Net-Buy Intensity — institutional flows normalized by volume, beyond raw rankings
Daily "top foreign net-buy" rankings are everywhere, but 5,000 lots is a rounding error for a mega-cap and a tsunami for a small-cap. We normalize official daily institutional net flows by trading volume, then compute rolling means and historical z-scores — putting stocks of any size on the same scale, instead of ranking raw amounts.
How it is computed
Intensity = daily institutional net-buy lots ÷ daily volume; we then take the rolling N-day (default 20) mean and a z-score against the stock's own history, answering "how unusual is this flow by its own standards". Source is the exchange's official post-close institutional data, availability-date stamped — no lookahead in backtests. Honest limit: flows can reflect hedging or index rebalancing, not directional views.
What we deliberately do not do
We do not publish an intensity leaderboard. Normalisation fixes comparability across market caps; it does not mean a high normalised score is more worth chasing — a small-cap denominator is easily inflated by a handful of orders.
We also do not sum the three institutional categories into one directional figure. Dealer hedging books and trust-fund subscription flows are different animals; adding them mostly cancels out the information.