If your system collection goes back as far as the 2600, which mine does, and if you remember the days where they were called “joysticks” and not “fight sticks”, and you wnat one joyustick to play everything from the 2660 to the Switch, or most systems, you need PCBs to pad hack to convert Directional and button presses into “joystick code”.
But some systems are so simple, that I heard you can use a “straight-line” connection. The Atari 2600 is the most popular example. Other ones I can think of (if I understand the 9-pin wiring charts right) that are not 2600 stick compatible are the Sega Master System, the Bally Astrocade, and MAYBE the 2600 Booster Grip (the joystick handle which taps the paddles for 2 extra digital buttons, packaged with 2600 Omega Race, depends how easy it it to tap the paddles as buttons. Again, a wiring noob 
that’s why I’m hiring people. )
How more early systems have joystick code
Intellivision uses 6 of the 9 pins as binary code to represent 64 possible states. Colecovision has variable voltage fire buttons to encode 2 fire buttons (for the standard controller, I don’t understand the Super Action coding), Vectrex and 5200 have analog sticks so to work with a fight stick, you need analog digitizations, which I found Mathew Gummo can do (at least for a PS1 Dual Shock 1, and I know a way to do it with a 5200.). 7800 has a weird grounding protocol to make it compatible with both old 2600 and new 7800 games. I don’t know if an NES or a TurboGrafx 16 is considered a straight wired-system or not. (TG 16 Street Fighter 6 button needs a PCB) I believe all other systems (based on what I read) need a pad-hacked PCB to work right on a fight stick, (unless you want to go thorough the computer science of the respective joystick languages, which is how console controller converters work.)