Short answer
If your only goal is speed, you should not consider classical strength training, as it will not improve your speed. That is, it will, but only initially and marginally. Although both kinds of training do train fast-twitch fibres, it is fast movement against resistance that trains speed performance, and slower movement with higher weight/intensity that trains strength performance. They are two different kinds of strength trained differently. Since your goal is to reach maximum velocity with low resistance, this is what you should train like (Specific Adaption of Imposed Demands, or SAID principle). What classical strength training may do for you, though, is improving the circumference of your muscle and allow for it to hold more energy reserves, hence if your problem is fatigue over longer periods of time, go for it. Also, it will eventually allow you to use exercises and weights for speed training you previously just couldn't do fast enough.
Long answer
Two kinds of strength
Fast-twitch muscle fibres are made for explosive tension. This can, depending on resistance, take two forms. As this article explains (including corresponding research), this translates to two different kinds of strength trained differently (here, using benchpress as example):
There have been a large number of people lately who have been using the terms "strength-speed" and "speed-strength" interchangeably. Unfortunately, this is incorrect. At its base, strength-speed means strength in conditions of speed. Speed-strength, on the other hand, means speed in conditions of strength (Ajan, 1988; Roman, 1986). What this essentially means is that strength-speed means that you move a heavy(er) weight as fast as you can. Typically, this is around 60% of a 1RM, and the bar moves at a specific velocity of .8-1.0m/s. In turn, speed-strength essentially means that you are trying to move as fast as you can, but your are moving a light(er) weight. Typically, this is around 25-40% of a 1RM, and the bar moves at 1.1-1.5m/s. (bolded mine)
The point is that the fast-twitch muscle fibres do, in a sense, try to do the same in both cases: move as fast as possible against a given resistance. Strength is strength is strength, isn't it? Yes, to an extent. But depending on resistance, the fibres have to coordinate and work differently. That is why (maximum) strength training should be different from speed training.
How to train maximum speed, specifically
As the article quotes from a paper (Jidovtseff, Quièvre, Hanon, & Crielaard, 2009), if your goal is to raise your maximum speed, you should train with "25-54% of [your] 1RM as fast as possible, which [should lead] to velocities of about 1.4 m/s to about 1.0 m/s". RM stands for Repetition Maximum, ie. the maximum weight (or resistance, or, more generally, intensity) with which you are barely able to perform one clean repetition. It does not matter which exercise we are talking about. As the article points out, the same figures have been found for squats. Your main takeaway should be that higher weights and/or slower execution do not result in improvements of maximum (!) speed performance using the very same exercise (if you already have some training).
Why use strength training at all, then?
Strength training will, over time, let you move higher weights faster, even if it does not directly improve your maximum speed. Thus, what it will eventually do is that it allows you to use weights and exercises that used to be too hard for you to do with velocities of 1 m/s and above in effective maximum speed training. Much less efficient than going for maximum speed training with appropriate resistance and velocity right from the start though.
For example, if you have limited means to generate resistance in terms of material, say given weights of medicine balls, given dumbbell weights, your own body weight, etc., it might be necessary to do maximum strength training first to enable you to use what you got in effective speed training later. Like if you want to use body weight (push-ups) but can't execute the exercise fast enough yet. Using appropriate resistance and velocities is to be preferred where possible.
A punch is not a benchpress
Additionally, punch speed (especially for cross punches) is not only a matter of arm and shoulder strength, but is based on a kinetic chain which involves feet, legs, hips, core stability, and torso rotation. As this article suggests, you will need to do exercises which train the specifics of all of the movement, which mainly involves: punching. A lot. Supplemented with high-speed exercises with the loads mentioned above. I would take this article with a pinch of salt as it lacks the insight mentioned above and thus bases its program on incomplete information. Using medicine balls with a specific weight as resistance in biomechanically very similar push-throws and distance thrown as speed parameter is a very sound suggestion though.
Some remarks on 'benchmarking' vs. individual training regimen
It does not make sense to base suggestions on body weight if not as rule of thumb for goals to reach for a basic fitness when it comes to maximum strength. If you do not use your body weight in speed exercises, it is irrelevant for your speed training. A training regimen should be all about your individual physiological conditions. That helps you most and fastest. A functional test should give you your personal repetition maximum per exercise, which you should evaluate on a regular basis (at least monthly). Knowing that, you can plan your exercise intensity (weights, resistance via rubber bands, angles if you use body weight, etc.). Knowing them, do as many reps per set as you can perform within the exercise parameters (clean, fast enough), 3-5 sets each. Once again:
Jidovtseff et al found that there were two separate and more explosive strengths on the bench press that they referred to as "power-velocity" [aka speed] and "strength-power" [aka (max) strength] (Jidovtseff, Quièvre, Hanon, & Crielaard, 2009). Interestingly enough, they found that power-velocity occurred when moving 25-54% of their 1RM as fast as possible, which led to velocities of about 1.4 m/s to about 1.0 m/s. Also, strength-power was found to be developed from 54-82% of their 1RM when moved as fast as possible, and it related to velocities of .8 m/s down to about .7 m/s. (from the first article)
This was known in Soviet sports science since the 80s, btw.