Kona 2026: What 32 Pro Bike Positions Actually Measure

After the Ironman World Championship we ran thirty-two professional bike positions through the same pose model Flapp uses on your clips. The median pro rode with the torso 20° above horizontal, the upper arm at 88° to the torso, the elbow bent to 84° and the forearms tilted 26° up, with the top of the helmet 18% of a torso length above the top of the back. The forearms are the story: not one of the thirty-two rode them flat.

About the photos. All race photos here are by Scott Flathouse (@scottflathousephoto), Kona 2026, used with his permission. More of his Kona work, and your own photos if you raced: scottflathousephoto.com/events.

The median Kona position

Angle	Median	Middle half	Full range	Flapp's triathlon range
Torso, from horizontal	20°	19–22°	12–27°	20–30°
Shoulder, torso to upper arm	88°	83–92°	74–100°	80–105°
Elbow	84°	76–88°	66–102°	75–110°
Forearm tilt, above horizontal	26°	23–28°	19–32°	5–30°
Helmet above the back, % of torso length	18%	14–21%	7–33%	up to 23%

1. The hands have gone up

In every one of the thirty-two positions the forearms were tilted at least 19° above horizontal. Eighteen were past 25°, and three passed 30°. A few years ago a flat forearm was the default; at Kona nobody rode one.

Triathlon never had the rule that held road time trials back. Under UCI rules, the armrests in a road time trial could tilt no more than 15° until 1 January 2023, when the limit went up to 30°. Those rules do not apply to Ironman. Its 2025 rules limit how far the extensions reach, not past the leading edge of the front wheel, and set no limit on tilt. So the pros at Kona could go as steep as they liked, and the middle half of the field settled between 23° and 28°.

The aerodynamics back them. In a 2025 study that modelled one professional rider and checked the model in a wind tunnel, very high hands alone cut drag area by 3%. Combined with narrow elbows the saving grew to 7.7% for high hands and 11.2% for very high hands. Mat Steinmetz, a fitter who works with professional triathletes, describes the same change from the other side: 10–20° of bar tilt supports the upper body, relaxes the shoulders and makes a pocket for the head.

Two cautions. The saving is rider-specific, and the same study found that a smaller frontal area did not automatically mean less drag. And the elbow pays for it: when the hands rise, the elbow tends to close. Across the field the two moved in opposite directions (r = −0.69). Seven riders, about a fifth of the field, sat below the 75° elbow floor in our own range. That floor was written for flatter forearms; we have not lowered it because we have not yet found a fitter or a study that names a lower number. If you ride tilted extensions, read your elbow angle together with your forearm tilt, not on its own.

2. The torso is not as low as you would think

The median torso was 20°, and half the field sat between 18° and 22°. That is lower than most age-groupers ride, but it is not the flat back of a track pursuit. Steinmetz makes the same point: the pros' back angles have hardly changed in recent years. What changed is where the hands and the reach are.

There is a reason the back stops where it does. Fintelman's work on nineteen trained cyclists measured what a lower torso buys and what it costs. Going from 24° to 16° cut frontal area by about 5%, and going on to 8° cut about another 5%. But maximal power and every physiological variable fell as the torso dropped, and the flattest position was significantly worse than all the others. His models put the best torso angle around 17° for speeds between 32 and 39 km/h, and found that a fully flat back is never optimal.

Almost half the field, fifteen of the thirty-two, rode below 20°, the lowest at 12°. Pros can hold that: years of conditioning, four or more hours at a time. Not every age-group athlete can hold that and still make their power, though many can move toward it in small steps. Flapp does not take points for a torso flatter than the range on aero bars, but it says plainly that this is the trade, and that it only pays if you can still breathe, push and hold it late in the race.

3. A closed shoulder is a choice, not an error

Most of the field sat close to the classic reference. The median upper arm was at 88° to the torso, and half the field sat between 83° and 92°. Fit guidance puts the reference at 90°: elbows under the shoulders. Fintelman's riders, in their own preferred positions, averaged 91° give or take 5°. Kona's median sits right on top of that.

Four positions closed the shoulder below 80°, the furthest to 74°. The people who wrote the guidance never called that a fault. Dan Empfield's original description of the measurement says an angle below 90° is the athlete's personal choice, and that it is an angle opened past 90° that can become a problem.

What to check if you ride with the shoulder that closed: whether the shoulders are shrugged up toward the ears, which is tiring and hard to breathe through, and whether the arms are carrying you or you are holding yourself up.

4. The head: how far the helmet rises above the back

The helmet is the part of the head that actually sits in the wind, so that is what we measure. The pose model also finds the rider's outline. Through the highest point of the back we draw a level line, level because the wind is, and measure how far the top of the helmet rises above it, as a share of torso length. Where a helmet's tail lies flat on the back the outline cannot show where back ends and helmet begins, and on sixteen of the thirty-two positions we moved a helmet point by hand, the way the app now lets you move them on your own photo.

