The playbook
What the large quantitative funds have made public about how they work, and what each lesson changed here.
Nobody outside these firms has their strategies, and anyone selling "the secret of" one of them is not telling the truth. What can be studied is the method. That study is the core of BaroqTech: every working day a piece of published research is read, turned into one testable idea for gold, and sent to the judge.
| Who | What is publicly known | What it changed here |
|---|---|---|
| Renaissance Technologies | Its Medallion fund is widely described as the best long-run record in the industry. By the published accounts it trades a very large number of small, short-lived patterns, is right barely more than half the time, and puts great effort into clean data and into measuring what each trade costs. | No strategy is judged by its win rate. Every test trade pays spread, commission, slippage and financing. No single strategy is allowed to carry the account. |
| AQR Capital Management | Publishes its research. A 2012 paper by Moskowitz, Ooi and Pedersen found that markets which rose over the past year tended to keep rising, across 58 futures markets, gold among them. | Tested here as "Daily momentum": turned down, too few trades to judge. A published result is a starting point, not a pass. |
| Man AHL, Winton | Decades of trend following on futures, with published work on sizing each position by how much the market has been moving lately. | Stops are set from gold's recent range, not from a fixed number of dollars. Sizing by recent movement is on the list to be tested. |
| Citadel, Millennium | Multi-strategy funds: many independent teams, each with a strict loss limit, and money moved quickly away from what stops working. Citadel describes five core strategies, commodities among them. | Each strategy has its own switch. There is a daily loss limit and an account floor. A strategy that stops earning on a live account goes back to demo. |
| D. E. Shaw, Two Sigma | Both describe their work as the scientific method applied to markets: a hypothesis first, then the data. | The reason for an idea is written down before it is tested, so a result cannot be explained after the fact. |
| Bailey and López de Prado; Harvey and Liu | Published tests for the oldest trap in the field: try enough ideas and one will look brilliant by accident. | The luck test in the judge is the one Bailey and López de Prado published. Every new idea tested raises the bar for all the others. |
BaroqTech has no connection with any firm or person named on this page, and does not have their strategies, their staff or their results. Nothing here suggests that software from BaroqTech will perform like any of these funds. They trade thousands of markets with hundreds of researchers; this is one market and a small set of rules.
What is read
Papers the firms and their researchers have published, books and long interviews about how they work, and academic studies of gold and of trading within the day. For example: "Time Series Momentum" (Moskowitz, Ooi and Pedersen), "The Deflated Sharpe Ratio" (Bailey and López de Prado), "Backtesting" (Harvey and Liu), "The Golden Dilemma" (Erb and Harvey), "Market Intraday Momentum" (Gao, Han, Li and Zhou), and Gregory Zuckerman's book on Renaissance, "The Man Who Solved the Market".
What happens to an idea
It is written down with its reason, tested on six years of gold prices, and judged by tests that cannot be adjusted afterwards. Then it goes on the record, pass or fail.