Skip to main content

Heading into Tokyo this year, Swimming Canada has ventured deeper than ever into data collection, hoping that new ways of parsing and interpreting races will lead the team back to the podium.

When Penny Oleksiak took the swimming world by storm in 2016, becoming the first Canadian to win four medals at a Summer Olympics, there was one aspect of her performances that became as unmistakable as her trips to the podium. As the races wore on, Oleksiak only seemed to get faster.

On the final lap, it appeared as if Oleksiak was gaining speed with every stroke. She quickly became known for her awe-inspiring finishes.

Oleksiak wasn’t alone. The same could be said for her teammate Taylor Ruck. When she won gold at the 2018 Pan Pacific Championships, the second half of Ruck’s 200-metre freestyle was something to behold. It was as though she was getting quicker as the finish line approached.

But it was all a mirage.

RAW
CORRECTED
What viewers see on their screen is an optical illusion.
Lead swimmers don’t speed up. The swimmers behind them are slowing down.
If we correct for the angle and movement of the camera, locking in on Ruck in the latter half of that 2018 race, the picture becomes clearer.
Though subtle, it’s most noticeable with the swimmer above her. As Ruck maintains her speed, Rikako Ikee of Japan tails off.
Such is the ruse of the strong finish. It’s a common misconception: The winner looks to be charging forward, but in reality, it’s the others that are falling back.
”It tricks you,” says Ryan Atkison, a former competitive swimmer and biomechanist with Canadian Sport Institute Ontario, who worked with Swimming Canada in the years leading up to Tokyo.
”If I’m watching, that’s what my mind is telling me. But my experience says otherwise … It’s almost impossible to speed up within a lap. Very few swimmers can do that.”

Atkison enjoys looking at swimming differently. Before Oleksiak burst onto the scene at the Rio Olympics, Canada’s coaches could see she had talent. But beginning in late 2015 and early 2016, the data on Penny told a hidden story.

As a 15-year-old, Oleksiak was inconsistent at the start of races. But after watching her finish strong at the end of the 100m butterfly, the coaches began pulling all the data they could get their hands on.

“We went back and looked at all the last 25m that we had on record, from the world-record swim, to the next fastest, and the next fastest, and the next fastest,” Atkison recalls. “She had the fastest last 25m in the world.”

It was a revelation. Filter out the noise from her starts and it was clear, Oleksiak was on the cusp of being world class. The data revealed more than the eye could see. “From there, we knew really what needed to be worked on,” Atkison says.

It was also a pivotal point for the program. It was around that time that Swimming Canada decided to get serious about data.

COUNTING
STROKES
AND KICKS

Ryan Atkison reviews practice footage with Canadian swimmer Sydney Pickrem in 2020. Melissa Tait/The Globe and Mail

Atkison, one of the architects of Canada’s data strategy, spent many hours over the past five years poolside with a video camera. His laptop is a vault stuffed with clips from meets around the world, documenting every swimmer on the roster, accompanied by spreadsheets, notes, charts and graphs.

Since the beginning of competitive swimming, the sport has always been awash in numbers, from race times and splits to rankings and beyond. Data mining is by no means new, but Swimming Canada decided after Rio it wanted to have better, more revelatory information on past performances at its disposal, and to be able to crunch those numbers faster, with comparisons between meets around the world, hoping to find glimmers of greatness that can be harnessed in each swimmer.

The goal was to measure and count everything – reaction times, stroke rates, stroke lengths, kicks, breaths, breakouts (the underwater transition from a dive to the initial strokes), the duration of each turn, and so on – so that the program could build a massive database, calculating how efficiently each swimmer moves through the water. As much as possible, the team needed to gather the video and data themselves, so that race comparisons were consistent, reliable, and always at their disposal.


To do this Atkison and company filmed from the pool deck, under the water during training, and often from the stands at major meets.
The challenge was to find a vantage point that best captured each performance, while at the same time manning multiple stopwatches.
Ensuring the data can be mined quickly and accurately is the key, so they can churn out rapid analysis or strategic insights from virtually anywhere in the world.
It could be from a laptop in the bowels of a stadium as the swimmers are preparing to race, or on an iPad in the stands between heats.

If a swimmer outperforms, or turns in a slower race than expected, it can’t be chalked up to a clutch performance or unexpected nerves. Somewhere in the data – in the kicks and the strokes, and the comparisons to past events – is likely the answer, a diagnosis of the problem.

