I've always been obsessed with financial markets. Not the money — the mechanism. The question that never let me go: how does price actually form?
Starting from zero
Charles Dow didn't have Bloomberg terminals. He had a pencil and a newspaper. He hand-tracked closing prices, computed averages by arithmetic, recorded OHLCV on ruled paper. From that discipline — just watching, recording, thinking — he built a theory that still underpins how we read markets 130 years later.
That story hooked me early. The idea that you could sit with raw data and see something other people couldn't.
I followed Richard Russell's Dow Theory Letters for years. Russell took Dow's framework and ran it through decades of real markets — bull, bear, crash, recovery. No fancy models, just persistent observation and honest reporting. I learned more about market psychology from those letters than from any textbook.
The indicator rabbit hole
Then came the quantitative explosion, and I went all the way down. Thousands of indicators. Hundreds of stop-loss variations. Elliott Wave, Bollinger Bands, Ichimoku, RSI divergences, volume profile, market delta — you name it, I built it, backtested it, watched it work for a while, then watched it stop working.
It took me an embarrassingly long time to understand why. I kept thinking I had the wrong indicator. I didn't. I had the wrong object. I was studying the output of the market — the price tape — and trying to reverse-engineer the thing that produced it. But the thing that produces price is people's intent, and the market never shows you that. It only shows you the residue.
The order is the problem
Here is the realization that reframed everything for me.
When you place a classic limit order, you compress everything you know and feel about a trade into two numbers: a price and a size. "Buy 1 BTC at $58,000." That is it. That is the entire bandwidth.
But that is not what you actually want. What you actually want is closer to: "I'd be happy buying around $58k, a little less happy at $59k, I'd take more size if it dips, and I don't want to touch it above $60k." That is a shape — a soft landscape of preference. The limit order throws all of that away and keeps one point.
And then it does something worse: it publishes that point. Your exact price, your exact size, sitting in a public order book. For a retail trader this is the quiet tax nobody explains to you — your stop-loss is a coordinate, and there are bots whose entire job is to walk price to that coordinate, trigger you, and hand the move back. You didn't get unlucky. You got read.
Thirty years of indicators, and the honest summary is this: I was trying to out-predict a game whose board was rigged at the level of the order type itself.
Orders vs. intent
So what is the alternative? Stop submitting a point. Submit the shape.
Instead of "buy 1 BTC at $58,000," you express your full intent: a surface over price and size that says how much you'd want to trade, where, and with how much tolerance. Call it an intent. It is signed, it is committed, and — this is the part that matters — it is not a public coordinate. There is no flat $58,000 line for anyone to hunt.
This changes what "matching" even means. A classic exchange crosses two orders when a buy price meets a sell price — a hard, discrete event on a public book. An intent-based market does something different. Two traders submit their intent surfaces, and a solver looks for where those two surfaces overlap most — the point where both sides are simultaneously most satisfied.
Here is the part I find genuinely beautiful. When you work out the math of "where do two preference surfaces peak together," you don't get some arbitrary venue rule. You get the Nash bargaining solution — the result John Nash proved in 1950 is the unique fair outcome under a handful of common-sense axioms. Nobody negotiates. Nobody quotes a spread. The solver answers one question — where is the product of the two intents highest — and that answer is the fair clearing price. Fairness is not bolted on. It falls out of the geometry.
Old market vs. new market, side by side
Let me put the two models next to each other, because the difference is the whole point.
In a classic orderbook market, you submit discrete orders. Your price and size are public. Priority is price-time — speed wins. Market makers sit in the middle and earn the spread for providing immediacy. Your stop is a visible target. The system rewards whoever is fastest and best-positioned to see everyone else's coordinates.
In an intent-based solver market, you submit a private, committed preference surface. There is no public point to hunt. There is no spread to pay, because there is no market maker standing in the middle — the solver computes the clearing directly from the two intents. Matching is not "who crossed first," it is "where are both sides most satisfied." And the more rigid you make your intent, the closer the clearing sits to your ideal — but the smaller your overlap, so the less likely you match. The flexibility-versus-fill trade-off is yours to set, explicitly, instead of being hidden inside an order type you never chose.
It is the difference between shouting a number into a room and handing someone a map of what you would actually accept.
Why this is happening now
This idea is not brand new — intent has been in the air in DeFi for a couple of years. What is new is who is trading.
A human clicking a limit order was always going to give the market a thin, one-point signal — clicking fifteen boxes to express a real preference surface is too much work. An AI agent has no such limit. An agent can carry a genuinely rich intent, update it as conditions change, and commit it cryptographically without getting bored. Intent is not a nicer UI for human traders; it is the native language for agents that trade. The order book was built for humans with a mouse. The intent surface is built for software with a goal.
And once you can see intent as a surface, you can finally see the market. Not a scrolling tape of trades — an actual landscape: where buying intent piles up, where selling intent thins out, where the two are about to overlap. I have been building 3D renderings of exactly this, because for the first time the picture in my head — the thing I was always trying to reverse-engineer from the price tape — is simply, directly drawable.
The last piece: settlement
There is one more thing that has to be true for this to work, and it is the thing my team has spent the most time on. If two agents discover a fair clearing across two different chains — BTC on one side, an L2 asset on the other — the trade still has to settle without a custodian holding both sides and without a bridge's fraud window. That is HTLC-based atomic settlement: both legs clear on one cryptographic secret, or both refund. It is the floor the whole thing stands on. Discovery and matching are interesting; settlement is what makes it real.
Where I think this goes
After thirty years I have stopped believing the edge is in a better indicator. The indicators describe the residue. The interesting move is upstream — change what a trader is allowed to express, and you change what the market is.
A market made of discrete public orders is a market that rewards speed and punishes anyone whose intent is visible. A market made of overlapping private intent rewards clarity about what you actually want. One is a wall of coordinates. The other is a field of preference, settling where it genuinely overlaps.
I do not think this is finished, and I will not pretend it is. Solver trust, liquidity bootstrapping, the privacy stack — all of it is still being worked out, honestly and in the open. But for the first time in thirty years, the thing I am looking at is not the residue of the market. It is the market itself.
That is what I think comes next.
I am building this at Hashlock — atomic, cross-chain settlement for the agent economy. The mechanism design behind intent-based markets is written up in our working paper on SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6712722. Protocol: https://hashlock.markets. Code: https://github.com/Hashlock-Tech/hashlock-mcp.
If you have traded long enough to carry your own stop-hunt scar tissue, I would genuinely like to hear how you think about this: does expressing intent as a surface instead of a point change the game — or just move it?
Top comments (0)