In a future time interval, an accurate velocity estimation model is the most important condition for the successful implementation of a control algorithm in a real system. In this study, Markov chain, which is a stochastic approach for future vehicle speed estimation, is discussed. The Markov chain is widely used for speed estimation over a short forecast horizon, especially in optimal control of autonomous and electric vehicles. Therefore, in this study, an algorithm is proposed that can improve the prediction accuracy efficiently. The proposed Markov chain-based algorithm is used in the dSPACE real-time simulation to estimate the velocities obtained from a real vehicle model used by different drivers. In the driving cycles, velocity is estimated along the 1 and 2 second prediction horizons, using the available velocity data as the model input. With the square of the mean squared error to evaluate the accuracy of the velocity estimate, the velocity estimation error of the 1st drive end is 1.3679 along the 1-second fore-horizon and the 2-second lead 1.8094 along the forecast horizon, the speed prediction error of the 2nd drive 1.8982 along the 1-second forecast horizon, and 2.7453 along the 2-second forecast horizon, and the 3rd lap The velocity ¨ estimation error of three is obtained as 0.6556 along the 1-second forecast horizon and 1.1634 along the 2-second forecast horizon.