Something I wrote on all Futaba modes:

There are 3 types of latency. (for those who do not know what latency is, it is the time difference from changing control on the input to the reaction on the output)

1) A steady but small latency time is by the speed of electronics and processor handeling. This one is easy to get used by the mind because it is a steady factor. It could be measured. by modifying the transmitter to let it controlled by a computer and an electronic sensor on a channel and measure the time difference of a change.
2) A non steady and very unpredictable time by interference which is not easy to measure. You need to setup a device to simulate all kinds of interference. This is related to like high voltage cables under/over the track, the mirror signal from a fence but also other devices where the receiver need to correct or re-read the signal.
3) Frame rate.

Sensing all channels within the transmitter is like a train running circles and over the whole round all channels as like train stations are located. In the servo with a 1.5ms center pulse width the move will start after it reads and decoded the pulse width. It depends where the train is when you change a channel value. If you have an old system with a 20ms framerate the latency can be 1.5ms (if the train is excact on time) up to 20ms (when the train just drove by and need to make an whole 20ms round). On a RC track with fast cars taking a corner that can be a difference of getting the apex right, getting launched over the curb or going one meter too wide. The faster the framerate is set, the smaller this variable and so more steady this factor will become. Lowering the center pulse width is also a thing to lower the steady latency, that is why Sanwa its SSR went to 300uS pulse width but it is also needed when you want higher framerates beyond 333Hz.