Pensacola · 2026-11-05

• Get into H1 upwind sooner: it is the biggest gain on the page. At 10–12.5 km/h H1 made 9.5–10.1 km/h of VMG against 7.0–7.7 in H2, about 2.4 km/h more. Development spent 36 % of that band in H1, Race Team 16 % and Geneva's fleet 28 %. Race Team's upwind was mostly sailing back to the start box, so part of that gap is the job they were doing, not how they sailed it.
• Start 4 is the model: on time, then quick. At 16:30:35 B15 was 2.2 s from the line at 16.9 km/h and still accelerating, then reached M1 in 58.7 s and the gate in 180.2 s, the quickest of the day to both. Start 1 had a similar speed but was 5.6 s late. Start 3 was on time but at 12.6 km/h. Aim for within 4 s and 15 km/h or more. In the no-foiling Geneva and Abu Dhabi races, 11 of the 12 starts that led to M1 were inside 4.2 s.
• Downwind, sail Development's wing. At 10–12.5 km/h in H2, Development ran 38.3° of rotation and +980 mm of clew, almost exactly Geneva's best boat CAN (38.5°, +938 mm). Race Team ran 34.8° and +664 mm, close to NZL's habit at Geneva (33.3°, +605 mm), and was 0.4 km/h down on VMG. Ease the wing to about 38°.
• Board Cycling pays on the raise, downwind, and the double pump did not earn its second movement. A single raise downwind gained 2.1 m over matched moments (78 raises), and a drop cost 1.2 m (82 drops). The three real double pumps gained 2.5, −0.8 and 2.1 m. To be worth two movements a double has to beat 4.2 m, and none did. Three tries cannot settle it, but nothing here says keep doing it. See the deep dive.
• The board limit was broken eight times. The worst was the starboard board at 16:23:28 (22:23:28 UTC): 9 movements in a minute and over the limit for 33 s. Keep an eye on the count in the pre-start.
• B15's wind instruments need calibrating before upwind numbers mean much. TWA reads about 2–5° wide upwind and TWS probably 1–2 km/h high, which makes upwind VMG read about 0.5–1.5 km/h low. Downwind angles are clean. Ask the techs to check it. Until then treat today's upwind gap to Geneva as smaller than it looks, and the 10–12.5 band as nearer 9–11.

• 14:47:15 (20:47:15 UTC) warm-up: 90 s in H1 at 26.5 km/h and 97° TWA in 12.8 km/h of wind, the quickest sailing of the day
• 14:56:56 (20:56:56 UTC) double raise downwind: speed barely changed
• 15:02:57 (21:02:57 UTC) Start 1: windward hull out 28 s before the gun, right after twist came off, but 5.6 s late
• 15:09:56 (21:09:56 UTC) Start 2: good gun at 16.7 km/h, then the board drops and speed falls to 13.7 km/h in 12 s
• 15:29:55 (21:29:55 UTC) double raise on a reach
• 15:57:03 (21:57:03 UTC) a timer ran in the Development block; left out of the starts because that block was free sailing
• 16:23:28 (22:23:28 UTC) starboard board over the limit for 33 s, peaking at 9 movements in the minute
• 16:25:40 (22:25:40 UTC) double drop downwind
• 16:30:35 (22:30:35 UTC) Start 4: the best start of the day

Getting out of H2 is worth more than any trim today. H1 at 10–12.5 km/h made about 2.4 km/h more VMG than H2 for both teams. Geneva's boats gained about 3 km/h the same way.

Part of the angle gap is the instruments. B15 reads about 60° in H1 and 54° in H2, against 50–51° for Geneva's fleet and NZL. The tack test below suggests 2–5° of that is calibration. Take 3–5° off and H1 VMG comes up to about 10.3–10.9 for Race Team and 11.0–11.6 for Development, against 11.1 for Geneva's fleet and 12.2 for NZL. H2 comes up to about 7.5–7.8 and 8.2–8.6, against 8.1 and 9.0. So Development is about level with the fleet, and both teams are still short of NZL at Geneva. The remaining angle gap looks real: at 18–20 km/h in H1 we are faster than Geneva (17.6–18.6) and lower. SPEED v2's 10 km/h row asks for 17 km/h at 49°, so try trading speed for height once H1 is established.

Twist in H1. Development ran 11.1° of wing twist, in line with the best boat FRA (10.4°), at 20.1 km/h. Race Team ran 3.0° at 18.1 km/h. Development was also flatter, at 25.7° of camber against 30.3°.

Jib in H2. Race Team's lead sat at 77 % against 56–59 % for Development and all of Geneva, but this is mostly the run back to the start box, so check it before changing anything. Cunningham was 64–66 kgf for both teams against 107 kgf for Geneva's fleet, 81 for NZL and 227 for FRA. Jib load was 186–203 kgf against 276 for the fleet.

