With recent technological advances, high-frequency transaction-by-transaction data are widely available to investors and researchers. To explore the microstructure of variability of stock prices on transaction-level intra-day data and to dynamically study patterns of comovement over multiple trading days, we propose a multiple day time series biclustering algorithm (CC-MDTSB) that extends the time series biclustering algorithm (CC-TSB). For identifying biclusters within each trading day, our algorithm provides a faster alternative to the random replacement method in the CC-TSB algorithm. Moreover, our algorithm does not require prespecification of the number of biclusters for each trading day. Instead, we set a threshold on the number of stocks within the biclusters to yield an adaptive stopping criterion for multiple day analysis. An analysis of the biclusters determined over multiple trading days enables us to study the dynamic behaviour of stocks over time. We effectively estimate the comovement probability of each m-tuple of stocks conditional on the other stocks within the dynamic biclusters and propose a method to forecast comovement days using a nonparametric double exponential smoothing procedure.Step 4. Identify the next bicluster: Eliminate the stocks in A 4 from A 1 and repeat steps 2-3. Let A 5 denote the next identified bicluster.Step 5. Stop search for biclusters: Let`D 1, : : : , L denote the number of identified biclusters for a trading day. Let S`denote the number of stocks in the`th bicluster within the trading day. Let U denote the total number of stocks (out of the S stocks) that belong in these L biclusters. If U ġ, continue to repeat steps 1-4. Otherwise, stop identifying biclusters for that trading day. We repeat this process for each of the T trading days by repeating steps 1-5.Parameter selection. The user-defined threshold˛is the tolerance of insertion or deletion of the rows or columns. When˛is large (small), the algorithm tends to give large (small) biclusters. We have used˛D 1.2 in our analysis, for which the distributions of the number and sizes of the biclusters are provided in Section 6.1. Â, the threshold Stat 2018; 7: e176 Comovement probability. Our CC-MDTSB algorithm calculates the comovement probability for any m-tuple of stocks. Table I shows the top 15 pairs of stocks with the largest comovement probability, that is, the probability of trading days for which the two stocks are in the same bicluster (i.e., are comoving) in 2013. For example, the estimated comovement probability of two stocks in the (oil) energy sector, CVX (Chevron) and XOM (Exxon Mobil), Stat 2018; 7: e176 9 of 17The 1-minute averaged log returns incorporate the volatility in the stock prices. The volatility-adjusted log returns represent the risk-adjusted asset log returns relative to the amount of risk the investment has taken over a given period Stat 2018; 7: e176 (wileyonlinelibrary.com) https://doi.