“Even the positioning of a camera or the number of cameras that we set up at a venue will influence the quality of the data that we can get,” Atkison says.

Fluctuations in stroke counts, or kicks, either upward or downward are signs of trouble – of lost efficiency, power and speed, or of panic and over-aggressiveness.

The team produces race reports for each swimmer, colour-coded in varying shades of green or red, depending on whether the numbers look good, or are behind where they should be at that stage of a race.

Eyeing one of these reports, Atkison points out a race in Tennessee in early 2020, where Oleksiak won gold in the 100m freestyle.


2020 Pro Swim Series

Penny Oleksiak - 100m Freestyle

50m

50-75m STROKE RATE: 44.5

75m

75-100m STROKE RATE: 43.8

2020 Pro Swim Series

Penny Oleksiak - 100m Freestyle

50m

50-75m STROKE RATE: 44.5

75m

75-100m STROKE RATE: 43.8

2020 Pro Swim Series

Penny Oleksiak - 100m Freestyle

50m

50-75m STROKE RATE: 44.5

75m

75-100m STROKE RATE: 43.8

2020 Pro Swim Series

Penny Oleksiak - 100m Freestyle

50m

50-75m STROKE RATE: 44.5

75-100m STROKE RATE: 43.8

2020 Pro Swim Series

Penny Oleksiak - 100m Freestyle

50m

50-75m STROKE RATE: 44.5

75m

75-100m STROKE RATE: 43.8

2020 Pro Swim Series

Penny Oleksiak - 100m Freestyle

0m

KICKS OFF THE START: 5

75m

KICKS OFF THE TURN: 3

2020 Pro Swim Series

Penny Oleksiak - 100m Freestyle

0m

KICKS OFF THE START: 5

75m

KICKS OFF THE TURN: 3

2020 Pro Swim Series

Penny Oleksiak - 100m Freestyle

0m

KICKS OFF THE START: 5

75m

KICKS OFF THE TURN: 3

2020 Pro Swim Series

Penny Oleksiak - 100m Freestyle

0m

KICKS OFF THE START: 5

75m

KICKS OFF THE TURN: 3

2020 Pro Swim Series

Penny Oleksiak - 100m Freestyle

0m

KICKS OFF THE START: 5

75m

KICKS OFF THE TURN: 3

Her time of 53.41 seconds – 0.02 ahead of American Simone Manuel who she tied for gold in 2016 – was her fastest race in years.
Oleksiak would later say it was the first race since she took a break from swimming in 2018 that she felt back in form.
But the data told a more nuanced story. Oleksiak’s stroke rate was lower than usual. That is, she had won in spite of a sub-optimal performance.
This, Atkison says, pointing to the race report, was a golden opportunity.
”43 to 46 cycles per minute is pretty low for that event,” Atkison says. “It’s actually extremely low. Nobody in the world swims like that. They’re more 48-to-52 cycles per minute.”
He then turns to the kick counts, which the team has watched back in slow motion on the video to ensure precision.
”Five kicks off the start, three kicks off the turn is pretty low,” he says.
All of this is good news. After all, Oleksiak won the race. What if these kinks were ironed out?
”This would lead me to believe that there’s a lot more left,” Atkison says. “So she has a lot of room to improve there if she just gets her stroke rate up higher, completes a few more kicks off the start and off the turn.”

It’s not that Swimming Canada never paid attention to such details in the past. It has just never gathered, catalogued and organized them in so much detail, or so universally for all of the athletes, with the goal of lightning-quick race diagnostics in the heat of the moment.

“Careful video race analysis takes time and a skilled eye,” says Aaron Maszko, a data analyst and former competitive swimmer who will be parsing races for the Canadian team during the Olympics.

“Getting key information to our coaches and athletes from one performance to the next may increase the chances for a medal, but there is a limited amount of time available to deliver accurate information,” Maszko says.

“What happened when the athlete began to fatigue? Can one athlete’s strength capitalize on someone else’s performance gap? These are the types of things we are looking to tease out between races.”

For the coaches, it’s a chance to see what’s beneath the surface of a competition.

“It’s critical,” says Ryan Mallette, one of Canada’s coaches. “The further you go in swimming, you’re less sure about what it’s going to take to get that extra tenth of a second. Getting some concrete information always helps you know what to do next.”