FRA is Geneva's best boat here, but its TWA reads about 2° narrower than the fleet's in the same tack test, so its VMG may be slightly flattered.

Time in each mode

H2 / RUA / H1 as a percentage of time above 8 km/h, in B15's own wind bands.

Against Geneva

Medians over 6 s periods, n = periods. The fleet is the median of each boat's median, and the best boat has the highest median VMG with at least 5 periods. Each row is its own median, so a column is not one moment.

H2, 7.5–10 km/h

H2, 10–12.5 km/h

RUA, 7.5–10 km/h

RUA, 10–12.5 km/h

H1, 10–12.5 km/h

Downwind is where B15 compares best, and these numbers are safe: the gybe test shows no angle error. At 10–12.5 km/h in H2 Development made 8.5 km/h of VMG and Race Team 8.1, against 9.5 for CAN, 7.9 for Geneva's fleet and 8.4 for NZL. At 7.5–10 Race Team made 7.3 and Development 7.0, against 7.5 for SWE, 6.3 for the fleet and 5.3 for NZL.

Ease the wing. In the 10–12.5 band Development's rotation and clew (38.3°, +980 mm) match CAN (38.5°, +938 mm), and it was 0.4 km/h up on Race Team (34.8°, +664 mm).

Jib. Both teams ran the sheet fully eased with no load. CAN ran 95 % and 50 kgf, and the fleet 97 % and 38 kgf. CAN also ran more twist, 15.7° against 9–10°. Both are worth a try.

Angle. At 124–126° TWA in this band we match CAN (127°) and the fleet (126°), and sit deeper than NZL at Geneva (122°), which answers the open thread on our downwind angle.

Time in each mode

Against Geneva

H2, 7.5–10 km/h

H2, 10–12.5 km/h

H2, 12.5–15 km/h

RUA, 10–12.5 km/h

Race Team only: the four start-to-M1 reaches, from just after the line to just before M1. All of it falls in the 10–12.5 km/h band. Geneva had no NZL periods in these cells, and only one boat in RUA, so that column is USA alone.

In RUA the reach was on the pace. 18.7 km/h at 95° against USA's 19.3 km/h at 92°. In H2 it was not. 14.6 km/h at 69° against USA's 18.2 km/h at 92° and the fleet's 16.1 at 87°, and that is almost all Start 2, which sailed high and slow.

Differences to try. Jib lead was 100 % against 79 % for USA, and cunningham 64–66 kgf against about 174. Clew in H2 was −259 mm against −34 mm for USA, so ease the wing on the reach.

Warm-up reference: 26.5 km/h in H1 at 97° with the windward board up, against 19.3 km/h on the best start. The reaches here are only 300 m, so the boat has little time to get there.

H2, 10–12.5 km/h

RUA, 10–12.5 km/h

Practice manoeuvres scoring under 25 are left out, as in every report, along with turns outside the team blocks. That keeps 17 of the day's 29 tacks and gybes. Samples are small, so treat these as a first read. VMG through is the measure, not metres lost.

Race Team's gybes at 7.5–10 km/h were good: 8.8 km/h of VMG through (5 gybes) against 8.9 for DEN, 7.7 for Geneva's fleet and 7.4 for NZL.

Development's turns at 10–12.5 km/h had more to find. Gybes made 8.5 km/h (5 gybes) against 11.7 for SWE, 8.4 for the fleet and 9.0 for NZL, and tacks 8.8 (3 tacks) against 10.5 for GBR and 9.1 for the fleet. SWE and GBR rest on 2 turns each, so the fleet figure is the safer one to aim at. Race Team's tacks are too few to compare.

The touchdown lists and the bear-away chart further down come from the standard report and mean nothing today, because the boat was not flying.

Against Geneva

Medians. A team or boat needs 2 turns to show, and the best boat has the highest VMG through. Geneva turns entered in H2, RUA or H1 only.

Four Race Team starts, with guns from the boat's own timer. The leg counter did not run, so M1 and the gate are the closest passes of the course file's marks, and the boat's own line position and logged start speed match these numbers on all four.

On time, then quick. Start 4 had both, and was the quickest to M1 and to the gate. In Geneva and Abu Dhabi races with no foiling, 11 of the 12 starts that led to M1 were within 4.2 s of the line, and on-time starts at 15 km/h or more did best.

Each start, in one line. Start 1 had the hull out early and flew the reach, but crossed 5.7 s late. Start 2 had a good gun, then speed fell 3 km/h as the board went down at the gun, and the hull never came out. Start 3 was on time but still at 52° and 12.6 km/h, and took 22 s to reach 90 % of reach speed. Start 4 came in fast from a 5-minute sequence, accelerating through the gun.

