Neo trading · AI-native

OUR TRADING PHILOSOPHY

Intelligence should
compound.

Our vision: the ultimate automated trading business, where human expertise and machine intelligence compound together.

Sankhe India Ventures 4 min read

AI is changing the cost of intelligence and how much of it a small team can put to work. In markets, that intelligence has economic leverage: a better decision can be applied repeatedly to capital. We are building a firm around this relationship, connecting expert judgment, specialised agents, and rigorous feedback across the entire trading process.

01 Where better decisions have economic leverage.

Markets bring decisions into direct contact with consequences. Better risk selection, more precise execution, or recognising when conditions have changed can affect how capital performs. That makes trading a demanding place to test whether additional intelligence creates economic value.

Our starting point is Indian index options. Around the same underlying market, different strikes and expiries present different prices for risk. The questions are precise: which risk is worth taking, at what price, for how long, and under what conditions should we step aside?

A single outcome is noisy. Repeated, well-recorded decisions give us something we can examine, challenge, and improve.

02 Intelligence cost compression.

The cost of intelligence shapes what a trading firm can afford to investigate. AI-driven cost compression expands that frontier: a lean team can explore more hypotheses, organise more evidence, and monitor more conditions. Research that was expensive to repeat can become part of the daily operating process.

We think in terms of return per intelligence token: how much useful decision-making each unit of AI reasoning enables relative to its cost. Does it uncover a weak assumption, improve execution, or help us avoid poorly compensated risk? Sometimes its most valuable contribution is a decision to do nothing.

Our thesis is that a trading edge comes from the combination of specialised intelligence, quality data, technology, human judgment, and disciplined execution. The advantage lies in how well these elements work together.

03 Give judgment a structure.

Exceptional traders accumulate judgment that is difficult to capture in a rulebook. Our architecture is designed to draw on insights from high-quality traders: the situations they recognise, the exceptions they watch for, and their reasoning for acting or staying out.

AI can help translate that expertise into explicit hypotheses, decision frameworks, and testable strategies. Its reasoning capabilities can also help weigh evidence, compare competing explanations, and propose responses to unfamiliar situations. We want to make expert judgment easier to examine, apply, and refine, with evaluation determining which insights earn a place in the system.

The architecture brings together rules-based strategies, strategies informed by AI judgment, and specialised trading agents. Each needs a clear job and a way to evaluate its decisions. Prices, exposures, and risk limits remain grounded in dependable calculations and explicit controls.

04 Keep a record of the decision.

A profit can follow a poor decision. A sound decision can lose money. Learning requires us to preserve what was known at the time, why a trade qualified, how it was managed, and what happened along the way.

We are designing our data and research workflow to retain that richer record, including opportunities we deliberately reject. Without those, we only study the choices we happened to make.

This gives each experiment a longer life. A rejected hypothesis can sharpen the next test. A live outcome can expose an unrealistic backtest assumption. A difficult session can reveal where a rule needs work. Improvements should enter the system through evaluation, with a traceable reason for the change.

05 Risk and execution are part of the idea.

A strategy has to work at the prices we can actually trade. Costs, slippage, liquidity, and the path of a position all belong in its evaluation. The return on paper is only the beginning of that work.

Capital preservation shapes the design: bounded exposure, clear conditions for pausing, and a deliberate progression from research to simulation to live deployment. Faster automation makes these controls more important. Human accountability remains part of the system as autonomy increases.

Staying selective preserves both capital and the ability to act when a better opportunity arrives.

06 Build a firm that can keep improving.

We approach trading with the discipline of technology-led operations: turning complex work into clear responsibilities, reliable workflows, and measurable feedback. That discipline connects the trading process, from data and research to execution and review.

The long-term direction is a lean, increasingly automated firm whose research capacity can grow through software and specialised agents. Its accumulated knowledge should remain useful beyond any one strategy, model, or individual.

That is the compounding we are working towards: capital managed with discipline, and a business that becomes better at deciding how to use it.