Inside the Black Box

Narang, who has worked in and allocated to quantitative funds, wrote a structural explanation of how systematic trading strategies are built, aimed at investors and allocators rather than at quants building them.

The organising contribution is decomposition. He argues that any quantitative trading system consists of identifiable modules, and that understanding a fund means understanding each. The alpha model generates the forecast and can be theory-driven, using price data through trend and mean reversion or fundamental data through value, growth, quality and sentiment, or data-driven, using statistical relationships without a prior theory. The risk model constrains exposures, either by limiting them or by penalising them in optimisation. The transaction cost model estimates what trading will cost, including commissions, spread and market impact, and Narang stresses that this module determines whether a signal is tradeable at all. The portfolio construction model combines the forecast, the risk constraints and the cost estimates into target positions, using optimisation or rules. The execution model implements the trades, choosing algorithms and venues. Data and research infrastructure sit underneath all of it.

He then covers high-frequency trading separately, explaining market making, arbitrage and the structural role these firms play, and addresses the arguments about their effect on markets.

The diligence chapters are the practical payoff: what an allocator should ask about each module, what answers indicate a problem, and how to think about capacity, crowding and the risk that many funds hold the same positions, which he illustrates with the August 2007 quant deleveraging.

The book explains structure rather than teaching construction, so a practitioner will find it introductory. It also predates the widespread application of machine learning methods, alternative data and the substantial changes in market structure and capacity since 2009. Later editions update parts of this.