Nov 2, 2026
•
12 min read
Oct 30, 2026
10 min read
Oct 25, 2026
11 min read
Oct 22, 2026
Oct 20, 2026
Oct 16, 2026
Oct 13, 2026
Oct 10, 2026
8 min read
Oct 8, 2026
Oct 6, 2026
9 min read
Nov 4, 2026
One number from each daily bar predicted the next day's winners for 25 years. I tested it before and after the paper, fixed its biggest flaw, and found the two things that can kill it.
The latest news, articles, and resources, sent to your inbox weekly.
Trading
Sep 3, 2025
4 min read
Algorithmic Trading Strategies
Sep 7, 2025
Sep 4, 2025
18 min read
Research Desk
A 2023 Oxford-Man paper trades each asset on its neighbours' trend. I rebuilt it on 21 ETFs: it beats classic trend following and trades half as much, but the placebo test tells a surprising story.
A 2025 paper adapts 40-year-old trend following to crypto. I tested it on Bitcoin and Ether, including the 18 months since, and tried to beat it with machine learning.
A 2023 Harvey paper fixes trend following's speed problem. I tested it on 100 years of US data and six markets, including the years since publication.
Trillions of dollars rebalance between stocks and bonds on dates everyone knows. A 2025 NBER paper says it costs investors $16 billion a year. I built the trade, tested it since 2002, and checked whether it still works.
Classic momentum lost 36% in one month in 2009. Strip out the market, manage the risk, and the same idea earns a much higher Sharpe ratio with less than half the drawdown.
