Researching Market Edges Through Economic Mechanisms, Not Rule Mining
Summary
This essay contrasts searching large numbers of trading rules with research that begins from a proposed market mechanism. It asks researchers to identify who takes the other side of a profitable trade, why that participant accepts the cost, and what constraints may keep the behavior recurring. Examples include portfolio rebalancing flows near month-end and leveraged perpetual-futures traders paying funding to obtain exposure. Backtests are framed as tools for evaluating implementations of an hypothesized edge, not as explanations of why returns might persist.
The author warns that broad searches, including AI-assisted searches, can find attractive historical patterns without establishing a durable source of returns. Multiple-testing corrections can address statistical false discoveries, but do not show that a pattern has a continuing economic cause. The proposed alternative is an iterative cycle of market observation, hypothesis formation, and data analysis that builds understanding over time. This is an argument about research practice, not a tested performance comparison; the examples are illustrative, and the piece does not quantify how often such mechanisms produce tradable profits.
Key ideas
- A research hypothesis should explain who pays for an edge and why that behavior may continue.
- Backtests evaluate rule implementations but do not by themselves reveal an edge’s economic mechanism.
- Searching many rules can uncover historical patterns that have no durable explanation.
- Statistical corrections address some false discoveries but cannot establish why returns should persist.
- Repeated market observation and hypothesis testing can build research intuition over time.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.