The term periodization refers to the practice of dividing the training process into distinct phases, each of which is defined by a specific purpose and made up of workouts that are intended to fulfill its purpose. Simply put, an athlete who practices periodization does different things at different points in the training cycle, whereas an athlete who does not periodize his training does the same thing week in and week out.

To put an analogy on it, an athlete who practices periodization is like a farmer, whereas an athlete who does not is like a factory worker. What sort of work does a farmer do? It depends entirely on when you visit the farm. In one season you may find him planting, in another administering pesticides, and in yet another harvesting. No matter when you visit the assembly line, however, you will find the factory worker putting screws in widgets.

Endurance athletes have not always practiced periodization in its current form. Like most modern training methods that we take for granted today, it had to be discovered. The idea that it is beneficial to train in different ways at different times is not terribly intuitive, which is why even now athletes who are not taught to periodize their training don’t.

While it hasn’t always existed, periodization was not discovered as a single event by a single individual—it’s too complex for that to have happened. Rather, it evolved piecemeal over time, with lots of different athletes and coaches representing a variety of endurance disciplines contributing to its development. In running, the most influential periodization model is the one that was created by the legendary New Zealand coach Arthur Lydiard. In this model, a phase of base-building that features increasing amounts of long, slow running is followed by a strength-building phase that features lots of hill running, then a speed phase dominated by short, fast intervals and finally a racing phase.

Since this model was developed in the 1950s, coaches and athletes have come to the conclusion that such a strict segregation of training types isn’t necessary. In his Hierarchy of Endurance Training Needs, noted exercise physiologist Stephen Seiler ranks “General Periodization Details” fourth in importance out of eight fundamental endurance training methods, remarking that there is a tendency to overrate the impact of sequencing different training stimuli in one way versus another. For example, some marathoners like to do a little bit of very high-intensity track work in the peak training weeks preceding a race, whereas others prefer to do almost all of their uptempo work at speeds closer to race pace. The evidence suggests that either can work.

Where there is less wiggle room is in how stress and rest are managed. Every runner, even those who don’t know the first thing about periodization, understand that their overall training workload should tend to increase as they move closer to race day. But the best results seem to come not when athletes continuously do about as much as their present fitness level allows, as intuition dictates, but rather when they intentionally do significantly less training than they could handle at some times and intentionally overreach—that is, taking on a training load that would break them if they sustained it for very long—at other times—and this is less intuitive.

Most people exercise with a “get in shape, stay in shape” mentality. In a typical scenario, a sedentary person sees an alarming number of the bathroom scale or has a scary doctor’s appointment and starts working out. Initially, he can’t do very much, but as he builds some fitness he does a little more and a little more until he reaches a point where he’s doing about all the exercise he cares to do. From that point on, he follows the same exercise routine for the rest of his life (slight exaggeration to make a point).

Endurance training doesn’t work like that—or shouldn’t. If your goal is to achieve peak performance in a race, you need to train in a way that, to put it crudely, burns you out, so that after the race you need a break and after the break, having voluntarily give up some fitness, you must ease gently back into a new cycle of training.

Several years ago Stephen McGregor, an exercise physiologist at Eastern Michigan University, shared with me some interesting data he collected from professional cyclists. These were athletes who logged all of their training on TrainingPeaks, whose Performance Management Chart quantifies fitness through a variable called chronic training load (or CTL). In most circumstances, CTL is a very accurate predictor of performance. As a general rule, the higher your CTL is, the fitter you truly are and the better you perform in a races and other endurance tests. But in analyzing the data or professional cyclists, McGregor found that, over the course of a season, their CTL and performance decoupled. Early in the season, as the riders increased their CTL, their performance improved as expected. Then, as they maintained their peak CTL into the racing season, their performance level held steady—for a while. But after three months or so of this, their performance level began to decline even as their CTL was maintained.

In other words, the same training that made the cyclists fitter initially burned them out over time. On its face, this seems like an avoidable mistake, but training less in order to achieve a sustainable CTL is no alternative because in that case their peak performance level wouldn’t be as high. Peak fitness and sustainable training loads are simply incompatible, and this is the number-one reason it’s necessary to periodize.

Always remember, you’re a farmer not a factory worker.