Don’t miss out on the latest CyclingTips updates.
You’re at the bottom of your favourite local climb, looking at the road snake its way up the hill ahead of you. You’ve ridden this climb a bunch of times but today is a little different — today you’re trying to set a new PB. You take a few deep breaths to compose yourself, take a sip of water, and then you’re off. The question is, how are you going to pace yourself?
Will you try to maintain a constant effort throughout? Will you try to start out a little easier and come home strong? Or will you start out as hard as you can and try to hold on for dear life?
Statistically speaking, your pacing strategy will likely depend on how experienced you are as a cyclist and, in all likelihood, how strong a rider you are. At least, that’s one of the findings from a new research paper that considers thousands of rides up one of Australia’s most well-known climbs.
Climbing 270m over the course of 5.54km — an average gradient of just below 5% — Norton Summit is the testing climb for cyclists in Adelaide. The main Strava segment for the ascent has been climbed more than 340,000 times by roughly 20,000 cyclists — plenty of data for researchers to work with.
The authors of this latest paper set out with three goals:
– Find out how each user paced themselves to get their best time.
– Find out which pacing methods were most prevalent.
– Determine whether there was a link between rider ability and pacing strategy.
Underpinning their analysis was a key assumption: “Given that some cyclists have attempted this segment many times, we expect that their personal best performance corresponds to their individual optimal pacing strategy.” That assumption won’t hold true for every rider, but given the average rider has ridden the climb roughly 17 times, it’s a reasonable starting point.
The researchers began their analysis by importing each rider’s best time and then cleaning up the data. They filtered out rides of longer than 25 minutes (slower than 13.3 km/h average) and efforts where the rider’s speed dropped below 8 km/h. They also filtered out any efforts with a GPS track that deviated too far from the average. This left them with a touch over 12,200 rides to work with.
The researchers then resampled each rider’s effort onto a grid with 512 points, to ensure they could accurately compare speed and power profiles from rider to rider, at the same points.
The researchers had two options when it came to determining a rider’s pacing strategy. They could consider the rider’s speed throughout the climb, or the power produced for the effort.
Using speed is problematic — changes in a rider’s speed throughout a climb are too strongly influenced by fluctuations in gradient, making it hard to accurately determine the pacing profile used. Fluctuations in power have a far smaller impact.
Unfortunately for the researchers, only around 20% of files they sampled contained power data. And so, for riders that hadn’t recorded power, the researchers set about estimating it. Using a mathematical model for road cycling power derived back in 1998, the researchers plugged in the gradient and the rider’s speed at each given point on the climb, allowing them to derive the corresponding power numbers.
Having plotted speed and power (estimated or recorded) for the entirety of each rider’s effort, they were then able to classify those efforts according to a particular pacing profile.
There are many ways you can pace yourself during an uphill time trial. In the case of this latest paper, the researchers considered six strategies, as laid out by one of the authors, Chris Abbiss, in a paper from 2014:
Positive: Your effort drops over the course of the climb.
Negative: Your effort increases over the course of the climb.
All-out: You start out as hard as you possibly can, and then fade from there.
Even: You keep your effort steady throughout.
Parabolic: You start hard, settle in, and then finish hard as well.
Variable: Your power varies often and considerably throughout.
In analysing the 12,202 efforts in their sample, the researchers observed five of the six profiles listed above — only the “all-out” strategy wasn’t represented. As the authors rightly note, such a strategy is only really viable for super-short efforts of up to a minute or two. The researchers did, however, observe a pacing profile that reminded them of the all-out strategy.
“We found in the data that some riders chose a profile similar to all-out, followed by a middle part with high power output, and finishing with decreasing power,” they write. The researchers dubbed this effort a “Negative Parabolic Effort.” (middle of the bottom row in the chart below).
How riders paced themselves
As you can see in the breakdown below, two pacing strategies were most common among the 12,202 efforts sampled: ‘Positive’ pacing (start hard, before slowing) and ‘Even’ pacing.
Parabolic positive: 545
Parabolic negative: 395
More interesting than the raw numbers is how these numbers line-up with rider ability. Slower riders who averaged up to 17 km/h (a time of more than 19:20 up Norton Summit), tended to opt for a positive pacing strategy. That is, they set out harder than they were able to maintain for the entirety of the effort.
This could be the result of inexperience — such riders might have overestimated the effort they could maintain, and faded as a result.
Faster riders tended towards a more consistent effort.
“In riders with better performances (17-27 km/h), the positive pacing profile is still common, but less frequently used in favor of even and negative pacing profiles,” the researchers write. This suggests greater experience — riders who are better at judging their effort and holding it all the way to the end, if not getting faster throughout.
This is even clearer among the best riders.
