UNDERSTANDING CALIBRATION + CLASSIFICATION

Calibration and Classification are two key parts of the Ergatta experience that make your workouts truly adapt and evolve with you. You interact with these elements every single time you hop on your rower, so we want to make sure their mechanism and function is crystal clear. Let’s break it down.

It should be expected that changing Classifications happens infrequently. Fitness changes can happen in small increments, while your Classification represents a broad swath of possible fitness levels. So, as you continue to workout and recalibrate with Ergatta, and your calibration hasn’t changed or changes infrequently, that isn’t unexpected! You can manually recalibrate if you feel things aren’t moving quickly enough; but otherwise, it isn’t unexpected to have Classification changes take more than two or three recalibrations.

You might be wondering: how does recalibration work exactly? The recalibration logic takes your average speed for a given interval and compares it to your expected speed for that interval. So what goes into coming up with your expected speed for an interval? Let’s begin with Pulse workouts. Pulse workouts can have one of three different “focuses”: split (Power), stroke rate (Precision), or a combination (Hybrid). Let’s take them in order. Split has the simplest expectation. We start with the prescribed Intensity Zone. For any working interval (an interval that is not warm-up, cool-down, or active rest), we expect your average speed to be in the Intensity Zone. Where in the Intensity Zone is dependent on the structure of the workout. For instance, if you are doing a 30-minute workout that is entirely in Steady, we predict that it will be more fatiguing than a 10-minute workout entirely in Steady.

All of the following examples will use equal intervals for simplicity, but I’ll note in advance that it does differ per interval based on the structure. For example, if your Steady zone is 2:10–2:00, the 10-minute workout might predict an average split per working interval of 2:05, while the 30-minute workout might predict 2:08. If your average speed is higher than the predicted speed (maybe your average split for each interval is 2:03 in both workouts), we recognize that as over-performing your expectation, and we would assign a positive growth score. If your average split were 2:00, the growth score would be higher. If your split were closer to 2:10, we would interpret that as a calibration too aggressive, and lower it. And if you decide that even 2:00 feels too easy and go into the Race zone? While the game penalizes you insofar as your percent achieved is lower, we would recognize that growth score as being higher. When considering your performance against expectation, we do not penalize you for going higher than your prescribed Intensity Zone. Because the goal with recalibration is to more accurately determine your fitness potential, we would interpret that as requiring a more positive growth score.

Precision and Hybrid workouts have similar conceptual implementations: we predict an average speed based on the prescription and your current Intensity Zones. With Precision workouts, we predict based on the prescribed SPM and an “analogous” Intensity Zone. For instance, a prescribed SPM of 24 would predict a split closer to the top end of Steady, while a prescribed SPM of 26 would predict closer to the lower end of Race. Again, the exact numbers depend on the workout structure, accounting for the strain of any given workout structure.

In short: for Pulse workouts, the easiest way to generate upward recalibrations is to perform near the top of your Intensity Zones.

For Races, we do a similar calculation of performance (time spent in active segments) against expected. Here, though, the “expected” number is transparent: it exactly matches the expected Race time shown on the overview screen prior to selecting competition. This expected Race time comes from applying your calibration to the workout structure. For instance, a single-segment 10km race would predict a slower average split than a 10km race with ten segments of 1km each and 0:30 rest in between each segment. And this would be slower than a 5km race with five segments of 1km each and 0:30 rest in between each segment.

This expected Race time also feeds into how we select your competitors when you use our auto-select competition feature. When we select competitors for you, we take the first 20 racers ahead of your projected time, and the first 20 below that projected time, and then pick a random assortment of 4 from each group.

So it should be apparent now: if you are performing better than your expected finish times in Races — directly evidenced by out-performing your competition when using auto-selected competitors — we will interpret that as a positive signal when recalibrating next.

As a final note, it’s important to know that we try to account for recovery workouts as well. We have a feature to exclude a workout from recalibration — look at the upper-right on the set-up screen just before starting a workout. In addition to this, we try to find outliers; if a growth score for a workout is anomalous relative to the rest of the workouts in that set, we’ll exclude it. We are more sensitive for this on the low end, so a recovery workout is more easily detected.