Aiming to prolong the duration of detection equipment deployed in deep water, a new type of hydrokinetic turbine with good self-starting ability should be developed to harness low-speed current energy in such deep-water scenarios. In this study, a novel ductless Archimedes screw turbine is proposed to improve the system’s startup performance for low-speed current applications. An experimentally verified numerical method was used to investigate the parametric sensitivity of several key geometrical parameters. Strong interaction effects between the lead angle and number of turns were observed. The turbine performance enhancement by increasing the number of turns can only be achieved at a large lead angle. A tradeoff between the lead angle and number of turns is necessary for the optimization design of the ductless Archimedes screw turbine. Multi-objective optimization was performed on the ductless Archimedes screw turbine to improve its self-starting ability and power coefficient by using the central composite design, the radial basis function neural network, and the non-dominated sorting genetic algorithm. Water flume experiment results showed that the maximum power coefficient and static torque coefficient of the optimized ductless Archimedes screw turbine can be improved by 36.7% and 143%, respectively, with respect to the initial design. These results indicate ductless Archimedes screw turbines are suitable for low-speed current applications in deep water.