Systematic proprietary trading desk

We do not guess
position sizes.
We compute them.

AN Trading Solutions Pvt. Ltd. runs a systematic F&O book across NSE and MCX. Sixteen strategies, weighted by a research process built to buy consistency with return — so the book survives the year it is wrong.

Strategies
16 across NSE & MCX
Research basis
5 years, 1-minute
Validation
Walk-forward, out-of-sample
Founded
2023
Base
New Delhi, India

01 — The book

A multi-strategy F&O book, sized by evidence.

We run a portfolio of systematic strategies across NSE index derivatives, single stocks and MCX commodities — futures and options, intraday and positional. Each strategy is independently researched, then sized as part of a whole.

The hard question on a desk like this is not what to trade. It is how much of each. Returns are abundant; consistency is not. Our allocation work exists to trade return for survivability, so that when one sleeve bleeds, another covers it.

01

Sixteen strategies

Trend, mean-reversion, option-buying and delta-neutral option-selling, spread across NSE and MCX so no single regime drives the book.

02

Minute-resolution research

Five years of 1-minute data, with exposure and margin modelled per minute rather than inferred from daily closes.

03

Weights, not hunches

A very large space of candidate allocations is evaluated on a full diagnostic battery, then filtered by robustness rules fixed in advance.

04

Walk-forward validated

Selection is judged out-of-sample. A book that only looks good on the data it was fitted to does not get deployed.

02 — What we trade

Four engines, deliberately uncorrelated.

The book is built so its parts compensate one another. Genuine negative correlation between the option-selling and directional sleeves is what makes the combined equity curve smoother than any strategy standing alone.

Directional core

NSE futures

Trend and moving-average systems on Nifty and BankNifty futures, run on intraday bars. The desk's longest-trusted sleeve.

  • 45MA — BankNifty
  • WAD — Nifty
  • TwoCandle + HalfTrend — Nifty
  • Directional Selling overlay

Commodities & stocks

MCX & cash

Trend systems on metals and energy, held at reduced weight because a five-year commodity bull market flatters long-biased results. Plus overnight stock positions.

  • WAD — Crude, Silver
  • HalfTrend — Gold, Silver, Crude
  • Supertrend 4H — Commodities
  • BTST + STBT — single stocks

Weights are reviewed as new out-of-sample data arrives. Sleeves that stop earning their risk budget are cut rather than carried — one commodity strategy was dropped outright for exactly this reason.

03 — How the book is chosen

Robustness rules, fixed before the search.

We do not run a mean-variance optimiser. Fed backtest-inflated inputs it becomes an error maximiser, allocating confidently to whichever strategy was luckiest.

Instead we sweep the whole feasible allocation space, then apply guardrails set in advance. Only a small fraction of candidates clear the full funnel — and the naive top performers are among the first to be rejected.

GuardrailWhat it enforces
Margin feasibilityJoint margin must fit the capital budget at every minute, with futures netted on signed position and options never netted.
Utilisation bandSustained utilisation is held well below the budget with headroom for a margin spike, rather than running the account hot to flatter returns.
Drawdown ceilingCandidates breaching the drawdown cap are discarded regardless of return.
Shock survivalThe book must survive a modelled global risk-off and still have margin left to keep trading. Several high-return candidates fail here and are rejected.
Loss compensationOn losing months, winning sleeves must offset the bulk of the damage — the property that makes the book a portfolio rather than a pile of bets.
Out-of-sampleWalk-forward folds on later years decide the ranking. In-sample performance alone is not treated as evidence.

04 — What the research showed

Findings we were willing to publish.

Results are backtested over a five-year window at 1-minute resolution, net of costs and with execution haircuts applied. Read the basis note below before drawing any conclusion from them.

  1. 01

    It holds up on crash days

    Across the worst single sessions in the study window, including the sharp June 2024 election-day fall, the book came through flat to positive. The option structures earn precisely when the directional sleeve hurts.

  2. 02

    It is regime-agnostic

    Average monthly result varies only modestly across low, mid and high volatility regimes — somewhat weaker in chop, not a collapse.

  3. 03

    Concentration was the real risk

    At equal weights, a single commodity sleeve accounted for well over half of portfolio variance. It was the largest contributor to both profit and risk, and its weight was cut hard.

  4. 04

    Live will be worse than backtest

    We plan around a harsh execution haircut — wins trimmed, losses widened — not around the headline simulated figure. The ranking of candidate books is unchanged under every haircut we tested.

Basis of figures. All results referenced on this site and in the Strategy Explorer are backtested on historical data, net of modelled costs, with execution haircuts applied. Backtested performance is hypothetical, carries the inherent limitations of simulated results, and does not represent actual trading. Known limitations include no deflation for multiple testing, and realised-only drawdown understating intra-trade mark-to-market swings on long-hold trend strategies. Past performance, whether simulated or actual, does not indicate future results.

Strategy explorer

See the book, strategy by strategy.

An interactive view of the desk's strategies — capital requirements, growth of capital, monthly and year-by-year P&L, and how a portfolio is assembled from them. Figures are backtested and shown net of costs with execution haircuts applied.

Open the Strategy Explorer

05 — Contact

Talk to the desk.

For allocation enquiries, research questions or a walk-through of the Strategy Explorer. Short and specific beats polished.