Assume one wants to forecast/trade over T bars, which we call the T-forecast. One could also do the T-forecast in two T/2-forecast steps, a T/2-forecast done immediately, and a second one done once T/2-bars worth of time have passed,and the prices for the first T/2 bars have been revealed. Which of the two approaches do you like, and why? Do you see any incoherence/inconsistence in your approach when compared to the alternative approach? Are there bounds on T below which or above which one should not trade/forecast, and why?