Individualization and the soul of training approaches
I’m happy that the current paradigm in powerlifting coaching is tending toward individualization and athlete-centric coaching versus previous models where the athlete had to fit the coaching system. In hindsight, looking back at the ways many coaches operated across sports, it now seems foolish to think that athletes will get their best performances when under the tutelage of specific coaching styles. Sure, coaches have optimized their approaches based on years of experience, but the gap still exists that athletes vary based on genetic factors like limb lengths, responses to volume, responses to intensity, frequency, exercise selection, level of arousal, and so on. Instead, we know that a good coach changes their training based on athlete response, over and over again, until the athlete is both objectively progressing well and internally feeling motivated and engaged with their training process.
In theory, the coach who is able to infinitely adjust their training parameters to match the athlete is the dream, perhaps guided by athlete training response and a rigorous system of trial and error like Emerging Strategies, or perhaps by a similar system with slightly more weekly variation that many other coaches employ.
The problem is that when we look in practice, it’s very clear that there are bounds to how varied any coaches’ training gets. First, I’d like to define a concept. The training space is the near-infinite range of all possible training plans. Every combination of every exercise you can and can’t think of, every partial or full range of motion, every tempo, every rep range, frequency, microcycle length, weekly progression, number of exercises per session, number of sessions in a day, every ____ (some undefined training variable I don’t even know). It all goes in there.
There are limits to the space of all possible training solutions and I think we need to reframe in subtle ways how we talk about individualization, or be honest with ourselves about the limits of individualization. As I’ll discuss, perhaps this has to do with the coaches’ own psychology, in inherent biases that make familiar paths more common, the limits of knowledge itself, how we go about testing new ideas, and even the nature of expertise itself.
As a first launch into this conversation, imagine someone you know who has a very unconventional training style. For me, the clearest example is how Ashton Rouska used to train, going heavy once a week on squat, bench, and deadlift and doing little else. I remember asking about his training leading in to the 2019 USA Powerlifting Raw Nationals. We were in the warmup room and I was just really curious. From the outside, it seemed that he was going pretty heavy all the time, and I never remembered seeing him do assistance exercises, bodybuilding work, or much at all in the way of movement variation besides sets of 5 and under.
It turns out this is exactly how he trained, it wasn’t an error of observation on my part. I thought to myself that if I was coaching Ashton, I might never have ended up on his current training plan. It’s not just that there’s a high probability of me not getting there, I’m talking never. What data would make me get rid of all accessories, or throw away my ideas about practice, frequency, the effect of a higher level athlete and their training needs, or throw away ideas about progression and periodization? What path or training insight would take me there? Even I found an athlete high in muscle mass and needing simple heavy loading to improve, I’d still give them another training day to practice the movement, or not be satisfied in current levels of muscle mass and add in work for quads, chest, back, and so on. How would I ever suppress my own biases to such an extreme degree?
I want to use Ashton and other extreme outlier training plans to talk more broadly about how we customize training and where the limits of that customization are. Given the standard middle of a bell curve for possible training programs a coach is likely to start off with for an athlete, how might one ever end up at Triphasic training, at capped intensity under 6 RPE, at Westside training, the Cube method, at super low training frequencies, singles only training, at getting rid of any hypertrophy work outside the main lift, or indeed other ideas that you’re probably (maybe) shaking your head at. What if those are viable training solutions for an athlete? If you’re unable to select those, does that make you less capable of a coach, or does it clarify your vision of what athletes need to progress? How do we move through the space of all possible training solutions in a way that leaves us maximally open to individualization while at the same time paying mind to science, observation, anecdote, and good data?
I hope you see the conundrum here. We say that we follow training response and individualize training, yet at the same time admitting there are potential ways of training that we have extreme low probabilities of ever choosing, based on our own prior experience and role models, our experience as athletes ourselves, the sources we choose to listen to, and onward into a connected web that represents the background by which we define our training philosophies.
Limits to the training space:
1) You don’t know what you don’t know.
One such limit is that as coaches, we’re unable to imagine training solutions that we can’t imagine. And I know this seems obvious, but if you’ve never heard of a paused squat or a Zercher squat, Sotts press, occlusion, rest pause, density training, fatigue drops, and many other training methods I don’t even know, you can’t program them. I can’t even write about training methods or exercises I don’t know--they’re entirely outside my scope to have any opinion or ability to program.
For even the most well-versed coaches, becoming entrenched in a set of familiar practices is common. The familiar comes easier, you develop a “style” of coaching that you turn to more and more, and stop seeking new ideas and new sources of information because athletes you work with are progressing well. This kind of rigor has some advantages as it likely makes you more efficient and practiced at the training you currently employ, but it essentially closes you off to new ideas. I see there being a balance here of new versus old, using common sense, experience, and exercise science to guide one’s path.
The point is that this exists, and is the first true boundary on the infinite possibility of training solutions: you cannot program what you are unable to think of.
2) Biases
Training biases come in all shapes and sizes. Perhaps you’re biased to the training that worked well for you personally, or the type of training you’ve seen work broadly. There’s biases to program exercises you see popular online (the availability heuristic), or biases toward programming what exercise science seems to show is viable. There’s a bias toward what other athletes you coach are doing*, or what other athletes you see online are doing. There’s a bias toward thinking of “old” training plans as inferior to more recent training (those poor people in the 1970s, if only they knew what we know today). There’s a bias either toward or against the way other countries tend to program for their athletes. There’s a bias toward seven-day microcycles. There’s probably a bias against accommodating resistance if I know my readers, and one against stretching, and one toward specificity in the main lifts.
Your biases create a rough map of what you think of as good solutions to training problems. There’s specific ranges for sets in a single session, amount of variation for main lifts, range of reps you’re likely to program, ranges of numbers of exercises per day, heavy sets per microcycle, and so on. You probably know these pretty well too.
The farther you move away from your tried and true training practices, the more an invisible string tries to pull you back to the familiar. The elastic pull of comfort in the known paths is constant at every check-in, every time the athlete trains. It takes work to be comfortable checking one’s biases and purely following training response where it occurs and trusting the athlete over one’s own intuitions.
3) Following the training data
The best coaches today make training plans based on good information, include a healthy dose of their experience, involve the athlete in the process, and have a system of checking to see what is working and what isn’t. That system then allows for subtle or major changes to training based on how well or horribly wrong training is going. There are real benefits to this approach, which allows coaches to parse what is happening and adjust when the athlete isn’t progressing.
One potential problem is that at almost no time does the athlete try out extremes that may be more optimal than the present approach. There is no trial period by which an athlete first tries a range of solutions (very low volume, very high volume, very low intensity, very high intensity) before carrying on with the winners of that test sample. I’ve seen studies designed this way, where subjects were run through three sample volume and rep combinations, eventually running the one that causes the highest increase in salivary testosterone response. You may have access to the testosterone responses of your athletes, but I certainly don’t.
There’s two problems here. One is that we don’t often know when we are at the fastest sustainable rate of progress, or when we’re lower than that number. Knowing if we’re at or below that number might help inform future training decisions, giving one the impetus to try out massive change. If an athlete has made no progress on a particular lift, the coach begins searching for new solutions beginning with easy, low-hanging fruit and searching outward for other solutions until the athlete starts making progress. What if an athlete adds 5kg to their squat in 8 weeks? The picture of whether this is or isn’t enough is unclear because there are no objective answers to the question “how much progress should I be making every 8 weeks?”
The other problem is that there’s typically no system for not progressing incrementally in changes to training, but trying out massive changes to training. Without such a system, we further limit ourselves to subtle iterations from the current training for any particular athlete. Normally coaches might swap competition squats for paused squats, pullups for lat pulldowns, add a single set, shift RPE by 1 up or down, and other small changes.
What is the system for arriving at massive changes? Every potential coaching decision you may ever make has a weighted probability of it occurring, much like a neural network in AI (actually it’s the OG neural network in reality) and typically it’s the athlete suggesting major sweeping changes, not the coach. Do we need such a system? Maybe, maybe not. But not being able to reach for jumps in changes to training and not steps further limits the training space.
4) AI and insights from other fields
One time I came up with a random training generator that chooses a random exercise, a random number of sets from one to ten, and a random number of reps from one to fifteen. This was more of a joke, but I think it helped me explore how truly varied training could be. I would look at some of the outcomes and laugh seeing how farfetched they were, but I see how in that same moment, I was rehearsing my own biases and my own sense of what feels normal and comfortable.
In other fields, researchers are using AI to come up with radical solutions to existing problems. In the new field of computational engineering, AIs are tasked with finding optimized solutions to problems. For instance, how do you build the frame of a car with the least material possible while also maximizing for support and rigidity? (See video). Or, how can you build the frame of a tower to maximize windows per room? The solutions were indeed radical, with almost organic-looking webs of metal frames that resemble a bird’s skeleton, beautiful in their alien way.
I hope you’re already thinking of how this might apply to powerlifting. Given enough data, could we task AIs with maximizing our own parameters? Could we minimize fatigue and the number of training days while maximizing strength outcomes? What would that look like, and is it something we’ve already thought of, or something new entirely? If it’s something we can’t currently think of, it represents yet another limitation to the possible training space. We’re starting to see that the weathered path and a few short forays into the unknown are the safe space for all of us. For good reason, but nevertheless, the clarification of what individualization is continues.
5) Expertise
Let’s come back down to earth here. There are experts in powerlifting coaching. What is an expert? Intuitively, I think an expert is a coach who delivers better results to more athletes, more of the time, quicker, with fewer negative outcomes like injuries or burnout. In my earlier research on what experts actually are, there’s a specific concept that kept coming up. Humans don’t usually pick from “the best of all possible options” every time they make decisions. At least, in chess, experts are usually the ones who can correctly represent large patterns of chess positions after just a few seconds of viewing. In powerlifting, this might be being able to see training patterns across time or quickly link faults in movement with verbal cues or training adjustments. Expertise is partly about the right kind of mental organization of information into useful patterns.
Thus, we see that it might not be necessary to understand the entirety of all options on the table for training plans if you know how to effectively move through the space. Experts are able to perceive large, meaningful patterns in their domain, are fast, and are able to represent problems at a deeper, more principled level. They’ve got lower error rates, they spend more time analyzing a problem qualitatively, and they focus on understanding first, and solutions second. There’s more on expertise. It’s a fascinating subject, but I don’t want to get too far off course.
Anyway, here’s the critical piece. Experts add constraints to reduce the search space to a smaller, more manageable area. Rather than looking at all possible squat options, let’s narrow in at only the ones that reduce acute loading, or increasing time in the bottom of the squat, or add extra work for the upper back. Or, let’s constrain the amount of training volume by using a system to understand which days should be easy or hard. Experts have systems that simplify and abstract patterns. It is therefore not likely to know all possible training variations if we’re able to effectively move between enough of them to deliver the right results at the right time to the right people.
CONCLUSION
The way “individualization” is commonly talked about assumes that coaches have access to all possible training options both in theory and in practice. There are limits to that, as I hope I’ve been able to show.
When we write training plans, we inherently build in things we value. We wouldn’t give them to an athlete if we didn’t think they had merit, and each coach has their own internal list of their most common exercises, progressions, and so on that serve as a basis on which to adjust. Training systems inherently have values on what constitutes a good training program. Every coaching system has boundaries. As a coach, if I don’t have total internal freedom to move across the entire training space, my ability to individualize is limited.
Lastly, limitations are both good and bad. Limitations increase creativity, not decrease it. Limitations reduce the amount of computation needed to look at all possible solutions and instead focus on a smaller subset that may have merit, dicing up a big problem into a more manageable chunk. On the flip side, limitations taken too far can blind someone to potential solutions. We need to walk a line of remaining open to new ideas, less entrenched in our ways.** We need to understand the real bounds on individualization, learn to check our natural biases and strive to find ways to arrive at more extreme coaching practices when required, else we limit ourselves from being able to help more people.
*Famed psychology researcher Daniel Kahneman spoke about grading the papers of his students, and how the prior score of the paper he just graded anchors and frames his decisions on the paper he is currently grading. There are clear implications here about training athletes and the necessity of keeping each athlete as their own island, not letting the decisions made for one athlete affect other athletes. One should effectively “cleanse their palette” before opening up the next athlete check-in to prevent overbleed and blended decision-making. I see examples of this in the real world, where everyone gets belt squats, or incline bench 12’s, or many other such training parameters that have less individualized reasons for their selection.
**An open question is whether or not powerlifting coaches need to be full chameleons, shifting and changing to the needs of the athlete, or whether there is a system you should select when you choose a coach that has its own values, values that persist and show up in training in subtle and large ways. Should coaches be true chameleons, or should they have some core identity that persists?