One rider was caught for an instant with the head ducked under the line of the back, which is not the position he rides, so his helmet reading is left out. In the other thirty-one, every pro had the helmet above the back: from 7% to 33% of a torso length, median 18%, half the field between 14% and 21%. Twenty-eight of the thirty-one were at 23% or lower; three rode with the head held higher, up to 33%. That is normal. You have to see the road, and an aero helmet's tail is shaped to meet the back, not to hide under it. What it costs to go higher has been measured only as a comparison of two positions: in a CFD study of time-trial helmets, head up against head down changed drag by up to 6.4%, more than switching between three helmets did.

5. Men and women: a first look

Twenty-one of the positions are men's and eleven are women's. We expected the two fields to sit differently. They do not, as far as this set can tell. The medians are close to identical: torso 20° for both, elbow 83° against 84°. The women's forearms were tilted 1° more (27° against 26°) and their shoulder sat 2° more open (90° against 87°), but the spread inside each group is far bigger than the gap between them, and in a statistical test none of these differences stands out from noise. Four of the eleven women's positions are broadcast stills, against two of the men's. Eleven positions are enough to say there is no big difference, not to rule out a small one. What shapes a fit here is the rider's proportions, range and trade-offs, not whether the rider is a man or a woman.

All thirty-two positions

Every position we measured, ordered from the flattest forearms to the steepest. Six of them are broadcast and social-media stills, softer and smaller than the photographs; their points were checked by hand like the rest. Where we could tell who the athlete is, the table names them. Every race photograph is shown below the table, credited to its photographer; the stills from video and the reader's photo are measured but not reproduced. Knee, hip and ankle are left out for the reasons in the next section.

#	Athlete	Field	Forearm tilt	Torso	Shoulder	Elbow	Helmet above the back
1	Kristian Blummenfelt	Men	19°	23°	86°	91°	+22%
2	Jenny Jendryschik	Women	19°	22°	100°	102°	+23%
3	Gregory Barnaby	Men	21°	25°	83°	88°	+14%
4	Vincent Luis	Men	21°	21°	86°	86°	+18%
5	Matthew Hanson	Men	22°	25°	97°	101°	+21%
6	Patrick Lange	Men	22°	25°	86°	89°	+19%
7	Lucy Charles-Barclay	Women	22°	15°	90°	83°	+19%
8	Ben Kanute	Men	22°	19°	89°	86°	+12%
9	Taylor Knibb	Women	24°	26°	87°	89°	+28%
10	Rudy Von Berg	Men	24°	20°	88°	84°	+14%
11	Cameron Wurf	Men	24°	19°	81°	76°	+20%
12	Hannah Berry	Women	24°	20°	93°	88°	+18%
13	Magnus Ditlev	Men	25°	20°	92°	88°	+17%
14	Gustav Iden	Men	25°	22°	74°	71°	+13%
15	Sam Long	Men	26°	22°	92°	88°	+18%
16	Kristian Høgenhaug	Men	26°	13°	96°	82°	+17%
17	—	Men	26°	27°	86°	87°	+23%
18	Matthew Hanson	Men	26°	25°	94°	93°	+23%
19	—	Women	26°	23°	87°	84°	+18%
20	—	Women	27°	21°	83°	77°	+33%
21	—	Men	28°	18°	79°	70°	+8%
22	—	Women	28°	17°	82°	71°	+22%
23	Imogen Simmonds	Women	28°	16°	91°	78°	+17%
24	Jason West	Men	28°	19°	76°	66°	+13%
25	—	Women	28°	18°	95°	84°	+20%
26	—	Men	29°	12°	88°	71°	—
27	Trevor Foley	Men	29°	19°	81°	70°	+14%
28	—	Men	29°	13°	95°	79°	+7%
29	Jana Uderstadt	Women	29°	21°	78°	70°	+14%
30	—	Men	31°	20°	88°	77°	+21%
31	—	Men	32°	21°	87°	76°	+19%
32	—	Women	32°	19°	91°	78°	+23%

Position 27 is Trevor Foley, from a photo a reader shared; an earlier Kona, not the 2026 race. Positions 5 and 18 are the same athlete, Matthew Hanson, in two frames: a broadcast still and a race photograph. We kept both; dropping either moves no median by more than half a degree. The one helmet left blank is the rider caught with the head ducked under the line of the back (section 4).

