An Investigation of Transactions Data for NYSE Stocks

Using transactions data, the behavior of returns and characteristics of trades at the micro level is examined. A minute-by-minute market return series is formed and tested for normality and autocorrelation. Evidence of differences in return distributions is found among overnight trades, trades during the first 30 minutes following the market opening, trades at the close, and trades during the remainder of the day. The latter distribution is found to be normal. Unusually high returns and standard deviations of returns are found at the beginning and the end of the trading day. When the beginningand end-of-the-day effects are omitted, autocorrelation in the market return series is reduced substantially. A number of patterns in trading are reported. THIS PAPER EMPIRICALLY INVESTIGATES the nature of the return-generating process and the characteristics of trade size, price changes, trading frequency, and trading interval at the level of individual trades for a large sample of NYSE stocks. Previous studies of market indices have used daily, weekly, or monthly