A simple stochastic gradient algorithm for adaptation of complex baseband weights of the transmit antenna at the base station by feedback from the mobile is presented. In order to converge to the optimal weights of the transmit antenna, the algorithm relies on two feedbacks from the mobile. The first feedback consist of a bit that indicates which of the two normalized perturbed weights sent along with the pilot signal from the base station delivered more power to the mobile. The second feedback consists of the difference of the powers delivered by the two normalized perturbed weights. The second feedback is sent only when the power difference decreases as compared to the previous minimum stored power difference. Utilizing these two feedbacks, our algorithm at the base station updates the complex base weights of the transmit antenna. It makes a coarse stochastic gradient algorithm and converges to those weights that maximize the power delivered to the receiver. This has been confirmed by simulations.