The CM entered the atmosphere blunt-end first, which became the hottest part during re-entry (see picture below left). So you can't place RCS thrusters on that part of the CM, as they would melt; you can only place them on the sides of the cone, where it stays cooler. Thus, the CM had no aft-facing ($-X$) RCS thrusters.
Note that the side of the CM toward the astronauts' feet ($+Z$) is also too hot for the RCS thrusters.
Also, the CM RCS was only required to rotate the attitude of the spacecraft, not provide translational control (unless used with the SM RCS). This removes the need to place RCS engines in all directions:
Initially, the CM RCS was intended to provide only a three-axis rotational capability. At approximately the same time that the requirement for SM reaction control deorbit capability was imposed, a CM RCS technique for translation was developed. This CM translation conferred a hybrid deorbit capability that involved the use of both CM RCS and SM RCS for total velocity increment ($\Delta v$) requirements.
Apollo Experience Report: Command and Service Module Reaction Control Systems, p. 2
Unlike the RCS engines used in Gemini/SM/LM/Shuttle, the CM RCS engines were designed to burn up while they were used:
A major difference between the CM RCS and SM RCS was the type of engines that were used. The SM engines were radiation-cooled, unlimited life engines (from a burn time standpoint). The CM engines were ablatively cooled, limited life engines and were used in a buried application. [...] The major difference between the CM and SM engines was the combustion chambers. Because the CM engines were buried in the CM skin, ablative chambers were used. [...] Although these changes represented a departure from the intent of using the same design on the Gemini spacecraft and on the Apollo CM, the improvement in product reliability and entry heating problems warranted the changes.
ibid, p. 9