Kinetic Alpha

A research and development practice for markets that are still being built.

New asset classes arrive in the wrong order. Futures get listed before the benchmark they settle on is governed. Indices get published before anyone measures the denominator. Credit gets extended against an asset nobody can mark. Kinetic Alpha works on the market structure underneath emerging assets and emerging technology in investing — classifying the asset, designing the instrument, auditing the benchmark, and working out what it costs to carry.

The output is research, and increasingly it is tooling: models, workbenches and decompositions built so the argument can actually be run against live data rather than read and admired.

What we do What you can run

The method

Six questions the work keeps returning to.

These are disciplines rather than topics, which is why the same six recur across compute, power, event contracts and digital assets. A new market usually needs all of them, in roughly this order.

Asset hierarchies

What is the thing, and how does it decompose?

Before an instrument can be designed on an asset, someone has to say what the asset is and how it behaves. Classification schemes that survive contact with the residuals, and decompositions that hold from index down to the individual risk factor.

e.g. Claim, Exposure, Regime

Exchange & contract mechanisms

What instrument should exist, and how should it settle?

Full instrument specification rather than commentary: settlement window, funding construction, delivery, margin treatment, and the no-arbitrage relationships that discipline a listed complex before it lists.

e.g. A Gold/Silver Ratio perpetual

Benchmark & index analysis

Who defines the number everything settles on?

Methodology review against IOSCO principles, denominator analysis for new asset classes, and the governance question that decides whether a published number can carry a contract at all.

e.g. Five Indices, One Price

Margin, clearing & collateral

What does it cost to carry, and what is the risk engine actually doing?

Margin frameworks reconstructed from published methodology and run on real histories — because the offset a clearing house grants is often a property of the model rather than of the markets.

e.g. Power, Compute, and the Margin Engine

Price formation & basis

Where does the price come from, and what does the spread between two of them mean?

Curve construction, cross-venue basis, and the spreads that only become visible once you hold two series to the same definition. Most of the findings on this site started as a basis nobody was quoting.

e.g. The Compute Crack Spread

Financing & credit structure

Who funds the asset, and how does paper become physical?

The capital stack behind a new asset class — who lends against it, on what collateral, and where the risk actually sits once the financing, not the instrument, becomes the binding constraint.

e.g. Where the Risk Goes

What you can run

The research generates instruments. The instruments get built.

Every workbench here came out of a specific piece of research and answers one question that could not be answered by reading. They are working tools rather than finished products — several are still being developed — but each runs on real data and each is open to anyone.

Energy complex decomposition

What am I actually exposed to across an energy book?

430+ risk factors across 470+ contracts on ICE, NYMEX-CME and Nodal, decomposed into dated factor legs with cross-exchange offset detection.

PowerGasCrudeMarginout of the hourly power work

Compute × Power workbench

What does a GPU-hour cost once you price its electricity properly?

Compute supply curve through to PJM, ERCOT, WECC and CAISO basis — spark spread, take-or-pay optimisation and trade synthesis across 14 modules.

ComputePowerBasisout of the crack-spread work

Perps × predictive margin

What does one portfolio-margin engine do with two unlike risks?

BTC and SPX perpetuals against event-contract strips on one underlying, with a 5,000-path Monte Carlo over an eight-cluster margin framework.

PerpsEvent marketsMonte Carloout of the margin work

Cross-venue divergence scanner

Where does the same question print at two different prices?

Kalshi against Polymarket against Manifold on live quotes, with the fee model that decides which gaps actually survive execution.

Event marketsArbitrageout of the cross-venue basis work

Portfolio allocator

How large should the position be, given what it consumes in margin?

Kelly and fractional-Kelly sizing solved against the margin framework, so size and margin consumption are decided together rather than in sequence.

SizingMarginout of the sleeve work

Tournament structure arbitrage

Do the outright and the match-path prices agree with each other?

Outright versus match-path dispersion, where the partition rule has real teeth — plus a best-of-seven series engine on the NBA Finals book.

Event marketsPartition ruleout of the correlation work

All tools, and how to read them

Where it gets applied

Four markets, and they keep meeting.

The method is the constant; these are the places it is currently pointed. They are not separate practices — the compute work runs on power prices, the event-contract work shares a margin engine with perpetuals, and the classification work underpins all of it.

Energy & Compute

The deepest body of work here. Compute is congealed electricity and almost none of it is priced that way yet — futures listed before benchmarks are governed, indices published before their denominators are measured.

Digital & Derivatives

Perpetuals came onshore and tokenized collateral reached the clearing house before the market-structure statute did. Both halves: the contract mechanics, and the settlement rail underneath them.

Predictive Markets

Event contracts scaled into a real asset class without growing a benchmark tier, a basis layer, or a quoted price for the correlation that combo flow already trades blind.

AI in Investing

Where the technology meets the mandate: an autonomy ladder for AI in wealth management, and a downloadable diligence tool for evaluating AI investment products against their own filings.

Working together

Research has to be applied, or it is just commentary.

The ways that happens are deliberately open-ended — a commissioned study, a benchmark methodology, an instrument specified ahead of listing, a tool built for one desk, or research co-sponsored with someone who needs the answer as much as we do. The method and the markets are the foundation; the delivery is whatever the problem wants.

Commissioned research

Venue and regulatory-pathway diligence, competitive teardowns of a benchmark or contract, and structural studies written to the same standard as the public work.

Index & benchmark design

Methodology design and IOSCO-principles review for a new asset class, including the denominator analysis most compute and energy indices currently skip.

Contract & product design

Full instrument specification ahead of listing — settlement, funding, delivery and margin treatment, with the pre-listing gap analysis a clearing house will ask for.

Tooling you can run

The workbenches above, built for a specific desk and its data, or a downloadable plugin like AI Tool Diligence. The research is the argument; the tool is where it gets used.

Start a conversation About the practice and the principal

The record