“An even profile was the most prevalent profile among the very best riders (27-30 km/h),” the paper reads. “This is important as it indicates that better performing athletes have faster performance times, not only because of greater average power output but also a greater capacity to minimize slowing when external load is high (i.e. high gradient) and therefore adopt a more even pacing profile.”
This is clear from a graph from the research paper, which shows that faster riders spend more of their time riding at a speed close to their average speed — suggestive of even pacing rather than an effort defined by lots of slowing down and speeding up (as seen in slower efforts).
Skewed by the Tour Down Under?
Dive into the Strava data from the Norton Summit climb and you’ll note something interesting about the best times. Of the 52 riders with an average speed of 27-30 km/h, 40 were pros that set their time during stage 4 of the 2016 Santos Tour Down Under. Those riders were riding in a bunch, in a road race, not riding solo in a time trial. Could it be that this latest research data is skewed as a result?
Could it be that those top riders were riding a reasonably even speed during the Tour Down Under, but that they would set a better time solo, using a different pacing strategy? I put that question to lead researcher, Dietmar Saupe, from the University of Konstanz in Germany.
“These 40 riders had times ranging from 11:06 to 11:43 so I gather they were not all in a single peloton from the start to the end,” he said. “So they may not all have ridden the segment as if they were purely self-paced. Nonetheless, they rode the segment the way it was recorded and so contributed a valid performance showing that it was possible to achieve that excellent performance with the type of pacing as used.
“But in principle it is correct to observe that there is a limitation that in general we cannot distinguish between riders that ride all by themselves and those that are “paced” by other riders as in a peloton or in a smaller group of riders. Regarding the fast riders above 27 km/h, I am also concerned a bit [with respect to] the small sample size. For other average speed intervals, like 20-23 km/h we have many, many more riders.”
So what can we learn from this study? What can we apply to our own riding?
“Regarding which pacing type should be applied to improve one’s performance is hard to say, based on our findings,” Saupe said. “It seems that at the low end of average speed more riders followed a positive strategy. So they started out with higher than average power and finished below average power. [But] if the magnitude of this difference is large, then this suggests that a suboptimal strategy was chosen in which too much of the energy reserves were spent right at the start.”
Given that the best riders seem to tend towards a relatively even pacing profile, that seems like a reasonable goal. But there’s more to it than that.
It’s well established that for a flat time trial, the optimal strategy is to maintain a constant power output. The same is true for an uphill time trial at a consistent gradient. However, the optimal strategy changes when you start introducing gradient changes (the case on virtually any climb).
The consensus from researchers is that you should increase your power when the gradient increases. As for how much you should increase it by, well, that’s a little more complicated.
It’s mathematically possible to calculate the optimal power profile for a given climb, by factoring in the nature of the climb (length, changes in gradient etc.) and the nature of the rider involved (energy resources, critical power etc.). For most riders, though, this level of detail isn’t easy to come by, nor is it particularly useful — it’s hard enough to maintain a constant power for a given effort, let alone achieve frequent and very specific variations in power while putting in a maximal effort.
Perhaps more useful is a 2016 study that simulated significant time gains for a TT rider who increased their power by 15% compared to their target average power when the gradient was 6% (on a sawtooth, up-and-down course). Research also suggests that the steeper the gradient, the more your power should probably increase from your baseline, particularly if you’re a lighter rider who handles steeper grades well.
It’s worth noting that as riders we almost instinctively increase our power whenever the gradient increases. There’s a tendency to push to maintain our speed and momentum which, when the gradient increases, leads to a power increase. Just don’t push it too hard — you need to be able to maintain the increased power for the duration of the steeper section (again, it comes back to even pacing) and then settle back into a rhythm once the road flattens off.
Speaking of which, the reverse of the above is true, too. If a climb flattens off from its regular gradient, or even starts descending briefly, you should reduce your power as a result. Again, you’ll likely find yourself doing this almost automatically.
So there you have it — want to improve at uphill time trials? Do as the pros do and keep your pacing as even as possible — especially when the gradient is consistent — but also be sure to vary your effort a little in parallel with changes in gradient.
Oh, and the fastest time up Norton Summit? A blistering 11:06, set during the 2016 Tour Down Under by none other than Thomas De Gendt. The Belgian averaged 427W and 29.6 km/h for that effort (roughly 6.2 W/kg), riding across to a breakaway on the climb. As the researchers write of De Gendt’s effort: “Presumably, the pacing of this athlete was close to optimal on this record ride.”
And as for the fastest (almost-entirely) solo effort? That belongs to Phil Gaimon, with a mind-boggling 11:14 (455W and 29.3 km/h average). You can see his full effort in the video below:
What’s your strategy when attempting a Strava PR or uphill time trial?