2017 FINA World Championships

Kylie Masse - 100m Backstroke

50m

2017 50-75m STROKE RATE: 48.5

75m

2017 75-100m STROKE RATE: 47.8

2017 FINA World Championships

Kylie Masse - 100m Backstroke

50m

2017 50-75m STROKE RATE: 48.5

75m

2017 75-100m STROKE RATE: 47.8

2017 FINA World Championships

Kylie Masse - 100m Backstroke

50m

2017 50-75m STROKE RATE: 48.5

75m

2017 75-100m STROKE RATE: 47.8

2017 FINA World Championships

Kylie Masse - 100m Backstroke

50m

2017 50-75m STROKE RATE: 48.5

75m

2017 75-100m STROKE RATE: 47.8

2017 FINA World Championships

Kylie Masse - 100m Backstroke

50m

2017 50-75m STROKE RATE: 48.5

75m

2017 75-100m STROKE RATE: 47.8

2017 FINA World Championships

Kylie Masse - 100m Backstroke

50m

2019 FINA WCH

2017 50-75m STROKE RATE: 48.5

2019 50-75m STROKE RATE: 47

75m

2017 75-100m STROKE RATE: 47.8

2019 75-100m STROKE RATE: 46.3

2017 FINA World Championships

Kylie Masse - 100m Backstroke

2019 FINA WCH

50m

2017 50-75m STROKE RATE: 48.5

2019 50-75m STROKE RATE: 47

75m

2017 75-100m STROKE RATE: 47.8

2019 75-100m STROKE RATE: 46.3

2017 FINA World Championships

Kylie Masse - 100m Backstroke

2019 FINA WCH

50m

2017 50-75m STROKE RATE: 48.5

2019 50-75m STROKE RATE: 47

75m

2017 75-100m STROKE RATE: 47.8

2019 75-100m STROKE RATE: 46.3

2017 FINA World Championships

Kylie Masse - 100m Backstroke

2019 FINA WCH

50m

2017 50-75m STROKE RATE: 48.5

2019 50-75m STROKE RATE: 47

75m

2017 75-100m STROKE RATE: 47.8

2019 75-100m STROKE RATE: 46.3

2017 FINA World Championships

Kylie Masse - 100m Backstroke

2019 FINA WCH

50m

2017 50-75m STROKE RATE: 48.5

2019 50-75m STROKE RATE: 47

75m

2017 75-100m STROKE RATE: 47.8

2019 75-100m STROKE RATE: 46.3

Say, for example, the coaches detect a difference in the way gold-medal contender Kylie Masse races in her early heats, but the reason isn’t obvious.
Coaches can quickly compare the results to her best races, including her 100m backstroke world record in 2017, and her 2019 world championship – two similar performances – looking for the wrinkle.
All of this could be done before Masse is done cooling off. If necessary, last-minute revelations or inspirations can be served up on an iPad pre-race before she walks out onto the pool deck.
”We’re archiving it in the database which has thousands of records, and we can pull it up and create those comparison reports, and do that reasonably quickly,” Atkison says.

Canada isn’t the only country deploying data this way, but swimming is far newer on the analytics scene compared to several of the major professional sports, such as baseball or hockey, where funding for research and development is less of an issue.

“We’re collecting data on what we’ve always known,” says Mallette, but the quality of the analysis has gotten better.

“With someone like Penny Oleksiak, we have five years of data on her now, and we know where she has performed at each level … It’s about collecting enough data on an individual to know when they are at their best.”

PARALYSIS
BY ANALYSIS

Composite image. Timothy Moore/The Globe and Mail

This philosophical shift comes with a few risks. Swimming is contested in the water, not in front of a laptop. Coaches know they must use the analysis sparingly, so that it doesn’t become a burden. Presented with too much information, some athletes will tune out.

Asked about the data revolution inside Swimming Canada, Ruck winces as though she’s just been asked to solve a complex math problem.

“I’m not a numbers person,” she says apologetically. “I don’t really think about stroke counts when I compete, because when I dive in the water my mind literally just goes blank.”

Ruck won double bronze medals in 2016 as part of Canada’s 4x100m and 4x200m relay teams, and spent the intervening years as part of Stanford University’s vaunted swim program. She enters the Tokyo Games hoping for another trip to the podium.

“I just focus on racing and holding the water as best as I can,” Ruck says, referencing the mechanics of pushing the water past her with maximum efficiency.

“After, when I get shown my video, sometimes I see straight away, I’m not holding the water like I’m used to, or my stroke count is way too high for the beginning of this race. And then that explains why I slow down at the end. Little things like that.”

