ChartChest logo ChartChest

ChartChest Research · Working Paper IV · August 2026

Reading the Opponent: Institutional Playbooks and an Adversarial Chess Engine

Papers I–III were about our own models. This paper turns outward, to the opponent — two engines that treat market data not as a signal to be fit, but as authored behavior to be read: a script being run, and a game being played.

← Part I ← Part II ← Part III Part V: Reading the Move → Part VI: The Context We Preserved →
Papers I–III were about our own models: what edge is, how we try to falsify it, and how we judge whether to trust a prediction. This paper turns outward, to the opponent — two engines that read the tape as authored behavior: a script being run, and a game being played.

Abstract

Most quantitative systems model the market as a noisy function. We also model it as an author — a set of sophisticated actors executing recognizable campaigns and playing a multi-agent game against each other and against slower flow. This paper describes, at the architecture level, two engines built on that premise. The first, a playbook decoder, reads price and volume as a sequence of scenes and asks which recurring institutional campaign is being run, who its intended victim is, and how far along it is. The second, a chess engine, treats the tape as a game between desks — detecting coordination, separating bait from genuine intent, classifying the state of play, anticipating the next move, and asking two questions almost no system asks: is there a human behind the desk automation, and is our own model being poisoned?

The most important lesson here is a failure we then repaired: a catalog of named plays decays for exactly the reason Papers I–III predict — naming a play makes it manufacturable — and the durable fix was to stop asking only "which known play is this?" and start asking "is this a real play I cannot yet name?"

1. From Judging Predictions to Reading the Opponent

Paper III separated prediction from judgment. This paper adds a third posture: reading the actor on the other side of the trade. A price series is the visible residue of decisions — accumulation disguised as weakness, distribution disguised as strength, bait placed to be chased. If those decisions follow recognizable campaigns, the right representation is not a feature vector but a narrative: a script with scenes, roles, and an intended victim. And if multiple actors are maneuvering against each other, the right representation is a game.

The two engines below are those two representations. They are deliberately not predictors. They are interpreters — they convert raw tape into a description of what the opponent appears to be doing, which the rest of the system can then reason over and, per Paper III, decline to trust.

2. The Playbook Decoder: Reading the Tape as a Script

The first engine treats a chart as choreography. Institutions do not enter and exit at random; they run campaigns with a beginning, middle, and end — a quiet build, a period of applied pressure, a release, a hand-off. Each campaign has a structure a domain expert can name, an intended victim (who is meant to provide liquidity or get shaken out), and a counter (what a prepared observer should do instead).

The output is not buy or sell. It is a reading: which script, how confident, how complete, and whether the last move looked like a genuine development or a planted decoy. We do not disclose the campaign catalog, the scene vocabulary, or the grammars — those are the engine.

3. The Law That Broke Our Own Engine: Named Plays Decay

Here is the most useful thing in this paper, and it is a failure of our own. A decoder built on a fixed catalog of named plays is, by construction, a set of learnable templates — and Papers I–III predict exactly what happens to learnable templates: they get manufactured and they decay. We observed this directly. Over time, the named-play readings from this engine lost almost all of their value to the downstream models.

The reason was not a coding bug. It was adversarial choreography: sophisticated actors began running campaigns with non-canonical orderings — the same underlying intent, deliberately staged in sequences that no fixed grammar recognized. Our catalog was written against an earlier era's choreography, and the opponent had moved on. The instant you enumerate "here are the plays that matter," you have handed the opponent the list of appearances to avoid or to counterfeit. A static catalog of named patterns is a snapshot of a moving target.

4. The Fix That Generalizes: Absolute Fit and Novelty

The repair is more interesting than the break, because it generalizes to any pattern-recognition system operating against an adaptive opponent. Our decoder had a subtle, fatal design property: it only ever measured relative fit — which of the known scripts is the least-bad explanation of the tape. Forced to choose, it would always nominate something, even when the honest answer was "none of these." It had no notion of absolute fit — does any known script actually explain this at all? — and therefore no notion of novelty — this is clearly a deliberate campaign, but it matches nothing I know.

The consequence is the important part. The predictive value that had drained out of "which named play is this?" reappeared in "this is a real play I cannot yet name." Novelty turned out to be one of the most independent signals we have ever measured — it fires in precisely the places the rest of the system, and the rigid catalog, are silent. In the language of Paper I, the durable question is not the name of the play; it is the presence of authored intent that resists naming. Naming decays; the fact that something is being staged does not.

5. The Chess Engine: The Market as a Game Between Desks

The second engine abandons the single-actor script frame for a multi-agent game. Its premise is that the interesting behavior on the tape is not one campaign but several desks maneuvering — some coordinating, some baiting, some defending — and that the state of that game can be read.

Again, the engine emits a description of the opponent's posture, not a trade. It is an interpreter of intent, and it is wired as an overlay whose influence must be earned.

6. Two Questions Almost No System Asks

We describe these as questions, not as mechanisms. How each fingerprint or poisoning signature is computed is not disclosed.

7. Offense Gated by Understanding, Again

The chess engine has an explicitly offensive mode — the point at which reading the game becomes acting on it. It is built on the same rule Paper III made structural: understanding gates action. The engine spends most of its time observing, accumulating conviction silently, and it is permitted to strike only after the state of play is genuinely understood. An opportunity that arrives before understanding is treated as bait by construction. A reader of the opponent that acted on every apparent opening would simply be the easiest desk at the table to trap; patience is not a temperament here, it is an architectural constraint.

8. Earning Influence: Engines as Overlays, Not Oracles

Neither engine is trusted on arrival. Both are deployed as overlays that begin with zero weight — pure observation — and their influence over decisions is allowed to grow only as their live readings are scored against outcomes, under the same champion/challenger discipline described across the series. An interpreter of the opponent is itself a hypothesis about the opponent; it must earn its authority from its live track record rather than assert it from its cleverness. This is the flywheel of Paper I applied to the engines themselves: usage and elapsed time decide how much they are believed.

9. What Elevated the Models (Shape Only)

None of this claims the opponent has been solved. It claims that reading the tape as authored behavior, and continuously renaming what we read as the opponent adapts, produced durable, independent lift under the hardened discipline of Paper II.

10. What This Paper Deliberately Does Not Disclose

As throughout the series, we describe roles and shapes, not instruments. We do not disclose the campaign catalog, the scene vocabulary, or the sequence grammars; the absolute-fit and novelty computations; the chess engine's coordination, bait, game-state, or anticipation mechanics; the fingerprints that distinguish human from algorithm or the signatures that reveal model poisoning; the conditions that license the offensive mode; or how engine weights are learned from outcomes. The value of these engines is not that they read the opponent — it is how they read the opponent, and that is the moat.

11. Conclusion

A signal system sees a curve. These engines see an author and a game. The playbook decoder reads the tape as a campaign with scenes and a victim; the chess engine reads it as desks maneuvering, complete with the human hands and the poisoned inputs most systems never look for. But the deepest lesson is the one our own engine taught us by breaking: the moment you finish naming the opponent's plays, the opponent has already changed them. Durable edge is therefore not a better catalog. It is a system that notices when its catalog has gone stale — that can recognize authored intent it cannot yet name — and keeps renaming the game as fast as the opponent rewrites it.

Correspondence: ChartChest Research. Fourth working paper in a six-part series. This public page communicates the architecture of the playbook and chess engines while withholding every implementation detail that would enable direct reproduction.