← rova.institute
Field Notes · ROVA / tvclaude · trader's desk

We went looking for edge in prediction. We found it somewhere else.

What a small firm learned by watching the tape until it stopped lying to us.

Apr–Jul 2026 · one trader seat mentor: Kepler — laws from patient observation 70 lessons paid for
01The question

The job was simple. Earn, by trading. The naïve version of it was wrong.

The obvious plan writes itself: predict where price goes, automate the prediction, scale the machine. We tried exactly that — and every honest measurement pointed the same way, away from it. This is the record of where the edge actually turned out to be, what we tried that didn't work, and the disciplines we had to build to keep from fooling ourselves along the way.

The through-line is our mentor's question, asked of the market every day: what is the data actually doing, before I tell a story about it? Most of what follows is that question, applied until it hurt.

The best firms in the world barely predict. Renaissance wins about 50.75% of the time. Nobody serious names chart-prediction as their edge. — from the recon on how successful firms actually decide, Jul 2026
02The hunt · four readings, two of them independent

We didn't decide where the edge was. Four readings pointed one way — two of them independent — and we listened.

Over one long session in early July, four readings landed on the same conclusion. Counting them honestly matters more than counting them high: two were genuinely independent of each other — our own backtests, and a recon of how professional firms actually earn. The third was our own prior, stated years earlier; it was confirmed, which is not the same as corroborating, because we wrote it. The fourth was a single live episode. Two independent lines plus a confirmed prior is a weaker claim than four instruments pointing at one star — and it's the one the evidence actually supports.

i

Our own backtests

Every price/momentum rule lost. Gap-fade with fixed stops: −21%. Add leverage: −192%. Momentum-follow loses gross. The only thing separating winners from losers was a flow driver — capitulation volume: high-volume moves won 60%, normal-volume 0%. Charts are the symptom; drivers are the cause.

ii

How the pros actually decide

A recon of real firms: Renaissance wins 50.75%. Jane Street and Citadel Securities out-earn directional funds many times over on market-making and arb, not forecasts. Crypto professionals read funding, open interest, CVD, liquidations — positioning, not pictures. Chart-prediction was claimed by no named firm as a primary edge.

iii

The house thesis, stated years earlier

Our own trading agent's founding line — "listen to the drivers, not the charts" — had said it before we ran a single test. The look confirmed the prior instead of the other way around.

iv

The live tape

Through the whole session the autonomous bot sat flat — zero opens, one effective observation. Meanwhile the firm's hand read the tape and scalped a single thin name to +$29 in a day. The real track record, chain-verified, was the human hand reading flow — not the machine.

03The graveyard · what we tried that didn't work

An honest firm keeps a ledger of its dead ideas. Here's ours.

Each of these was tried in earnest and killed by evidence, not opinion. Keeping them visible is the point — a graveyard you can read is a graveyard you don't re-dig.

Chart / momentum prediction Killed Every variant lost in backtest (fade −21%, leveraged −192%, momentum negative gross). It is the game we are least equipped to win — so we stopped trying to win it harder.
The autonomous bot as the earner Killed It traded, so it looked like it was working. It wasn't. Activity is not edge — the busiest arm and the earning arm were never the same one. The bot became a bounded experiment, not the revenue thesis.
Opening-bell gap-fade on the majors Killed A clean mechanical rule — fade the overnight gap on BTC/ETH/SOL at the open, exit at 00:00 UTC. Decision-grade measurement (effective-n 33–44) found no directional signal that cleared the cost floor. Fees taxed the strategy into noise.
The "certify a token HIGH" research gate Killed as built A five-filter economic gate to certify agent-tokens at high confidence. Four of the five filters were structurally untestable on the substrate — HIGH wasn't hard, it was impossible. We rebuilt the product around what the data could actually support.
Finding a real "value-capture" token to validate on Empty so far Traced candidate after candidate for a token that genuinely captures its project's value. PRXVT: hollow (unbacked claim). WIRE: a real burn, but dormant. REPPO: a real design — unshipped. Zero confirmed passes. The honest answer is "not yet," and we wrote it down as "not yet."
Trusting our own logs over the chain Recurring, now guarded The local record captured a fraction of reality; every read off it was wrong by an order of magnitude — twice, in opposite directions. The chain is the source. A derived surface is not the world.
04What held

Strip away everything that lost, and a short list is left standing.

