The Financial Services Agency's update on high-speed trading, published on September 30, 2026, shows that high-speed traders account for far more of Japan's order traffic than of its executed turnover. Registration has been required since the Financial Instruments and Exchange Act amendment took effect in April 2018.
Who is registered. There were 52 registered high-speed traders at the end of June 2026. January to June brought one new registration and two exits. The trading statistics below cover a wider group: registered high-speed traders plus financial instruments business operators that conduct high-speed trading.
Cash equities. The FSA's data cover 4,437 Tokyo Stock Exchange stocks. High-speed traders' share of trading value has mostly run in the high 30s per cent. The FSA gives the strategy split as roughly 15% market-making, 10% arbitrage, 40% directional and 35% other. The directional share has been falling since 2026 as total turnover rose.
The order-count and turnover shares differ sharply. High-speed traders trading through colocation, meaning servers placed next to the exchange's trading system, account for about 85% of order counts but about 45% of trading value. Those outside colocation are under 1% on both measures. By order type, high-speed traders account for about 80% of new orders and about 90% of cancellations, but about 40% of executions. In cash equities, the cancellation count excludes expired immediate-or-cancel orders.
Derivatives. The Osaka Exchange data cover 55 futures and options products. High-speed traders' share of trading value has run at 35% to 50%. Their share of new and cancelled orders is about 99%, and of executions about 65%. In this data the cancellation count includes expired fill-and-kill and fill-or-kill orders, so it is not comparable with the equity figure.
By product, the share has run at 60% to 65% in Nikkei 225 mini futures and 40% to 60% in Nikkei 225 futures. It has been 25% to 60% in TOPIX futures and 25% to 40% in long-term JGB futures. The FSA's figures are chart-based ranges, and its brief reports no price or execution-quality findings.