The hull. Reach speed followed how much of the reach had the windward hull flying: 100 %, 0 %, 47 % and 84 % gave 19.0, 14.7, 16.6 and 19.3 km/h. Taking twist off was the most reliable trigger with the windward board down at 14 km/h or more: 4 of 6 twist drops lifted the hull within 8 s.

Geneva's start columns use only races where nobody foiled in the first 30 s. DEN rests on 2 starts and NZL on 2, so read those as pointers. A timer also ran at 15:57:03 in the Development block, which is left out because that block was free sailing.

There were no race roundings today, so this is the bottom gate on the four Race Team starts. M1 is a virtual mark.

Both MB2 roundings held their speed. Both MS2 roundings did not. Starts 1 and 4 gybed once and rounded MB2 with a low of 14.9 and 11.0 km/h. Starts 2 and 3 finished east of MS2, gybed back to enter the gate and gybed again at the mark, with lows of 7.4 and 8.1 km/h. From M1, either gybe early and take MB2, or aim just west of MS2 so that one turn takes you round it.

Short answer: Board Cycling gains a little downwind, mostly from the raise. The double pump showed no sign of being worth its extra movement, though three tries cannot prove it.

How this was measured. Movements use the widget's rule (a 200 mm reversal), which gave 181 port and 231 starboard, and nearly all were full 1.9 m strokes. Boats tend to pump when already slow, and speed recovers anyway, so a before-and-after number flatters pumping. Each stroke is compared with the 15 closest moments that had no board movement, matched on speed, speed trend, TWA and TWS. The gain is metres over 15 s beyond those moments.

The boost comes on the raise. Speed rises about 0.7–1.3 km/h within 4–6 s. From a windward raise it fades by about 12 s, and from a leeward raise it holds. Dropping costs a little, especially the windward board.

A full cycle. Downwind, down then up on one board nets +3.0 m for two movements (31 cycles), or about 1.5 m per movement. Up then down nets nothing (39 cycles). Upwind, up then down nets +6.4 m (20 cycles) and down then up +1.1 m (15 cycles). Do not cycle on the reach: it lost ground.

Over longer stretches. In 30 s windows of straight H2 sailing, at the same wind and angle, each extra movement went with +0.14 km/h of boat speed and +0.13 km/h of VMG downwind (156 windows, a weak link at r = 0.21). Upwind there was no visible gain (+0.06 km/h, VMG −0.05, 70 windows). Downwind that works out at about 1.2 m per movement, which agrees with the stroke-by-stroke figure.

Double pumps

There were 17 moments with both boards moving the same way within 2 s, but only 3 were real pumps on straight sailing. The other 14 were boards moving around a gybe or a tack, including some up to 17 s ahead of a gybe, or a partial move.

Was it worth it? Not on this evidence. A double costs one movement on each board, so it has to gain more than two singles to pay. Downwind that means beating 4.2 m, and the best double made 2.5 m. It may also be structurally hard to win. Each board has to go down before it can come up, so 6 movements a minute means at most 3 raises per board, and 6 raises across both boards however you pair them. Doubling cannot add raises. It only stacks two boosts of about 6–8 s on top of each other, where spacing them keeps the boost going for longer.

The double drop at 16:25:40 did beat a typical single drop, so it is not all one way. Singles also vary a lot (the middle half of downwind raises runs from −0.4 to +4.4 m), and three doubles cannot settle this. To settle it: run about 10 double raises against 10 single raises, back to back, in steady H2 downwind.

The limit

Each board was at 2 or fewer movements left for a good share of the session: 14 % of the time on port and 24 % on starboard. The limit was exceeded eight times:

Race Team moved each board about 2 times a minute on average and Development about 1.2, so the limit is a matter of bursts, mostly in the pre-start and around gybes.

B15 sails alone, so its wind cannot be calibrated against other boats the way Geneva's was. Two checks suggest it needs attention.

TWA reads wide upwind. At a tack the wind does not flip, so half the heading change is a check on TWA. Over 8 steady tacks B15 reported 53.2° against 45.2° from the headings, a gap of 6.8°. The same test on 311 Geneva tacks gave 1.5°, so B15 is about 5° wider than normal on this test. A second test, the jump in TWD at each tack, gives only about 2°. So the error is somewhere around 2–5°. Downwind is clean: 12 gybes gave 2.9° against 3.5° for Geneva.

VMG is worked out from TWA, so upwind VMG reads low by about 0.5–1.5 km/h. Downwind is not affected.

TWS looks high. B15 read 10.4–11.4 km/h while the two mark boats reporting wind read 7.8–9.1, a gap of about 2.4 km/h. At Geneva the gap between boats and marks was about 0–1 km/h. Today's bands may therefore be 1–2 km/h too high, so the 10–12.5 band may really be nearer 9–11. