This blog post is an effort to codify my own mental model for approaching powerlifting programming. I expect nothing here to be particularly revolutionary, and in fact I suspect most coaches to have some version of the same thing. This is merely my effort to take what has been, up until now, a fairly nebulous and ill-defined idea and turn it into something concrete. I also don’t claim to have created any of the ideas here - merely synthesised the ideas of dozens of men and women much smarter than me into something that works for me.
Uncertainty in Coaching
The first and most important pillar of this philosophy is that we cannot have perfect information. Because powerlifting deals with humans (which are unpredictable) and human physiology (which is weird at the best of times), we can never be certain that something we’re doing is going to work. Even a decision we’re very confident in might be only be 90% certain. Even if we controlled every plausible variable (training, sleeping, nutrition, etc), there are things outside our control which could throw a spanner in the works (e.g. work or other life stress interfering with training.)
As an aside, excessively high confidence is a hallmark of a charlatan, something we see incredibly often in the fitness industry already.
Further, we don’t even have the benefit of perfect hindsight. Even if a decision we made worked, we don’t know what what would have happened if we tried a different path. Sure, whatever training decision we made resulted in a 10 kg increase on an athlete’s squat (a great outcome), but could a different decision have created a 12.5 kg increase (an even better outcome)? (For this reason, the word “optimal” has no place in powerlifting training or coaching, but that’s a matter for another day.)
Finally, even perfect decision making doesn’t necessarily guarantee good outcomes - it increases their odds, but sports is a wicked problem space. (Conversely, even making bad decisions doesn’t preclude good outcomes - but it might make them less frequent or less likely.)
As a result of this, we can consider every decision we make as a bet - a gamble on the future. As coaches, we have a responsibility to place these bets as effectively as possible.
At this point, I’m going to introduce the concept of a confidence, which we’ll use going forward. A confidence should really be expressed as a percentage - we might say I’m 90% sure we can improve your squat by 2.5 to 7.5 kg over the next 12 weeks. Here, we’re going to use it both loosely and qualitatively, as I think giving specific percentages implies a level of precision we don’t have. For the purposes of this, it doesn’t matter if it’s 90% or 92% or 95%; it’s only going to matter that I’m more confident in one pathway than another. Because this is a blog about powerlifting, we’re going to consider a “good” outcome one which increases our maxes by an amount we consider worthwhile, but this conceptual process applies similarly to any other pursuit.
Decision Trees
A simple decision tree
A decision tree is, as the name suggests, a tree of all of the possible decisions a powerlifting coach can make. Many of these decisions will relate to one another, and thus become layers in a decision tree. Many others will be independent, and thus make our tree enormously large.
Our tree is (probably) impossible to enumerate. There are tens of thousands of decisions a coach makes, some explicit, many implicit, each of which multiplies the size of our tree. Even the most foundational decisions (how many days per week should this athlete train? Which days? What frequency for each lift? Which days? Should they also do cardio? How much? How often?) results in thousands of possible outcomes, and that’s before we even get to the more granular decisions (What RPE or weight should this set be done at? Which accessory movement should be done for this athlete’s biceps? Should they even do bicep work?)
Complicating the process even more, although the pathways on these trees remain more or less fixed, the set of pathways which result in outcomes which we consider to be good changes over time - the outcomes are not static. We still have the same questions to answer (How many sets of squats should this athlete do?) but the answer changes (as a novice lifter maybe 10 is sufficient, maybe they need 18 or 20 by the time they’re experience and much stronger.)
Although I’ve primarily focussed these examples on programming for powerlifting, we can apply the same ideas to other aspects of coaching. What cues should I give this athlete? How should I encourage them, or should I look to encourage them at all?
Fortunately, we have a partner in the creation of the decision tree - the athlete themselves. Even if the athlete offers no opinions about coaching (which can happen), their logistics limit the tree. They might only want to train 4 or 5 days per week, meaning we don’t even need to consider the options falling between 1 and 3 or 6 and 7 times per week (or, technically, 0 days per week.) Athlete limitations set the shape of the decision tree, but they don’t necessarily help us traverse it.
Putting It Together
Putting together the concepts of uncertainty in decision making and decision trees, we come up with an idea - there exists a set of pathways through our decision tree, a set of decisions we can make, which lead to “good” outcomes. We don’t know what those decisions are, and even if we get it right once, we don’t know if we got it perfect, and we don’t know that doing the same thing again will lead to another good outcome.
Sounds grim.
We Don’t Have To Traverse The Entire Tree
All of this sounds very bleak so far. We have infinite options, don’t know enough, and even everything we do know changes constantly. It’s time for some good news.
Firstly - there’s lots of options which will lead to good outcomes.
Just like it says on the tin, there isn’t just one pathway to success in powerlifting. There are lots of ways to get strong, there are lots of ways to peak, there are lots of ways to communicate. Not everything you try will work, and not everything that works will work for everyone, but we are not searching for a single needle in an entire haystack. Success is not unique.
Secondly - we don’t have to traverse the entire tree.
We don’t have perfect information, but we do know some thing. Every person is different and unique, but largely we all fall within a few standard deviations of one another on some important dimensions. I don’t know how many sets of squats per week someone needs to get stronger just from looking at them, but there’s a good chance something between 5 and 20 is going to work. If I know their previous training, that’s somewhere I can start. If I know someone morphologically similar to them, that’s somewhere I can start. If I have neither of those, I can fall back to general literature, and use that to give me a starting point for my search.
We also have heuristics and other tools that help narrow our search - some of these come directly from our own experience (it seems like people with long legs need more quad work, lets include that), others from literature (women tend to use their pecs less when benching, let’s include more pec work), and yet others from biomechanical analysis (this lifter moves loading towards their anterior chain while squatting, they must have a stronger anterior chain.) All of these are tools we use to limit how much our decision tree we have to span before finding a set of decisions that works for us.
Finally - if we miss, we’re not left with nothing.
When we write a training program we are, implicitly or explicitly, making a hypothesis, or a guess as to what is going to happen. At minimum, that hypothesis is “this training will cause positive adaptations in this athlete”, though they might be much more nuanced if you’re trying specific things for a lift or an athlete. In addition, our responses are not binary (a good response or a bad response), but rather graded - a program can be anywhere between “really bad” (the athlete got weaker, or injured, or hated training) through to “really good” (the athlete hit new 1RMs on every lift) and anything in between.
Even one data point tells us something (even if it’s just “don’t do that”), but by stringing together a set of hypotheses and graded responses, we can narrow down our search towards what constitutes effective training for a given athlete. By making several similar, but modestly different training programs in a row, we can get a fairly good idea of the changes we need to make to move towards effective and efficient training by the trend of responses to those programs.
What Does an Experienced Coach Get Us
Although any coach (or indeed, any person) can follow this process, a good and experienced coach is going to get you three things - a better starting point, faster convergence, and better stickiness.
Of course, no coach is going to get it perfect the first time, every time. Even very good coaches are going to get it very wrong occasionally. What we expect to see, however, is that an attentive and experienced coach is likely to get a better starting point, on average, than a less experienced coach. As an incidental aside, I suspect this is at least partially why some coaches end up with a specialisation in their roster in terms of athlete weight classes or sex.
A more experienced coach should be able to converge on a “good” solution better. There’s a combination of factors that contribute to this, but two in particular come to mind - faster recognition of trends, and more useful heuristics. A coach who has seen every variation of trend and response before can, at least in theory, see trends faster than someone who hasn’t; because they’ve seen all the trends before. There may be small pieces of information they can tie into the pattern a less experienced coach can’t, or they may be able to project trends more accurately.
Finally, a more experienced coach may be able to predict or react to factors which move us away from “good” more quickly. It’s no secret that major changes to life stress can affect our training and recovery - we might expect a better coach to be able to react to changes more quickly, and potentially even forecast effects based on scheduled events (for example, an experienced coach might be able to forecast the effect of university exams, or react more quickly or effectively to a period of high work stress.)
Concluding Thoughts
This has been a philosophical ramble through the art of powerlifting coaching - an effort to codify the thought processes that go into the day-to-day decision making we all use. Although nothing here can be immediately actioned, I think it’s useful to conceptualise our own thought processes as we make decisions, and it might even help change the way we make decisions over time.