HALFPLANE — research (v2a + parallax + breathing)

01 Architecture_of_Understanding

Architecture of Understanding

We are three operators.

We do not read charts. We do not weigh macro. We do not interpret sentiment.

We treat the market as a geometric object — a high-dimensional surface with curvature, boundaries, and a small number of invariants worth knowing. Our work is to find the places where that geometry is most legible.

We do not guess. We isolate.

02 The_Method

The Method

A market, read carefully, encodes its own forward distribution of beliefs. When that distribution is misshaped — which it is, with measurable regularity — there is structure to extract. We extract it mathematically.

Composition No single model carries the system. A small ensemble of specialised learners, each responsible for a region the others would distort. Geometry We do not predict prices. We classify regions. The boundary between regions is where the math becomes the strategy. Asymmetry What reads as noise has shape. The shape is not random. We map it, then we trade against the parts of it that other participants are forced to ignore. Scale The work spans more than one timescale. Different learners answer different questions, composed into a single coherent view.

03 The_Constraints

The Constraints

The strategy is capacity-constrained. We run with a hard ceiling. This is not a product — it is a research position.

Ceiling Hard capacity limit, set before the strategy degrades — not after. The discipline is in knowing when edge is being extracted versus when it is being diluted. We close before we dilute. Correlation Pure alpha. Uncorrelated to long equity, to volatility carry, to any macro regime we have measured against. The source of return is structural, not exposure. Research We publish what we learn. Research logs, model notes, and market microstructure breakdowns distributed bi-weekly to a closed list. No marketing material. Access Inbound only. If we are not the right structure for your mandate, we will say so.

VENUES NSE · BSE

root@halfplane:~/archive_access

bi-weekly · encrypted · no third-party

The Open Kitchen

We publish bi-weekly research logs, non-linear machine-learning papers, and quantitative breakdowns of Indian market microstructure anomalies. No marketing. No funnels. The lab notebook, routed directly.

NO_TRACKING · NO_LINKEDIN · NO_FUNNEL awaiting input

[ FEED INITIALIZED // LOGS ROUTED · feed_id 0xA47C ]