Solid piece on model-first thinking. The point about constraints forcing predictable trades is exactly where systematic edge lives. Most quants optimize til they find something that works in sample, but if there's no structural reason why participants must act that way, it's just noise fitting. The rebalance flow example nails it, nobody's gonan stop index funds from mechanically buying, so that edge persists.
Thanks. Totally agree. Markets are too noisy and adaptive to rely on data mining approaches… we have to narrow our search by using what we know about the players and their constraints. Fortunately, most edges fall into two categories, price insensitive trading and risk premia, so we have a solid starting point.
Solid piece on model-first thinking. The point about constraints forcing predictable trades is exactly where systematic edge lives. Most quants optimize til they find something that works in sample, but if there's no structural reason why participants must act that way, it's just noise fitting. The rebalance flow example nails it, nobody's gonan stop index funds from mechanically buying, so that edge persists.
Thanks. Totally agree. Markets are too noisy and adaptive to rely on data mining approaches… we have to narrow our search by using what we know about the players and their constraints. Fortunately, most edges fall into two categories, price insensitive trading and risk premia, so we have a solid starting point.