Overnight versus Intraday Returns on the Tokyo Stock Exchange

A close-to-close return is the sum of two different trades. The overnight piece is the gap from the previous close to the next open. The intraday piece is the open-to-close move. Lou, Polk, and Skouras showed that, in the US, those two pieces do not carry the same premia: a large part of the equity premium has sat in the overnight gap, and a large part of the reversal and other high-turnover effects has sat in the day session.

This page repeats the decomposition on a fixed panel of Tokyo large caps (autos, banks, technology, trading companies, pharmaceuticals). It is a property of the session, not a stock-picking backtest. The Nikkei 225 is the reference index in the header; the averages themselves are equal-weight across the names with a valid open and close that day.

Yahoo's adjusted close includes splits and dividends. The raw open does not. The open is rescaled by the same factor that takes the raw close into the adjusted close,

$${\tilde{O}}_t=O_t\cdot \frac{P_t^{\mathrm{adj}}}{P_t},r_t^{\mathrm{on}}=\frac{{\tilde{O}}_t}{P_{t-1}^{\mathrm{adj}}}-1,r_t^{\mathrm{day}}=\frac{P_t^{\mathrm{adj}}}{{\tilde{O}}_t}-1$$

Each day the cross-sectional mean is computed only if at least half the panel, and at least eight names, have a finite return. Session returns wider than 25% are dropped before that average. Those prints are split mismatches between the raw open and the adjusted close, not tradable gaps.

Inference is a Newey–West t-test with five lags on the daily mean: overnight equals zero, intraday equals zero, and the paired difference equals zero. Five lags cover a trading week of overlapping microstructure. The wealth chart also shows the overnight path after subtracting 1 bp per day, which is a financing sketch, not a claim that the open auction can be traded at that cost.

Four wealth paths are compounded from the daily equal-weight means: gross overnight, overnight net of 1 bp, open-to-close, and close-to-close. The annual table compounds the overnight and intraday means inside each calendar year.

The Tokyo cash open is an auction with a different liquidity schedule from the US open. A positive overnight mean is not evidence that a market-on-open order captures it.

Equal weight gives a small cap the same vote as Toyota. A cap-weighted version would sit closer to the Nikkei and would be a different measurement.

Yahoo Finance adjusted closes are not a point-in-time index membership file. The constituent lists are a liquid large-cap panel, fixed in the repository, not the official index as it stood on each rebalance date.

The risk-free rate is set to zero in every Sharpe ratio. A local bill rate would lower the ratios and would not, by itself, change a long-short spread.

Costs are a flat one-way haircut on weight turnover. They do not include borrow fees, lot-size rules, auction imbalance, stamp duty, or the Shanghai and Hong Kong short-sale bans that bind a live short book.

This is a research record. It is not a recommendation to trade the sleeve.