What decides a 100m freestyle? A regression study of 13 races.
Original research from EasySpeed. We measured 13 elite 100m freestyles segment by segment and ran the regression. The race is not won where most coaches think.
The verdict
The race is decided at 75-85m. Not the start. Not the speed zone. Not the turn. One ten-meter segment, three-quarters into the race, predicts the final time. Everything else is consequence.
Five things we found
- 1.The race is decided at 75-85m — the only ten-meter segment that significantly predicts final time.
- 2.The speed zone is a trap. Going fast in the middle of the race correlates -0.890 with the ability to hold speed later.
- 3.Turns are free speed. Investing in underwater quality costs nothing downstream and pays off in the post-turn segment.
- 4.Distance per stroke (DPS) on the first lap predicts whether the swimmer holds speed in the second half. Raw speed does not.
- 5.The 85-95m segment correlates with nothing in the race. It's a separate neuromuscular system, built in training, not raced.
How we measured it
Dataset: 13 races of the same swimmer, 100m freestyle long course, finishing in a range of 51.94 to 52.95 seconds. Every race was measured through the EasySpeed platform: ten-meter segment speeds, distance per stroke (DPS) per lap, stroke rate, turn time, underwater kicks, breakout distance, and reaction time.
Method: Pearson correlation between every measured variable and final race time. With n=13, a correlation of |r| > 0.553 reaches statistical significance at p<0.05. We also ran correlations between intermediate variables to identify the trade-off structure underneath the race.
Limits of this studyThis is a single-subject longitudinal study, not a population cohort. The findings are hypothesis-generating: they describe what changed across one swimmer's career, not what is universally true. We are publishing because the patterns match independent coaching observations — Brett Hawke's teaching on the speed-zone trap, Cam McEvoy's first-half restraint — and because making the data public is more useful than keeping it private.
Zone structure of the race
We divided the 100m into five functional zones. These are the units the rest of the study refers to.
Finding 1
The race is decided at 75-85m.
Across the 13 races, the speed at 75-85m was the only ten-meter segment that significantly predicted final time (r = -0.585, p < 0.05). No other ten-meter segment of the race — not the start, not the speed zone, not the finish sprint — reached significance against final time.
The adjacent segment, 65-75m, does not predict the 75-85m result (r = 0.086). This means the collapse is delayed, not gradual. A swimmer can look fine through 75m and lose the race in the next ten meters.
Finding 2
The speed zone is a trap.
Speed in the speed zone (15-45m) has essentially zero correlation with final time (r = -0.054). Going fast in the middle of the race does not, on its own, make the race faster.
But it correlates strongly and negatively with the ability to hold speed afterward: r = -0.890 with maintenance speed across the second half, r = -0.693 with speed in the decisive 75-85m segment, and r = -0.631 with finish speed. Faster speed-zone speed = bigger collapse later. This is the energetic cost surfacing one zone too late to feel it.
Finding 3
Distance per stroke beats raw speed.
DPS on the first lap correlates +0.597 with maintenance speed and +0.553 with finish speed. The more efficient the swimmer is in the first half, the better they hold speed in the second. The relationship is reversed for stroke count: more strokes on lap one correlates -0.555 with finish speed.
DPS alone does not predict final time (r = -0.018). What matters is the interaction: efficient swimmers convert their efficiency into late-race speed. Inefficient swimmers cannot. The race is won by efficiency, not raw speed.
Finding 4
Turns are free speed.
A 'slower' turn time — meaning more time spent underwater off the wall — correlates +0.584 with speed in the post-turn 65-75m segment. Time invested in underwater quality pays off immediately downstream.
Turn time has essentially no relationship with finish speed (r = 0.171). There is no energetic cost to a longer underwater. The turn is the one place in the race where slow input produces fast output, at zero downstream price.
Finding 5
The mechanism: tempo inflation.
Tempo change between lap one and lap two correlates -0.587 with speed at 75-85m. When the swimmer fatigues, they compensate by spinning their arms faster — higher rate, worse DPS — which is energetically expensive and collapses the decisive segment.
This is the why behind the trap. A fast first half forces a tempo inflation in the second half. The tempo inflation eats the 75-85m segment. The 75-85m segment loses the race. The chain is consistent across the dataset.