Not all of the swimmers are data-heads. Some love it, others find it a distraction. The challenge for the coaches is to find different ways to package the information.

“If we are doing [analysis on] how many strokes you take in a second, or how many metres you move in a second, I don’t really think about that as much,” Oleksiak says. “I think a lot about biomechanics, like where my hand is placed in the water and how I’m pulling the water, and what angle my head is at.”

Oleksiak’s coaches use data as a motivator with her. Having risen up the ranks so quickly since age 14 or 15, Oleksiak has been exposed to no shortage of praise about how great she is in the pool. She often sees through it, Atkison says.

“She’s got a high B.S. meter,” he says. But if the coaches want to show her something specific that will shave milliseconds off her races, that’s where the data makes a difference.

“Somebody like her seems to respond really well to that motivation. She wants to be the best in the world. And I think the closer you get and more you expose that the possibility is there, the more the drive, it just becomes positively reinforced.”


Swim Canada head coach Ben Titley at the PanAm Sport Centre in Toronto. Melissa Tait/The Globe and Mail

Ben Titley, Oleksiak’s coach, is a well-known motivator. He also prefers to follow the evidence. However, selling coaches on using more data analysis isn’t always easy. Training time is precious and bogging the swimmers down with too much information is a problem. That’s where the analysts get pushback.

Titley has a rule: three things in nine minutes. Discussions are kept short and focus on a few major points. Anything beyond that risks overload, and distracts from the task at hand. It is what the coaches call “Paralysis by analysis.”

They want the data served up in digestible chunks, which is what spawned the team’s colour-coded race reports.

“Most coaches don’t want to see that level of detail with the numbers, so we just try to summarize it and compress it to more relatable pieces of information,” Atkison says.

“What are the three areas that you need to change? Maybe it’s just the start, the middle turn and the finish, and those three data points are the only things that are presented to the athlete, and they compare with a different performance and show that they have room to improve, or the magnitude is such that it’s important.”

How the data is deployed differs even between coaches. Analysts might debrief one swimmer directly. With another, they may leave it up to a coach who prefers to control that process and relay the information.

“It does tend to bleed over into coaching style and coaching in general. … So it’s complicated,” he says. “It really depends.”

THE
DATA PAYOFF

Melissa Tait/The Globe and Mail

The strategy to improve Oleksiak’s starts has shown returns. Prior to the pandemic-induced delay of the Summer Games, she was already getting faster on the first 25m.

Part of that is her strength. No longer the precocious 16-year-old she was in Rio, Oleksiak, Ruck and Canada’s other young swimmers have gotten stronger. As Titley says, the teenagers who surprised in 2016, are now adults and capable of more if they can harness it.


The focus on every measurable aspect of Oleksiak’s starts – the reaction time, the cadence of her strokes, kicks and breathing – has made her faster.
”She’s starting to be one of the fastest through the first 25m in some of her races, which was never the case before, so we’re actually getting towards something new that’s a world-leading skill that’s the exact opposite of where she was before,” Atkison says.

There are other ways Canada may deploy data in Tokyo that will be completely different from past Olympics. How the relay team is selected, if substitutions are needed, could be one of them.

“Some of the areas that we hadn’t considered in the past is around relays,” Atkison says. “How other competitors are doing in their relays and who’s racing, and what’s the likelihood we can make a final with a certain collection of athletes, and using data in that way to inform decision making.”

All of this will be done under the sport’s brightest lights, and intense pressure.

“Any major competition on the world stage is stressful,” Maszko says. “For swimming, the Olympics is our pinnacle event.”

But there is still much that has yet to be harnessed.

While stroke counts, kicks, starts, turns, metres-per-second and a host of other data are now scrutinized through every meet, then logged into the database with accompanying video, there are areas that need more work.

Atkison thinks of what he calls the “messy five.” The last 5m before the wall in any race. At times it can be a free-for-all, where some swimmers break their form in a frantic lunge for the finish, and where competitions are won and lost.

As far as the data is concerned, it’s a relatively untapped frontier.

“We don’t do a good job of collecting what’s going on in those last 5m, because it’s hard, and it’s messy.” Atkison says. “But that’s where you get a lot of really useful information.”

In that case, the revolution is still a work in progress.


Video by Melissa Tait and Timothy Moore; Race footage provided by Swim Canada; Race footage compositing by Timothy Moore.

Follow related authors and topics

Authors and topics you follow will be added to your personal news feed in Following.

Interact with The Globe

Trending