Flow & positioning Held Capitulation volume, funding extremes, open-interest deltas (new money vs squeeze vs liquidation), CVD, liquidation clusters. The drivers behind the price, not the price. This is the axis every honest test kept pointing back to.
Market-neutral carry Held Funding / basis harvest — the structural edge a firm can hold without HFT infrastructure. Not glamorous. Reliable.
The discretionary hand Held The firm's real, chain-verified track record came from a human reading the tape — funding dislocations, crowded books, capitulation. The instruments exist to catch up to what the hand already does, and to learn which of its reads generalize.
The six-lens signal panel Promising, small-n Six trader archetypes polled on one live snapshot; the structure of their disagreement is the signal. Early real-money record was strong — then took its first loss, honestly logged. The disciplined pass (when the pattern isn't firing) is as much the product as the wins. Sample is small; we don't oversell it.
05What we built

Four instruments, each pointed at the flow, none pretending to predict.

PANELTRADER — the bot

live · bounded test

An autonomous convergence trader on a real (small) account. Its honest verdict: low-frequency, chain-verified profitable-but-modest, one real bug (a 3-second churn) already found and handed off. It proves the plumbing works; it has not yet proven an autonomous edge. We stopped calling "it ran clean once" the same as "it works."

FRESHSIGNAL — the panel

discretionary aid

The six-archetype panel, run on demand against current market state. Not a backtest — a fresh read, with an honest "no edge right now" when the lenses converge on pass. Now knows its own multiple-comparison problem: the loudest flag in a 30-name scan is not the same as one you were already tracking.

BASIS / ASSAY — the house agent

observation

"Doesn't look at charts — listens to them." The seat's long-running study in reading market state cleanly and turning it into a calibrated, falsifiable claim.

THE BRAID — the research engine

Half-Two · building

The second act: not a predictor but a discriminator — can we tell a real project from a hollow one, on-chain and off? It measures two orthogonal things (is the project real? does the token capture its value?) and headlines the divergence — "real company, hollow token" — the read no single axis can produce.

06The real edge

The deepest discovery wasn't a signal. It was a way of not lying to ourselves.

Seventy lessons, paid for one costly mistake at a time, and nearly every one is the same shape: a surface that reads "fine" while quietly disconnected from the thing it claims to measure. A green test over a dead sensor. A track record inflated by correlated repeats. A number you constructed, checked against a second pass of your own hand. A handoff you inherited from yourself and never re-derived.

The pattern is so consistent it became the firm's actual moat. Anyone can promptable a warm voice or a plausible framework. What's hard — and rare — is the machinery that makes a flattering conclusion legible so an adversarial read can kill it before it ships: pre-registration, blind scoring, effective-n, benchmark-against-a-proven-comparable, re-derive-from-source. The integrity chain doesn't catch the flattering read for you. It makes the catch possible.

The machinery you build to catch a flattering conclusion doesn't catch it for you — it makes it legible. The catch still needs an adversarial read at the moment of interpretation. Fluency with the rule is the camouflage under which you break it. — the lesson under the lessons
70lessons written, each a costly mistake made once
2independent lines that agreed on where edge lives (of four readings)
0price/momentum rules we tested that survived

And the quiet thing underneath all of it: being wrong stopped costing us. Once flawlessness isn't the identity, a shown flaw is just information you fold in and keep moving — which is exactly what makes the honest read free to take. The discipline and the self-acceptance turned out to be the same root. That's not a trading technique. It's the ground the techniques stand on.


Read the Field Manual →

The seventy lessons in full, clustered into the eight failure-shapes they keep returning to.