The race photographs are by Scott Flathouse (@scottflathousephoto), used with his permission. If you raced Kona, you may find your own photos from the day on his site. The labels are Flapp's own readings for a triathlon position: green inside the range, amber just outside it or below it at no cost, red outside it. A ringed point is one we moved by hand, and the green dot by the shoulders is the top of the back the helmet reading starts from. The legs are drawn but not measured, for the reasons in the next section.

What we did not measure, and why

No knee angles and no saddle verdicts. In a single race photo the crank is wherever it happened to be, so the knee angle is not a saddle-height reading. There is a second reason. Twenty-two of the thirty-two positions were seen from the drive side, the side with the chainring, because most riders crossed the frame in the same direction past photographers on one side of the road. On two bikes of our own we measured that the drive side reads the knee at the bottom of the stroke 8–10° lower than it really is: the ankle point slides onto the dark chainring behind it. So we left the legs out rather than publish numbers we know are biased.

How we measured: thirty-two professional positions. The race photographs are Scott Flathouse's, published with his permission. Twenty-five come from photographs of the 2026 race and six from stills of the race broadcast and of riders' own social-media posts; the twenty-sixth photograph is Trevor Foley, from an earlier Kona, which a reader shared in the comments, and we do not know who took it. Eight frames were left out: one where the model had found the passenger on the camera motorbike instead of the rider, and Flapp returned no score rather than a wrong one; two broadcast stills too blurred for the model to find a rider at all; three that showed a rider already counted; one still too poor to trust; and one broadcast still replaced by a sharper photograph of the same rider. Further screen grabs of riders already in the set were not counted twice. Three age-group athletes are kept out of these numbers too; we are collecting those to compare with the pros.

On every position, angles were measured in 2D on the side of the body nearest the camera, from four points: shoulder, elbow, wrist and hip. A pose model's points are its best guess at where each joint is, not the bony landmarks a fitter would mark by hand, and in the first version of this article several of them missed. So we went through every photo and every still and moved each point that missed into the joint by hand, the same way the app lets you correct a point on your own photo. 76 of the 128 joint points moved, by a median of about 2 cm. One 2026 photograph is itself a still from video, where the model had mixed up the joints; every point on it was placed by hand. On the other photographs the largest move was about 8 cm. Centimetres here are estimated from each rider's own hip-to-shoulder distance, taken as 50 cm. The helmet reading is taken from the rider's outline, not from these four points.

When we first corrected the points, on the original fifteen photos, the medians moved by 2° at most; single riders moved more. That is why we report the field rather than anyone's exact fit. Some athletes may appear in more than one frame that we could not tell apart; where two frames clearly showed the same rider we kept one, with the one exception named under the table. It is one race and one roadside course, so read these as the shape of a field, not as anyone's exact fit. The names come from the race number on the bike, checked against the published pro start list, or from the athlete being clearly recognisable in the frame; where we could not be sure, the row is left blank. The point is still the position, not the person.

What to take for your own position

• Try some forearm tilt before you chase a lower torso. It is the trend with the best evidence and it is cheap to adjust. Every position here had at least 19°, and the middle half sat between 23° and 28°. Keep the elbows narrow; that is where most of the measured saving came from.
• Do not copy a 15° torso. Find the lowest angle you can hold for your race distance and still make your power, then test it in training at race effort.
• A closed shoulder is fine if you are not shrugging and you can breathe.
• Keep the helmet within the pros' range. Twenty-eight of the thirty-one Kona pros we could read kept the helmet within 23% of a torso length above the back. If yours rises more, drop the head a little, as far as you can while still seeing the road.
• Measure yourself the same way. Film side-on, camera at hip height, and include the bottom of the pedal stroke. If a point lands off a joint, drag it into place before you read the numbers. Then you can put your own numbers next to the table above.

What we changed in Flapp

Measuring this field showed that three of our own rules were behind it, and we changed each one only as far as a source supports. The forearm-tilt range now runs to 30°. That is the ceiling road time trials allow under UCI rules; triathlon sets none, and three Kona positions went past it, so read it as a reference, not a rule. A shoulder closed below the range on aero bars is amber and costs a third of the penalty, because no source calls it a fault. The aero estimate, which used to put every torso below 28° in one bucket, now separates three steps using Fintelman's measured frontal areas. And a torso flatter than the range costs no points, but comes with the concern you read above. For the head, the app measures the helmet above the back, lets you move both points it is measured between, and asks you to check the top-of-back point when the helmet's tail lies flat on it. A helmet up to 23% of a torso length above the back, where twenty-eight of the thirty-one Kona helmet readings sat, costs nothing; the three riders above it do not widen that for everyone. Above that it is amber with a small penalty, never red, because the one study we have compares two head positions and draws no line between them.

From the same race: pros and age-groupers bend the knee the same, but the pros sit 4–5 cm further forward over the crank.

Measure your own position