darshan is real.

I'm Darshan. You caught me mid-build. Have a look around.

I've written a number on my laptop. Don't peek. Give me any 3-digit number.

You picked the number.
The rules picked the answer.

Try a thousand numbers and you'll land on 1089 every time. Under what looked like free choice sat a structure you couldn't see.

Markets look free too. Lattice Chiti is a quantitative firm being built to search for hidden structure there: maths to know where to look, AI and predictive models to look further than any person can, and research honest enough to say when nothing is there.

Curious about markets, or here to research?

One prize. Three doors. Your gut is about to lie to you.

A prize, three doors, one choice. Play on instinct and you'll win 1 time in 3. Know one small piece of maths and you'll win 2 in 3. Pick a door, and let's find out which kind of player you are.

Go on, pick one.

if you stay, you win

–

0 games

if you switch, you win

–

0 games

Your gut said 50/50. The maths says 2 in 3.

In 1990 a magazine printed this answer. Around ten thousand readers wrote in to say it was wrong, many of them with PhDs. They were the ones who were wrong.

Don't trust our doors. Check every case.

You picked door 1. The prize can only be in three places:

Still suspicious? Three cups, one coin, a friend. Kitchen table, five minutes.

This is what research is for.

Your eyes saw two closed doors and called it even. The truth was in the structure underneath: who knew what, and when. A gut feeling can't see that. Writing down every case can.

Markets ask this question every second. Two moves look equally likely, and underneath, something may be tilting the odds, or it's a mirage. We're building the instruments that tell the difference: maths sharp enough to see the tilt, AI patient enough to test it a million times, and the discipline to believe neither until it's proven.

Still being built. Want to build it with us? The researcher desk is one jump away.

Textbook name: the Monty Hall problem (conditional probability).

Seeing the move is the easy part.

A chart can show that price moved. Understanding how the market was behaving around that move takes more work. That gap is what we study.

Drawn for illustration. Not market data, and not a result.

1 · the move

In 1900, five years before Einstein explained the jitter of pollen in water, a young Frenchman named Louis Bachelier used the same mathematics to describe prices on the Paris Bourse. His professors shrugged. Today it's the foundation of modern finance. A price line is the footprint of millions of decisions. It tells you that something happened. Never why.

Textbook name: random walk, Brownian motion.

2 · the behaviour

Take two games. Each one, played alone, is a guaranteed loser. Alternate between them, and you can win. It's called Parrondo's paradox, and its lesson is strange: which game you're in can matter more than any single move. Markets switch games too. Calm, restless, drifting. Mathematicians call these hidden states, because the line never announces which game it's playing.

Textbook name: Parrondo's paradox, hidden Markov models.

3 · the hand-over

Water at 99°C looks exactly like water at 20°C. One degree later, everything changes. Physicists call it a phase transition, and near one, systems start to flicker: small nudges take longer to fade. Add people who are all guessing what everyone else will guess (Keynes called it a beauty contest), and a crowd can flip all at once. Do markets have boiling points? That's the question.

Textbook name: phase transitions, critical slowing down, Keynesian beauty contest.

4 · the structure

Take every number in a newspaper: prices, populations, river lengths. You'd expect the first digits, 1 to 9, to show up equally often. They don't. Almost a third start with 1, and fewer than one in twenty start with 9. It's called Benford's law, and auditors use it to catch faked accounts, because made-up numbers don't follow it. Real data has a fingerprint. The prize is finding the fingerprints that hold: patterns that survive every test, and threads between markets that only appear when you look at them together. The lattice behind these words is that dream, drawn before it's found.

Textbook name: Benford's law (first-digit law).

02the buildscaffolding goes up here

The workshop.

Every quant firm runs on its instruments. Here are ours: one running, the rest on the bench.

Direction, said openly This company is planned as a quant firm: market research done with maths, AI and predictive models, where a claim has to survive a test before it counts. Further out, the firm may run liquidity services or market making. Those are plans and a direction. Neither is offered today.

our market data engine

The Loom

Markets speak in millions of tiny messages a day. The Loom listens, checks every file it keeps, and weaves them into one clean, verified record. Built in-house, tested, and running since the day Lattice Chiti began.