Finding 6
85-95m is a separate system.
The 85-95m segment correlates with nothing in the race. Not the speed zone (r = 0.372). Not DPS on lap one (r = -0.143). Not the 75-85m segment immediately before it (r = -0.173).
This is the neuromuscular sprint reserve — a separate physical system built in training, not raced. A coach who looks at a fading swimmer and prescribes more sprint work is treating the wrong problem. The 75-85m fade is a strategy problem. The 85-95m sprint is a training problem. They are not the same.
Two races, one thesis
Slowest first half on record (25.65). Slowest speed zone (1.81 m/s). Best DPS (1.61m), fewest strokes (31). Only negative split in the dataset. Best finish speed (1.83 m/s). Produced by four months of short, quality-over-volume practices.
Best turn ever recorded (10.11s). Moderate everything else. The personal best was carried by turn quality, not by race execution. The execution was actually worse than the 52.35 race.
Imagine the 52.35 execution with the PB turn. That is 51-low. The race we are after is already inside the data — it just hasn't been swum together yet.
What this means for coaches
- 1Restrain the speed zone.If the speed zone is a trap, the coaching cue is to swim it relaxed, not aggressive. The athletes who go fastest here lose the most later. Slower is faster.
- 2Drill DPS, not stroke count.Distance per stroke on lap one is the single best predictor of late-race speed. The training intervention is technique work that lengthens the stroke, not work that drives the cycle rate.
- 3Invest in the underwater.Time spent underwater off the wall is the only training investment in this study that produces speed at zero downstream cost. Coaches who don't drill underwater leave free time on the table.
- 4Don't conflate the fade with the sprint.The 75-85m fade is a pacing and efficiency problem. The 85-95m sprint is a neuromuscular reserve problem. They have different fixes. Treating them as one is why generic 'finish stronger' work doesn't move the needle.
- 5The second half predicts the race.Second-half time correlates +0.589 with final time. First-half time correlates only +0.227 (not significant). When evaluating a race, look at the second half first. That is the part that decided the result.
Frequently asked questions
What determines who wins a 100m freestyle?
In this study of 13 elite 100m freestyles, the single strongest predictor of final time was speed at 75-85m. No other ten-meter segment reached statistical significance. The race is decided at 75-85m, and what happens earlier in the race only matters insofar as it affects that segment.
Does swimming faster in the middle of the race make you faster overall?
No. Speed in the 15-45m zone has essentially zero correlation with final time (r = -0.054). But it has a very strong negative correlation with the ability to hold speed in the second half (r = -0.890). Faster middle = bigger collapse later. The speed zone is an energy trap, not a time gain.
When does fatigue actually set in during a 100m freestyle?
The visible collapse occurs at 75-85m. But the cause is set much earlier: a fast first half forces a tempo inflation in the second half, which burns through energy and produces the 75-85m collapse. Fatigue is not gradual — it is delayed. The 65-75m segment can look normal while the 75-85m segment falls apart.
Is the start the most important part of a 100m sprint?
Not in this dataset. The 0-15m time, dolphin kick count, and block-to-entry time all failed to predict final time. The start phase is too consistent across races (only 284ms of variation across all 13 races) to differentiate them. Start technique matters for absolute speed, but it does not differentiate races within an established swimmer.
How can a turn be 'free speed'?
Time spent in the underwater portion of the turn — what looks like a 'slower' turn — correlates +0.584 with speed coming out of the turn, and has essentially no correlation with finish speed. Underwater work converts directly to post-turn speed without spending any energy that's needed later. It is the only zone in the race where slower input produces faster output at no cost.
How was this study conducted?
13 races of one swimmer (100m freestyle long course, range 51.94–52.95 seconds) were measured using EasySpeed: ten-meter segment speeds, DPS per lap, stroke rate, turn time, underwater kicks, breakout distance, reaction time. Pearson correlations were run between every variable and final time, and between variables to identify trade-offs. With n=13, |r| > 0.553 reaches significance at p<0.05. The study is single-subject longitudinal, so findings are hypothesis-generating, not population-level claims.
Get this analysis on your own races.
Upload one race. EasySpeed returns the same metrics this study is built on — segment speeds, DPS, stroke rate, turn breakdown, underwater quality. Free to start.
Analyze a race