For now it weaves a single instrument, deliberately, because our infrastructure only stretches so far. Every bit of capacity we add is another thread: more instruments, more markets, from stocks to crypto, until The Loom is weaving the whole picture.

built · tested · active

predictive models

Prism

A prism splits white light into hidden colours. Prism does it to markets: models that try to tell which hidden state a market is in, and what tends to come next.

in development

proprietary algorithms

Kernel

The house rules. Algorithms written in-house, and every one has to prove itself on simulated markets before it's allowed near a real one.

in development

our research system

The Trellis

A trellis is a lattice that helps things grow in the right direction. Ours is where researchers start. Most research dies in setup: missing data, broken tools, notes nobody can find. The Trellis is built so yours doesn't. Ideas go in at the bottom, and only the ones that survive every checkpoint climb to the top.

in development

our own neural network

Chiti

A network trained on what the Loom weaves, built to see structure that formulas miss.

upcoming

AI runs through all of it. It writes, checks and argues with us at every step, so a small team moves at a speed one person never could. Infrastructure is our only speed limit. Every bit we add turns into faster research and more threads for the Loom.

Why start with simulated data?

With real markets, we don't know the full story behind every change. Simulated data lets us test a method against situations we deliberately create, so we can see where it works and where it misleads us. Passing those checks still leaves the question of how well it works on real data.

Site notice

On site today

  • ✓ The Loom: weaving one market, tested and running since day one
  • ✓ Research on one market, fed by the Loom since day one, with AI in every step
  • ✓ Prism, Kernel and the Trellis, under construction
  • ✓ One founder, a lot of cold coffee, and room for researchers

Not on site yet

  • ○ A product you can use
  • ○ Chiti, our own neural network
  • ○ Market making and liquidity services (a milestone ahead, not an offer)
  • ○ Customers or a release date. When they arrive, this board will say so.

mind the gaps. they're real.

Seen enough scaffolding? The researcher desk is this way.

A note from Darshan

Hi. I'm Darshan, the one with the coffee upstairs.

Plenty of great firms started as one person at a desk with a question that wouldn't leave them alone. Mine is easy to ask and hard to answer: markets look like chaos, so where is the structure hiding, and how do you prove it's real?

So I started building. First the Loom, so the data would be clean. Then the tools, so every idea could be tested instead of believed. AI sits beside me through all of it, which means one person at one desk can move like a small team. Most nights, the coffee goes cold before the work does.

Here's the plan, said out loud: a quantitative firm built on maths, AI and research that has to earn every claim, then liquidity services and market making. Market making is a milestone, not the finish line. The real dream is a technology company: our own AI, our own infrastructure, our own research engine, one lattice where every piece holds up the others. Markets are where we'll prove it works. After that, structured intelligence goes wherever chaos is hiding a pattern, far beyond finance. Today it's scaffolding, and I'd rather show you honest scaffolding than a pretty fake building.

If you're reading this, you're early. That's the best time to show up: the drawings are still on the table, and your fingerprints can end up on the final structure.

And if you're building something of your own: start before you feel ready. I did.

Pull up a chair. Bring a hard question.

Darshan

written at the desk
you saw upstairs ↑

Researchers welcome

What would you like to investigate?

If you enjoy working with data, questioning a method, or looking closely at a result that shouldn't be there, I'd like to hear from you.

You might come from mathematics, statistics, programming, or somewhere else entirely. Tell me what interests you and how you'd like to contribute.

We're early. This is an invitation to talk about research together, not an advertised role with fixed terms.

The researcher desk three short questions

These are Darshan's questions, written in advance. Nobody is typing on the other side, and nothing you write leaves this page until you send it from your own mail app.

    03top tierdrawn, not built ↓

    Where this is going.

    Picture the finished lattice. The Loom weaving every market on Earth, from stocks to crypto. Prism and Chiti reading the hidden states inside them. The Trellis full of researchers, their ideas climbing toward proof. A top-tier quant house that makes markets, and a technology company that carries structured intelligence far beyond them. The bright structure behind these words is that dream, drawn in full.

    It isn't built, and none of it is on offer. We're near the ground, fixing the first beams. If you scroll back up you can watch how much is still pencil.

    Get early access

    Honest note: there is nothing to release yet, and no date. This button opens an email to Darshan saying you'd like to try things when they exist. That's all it does.

    Start a conversation.

    A question about the research? Write to Darshan.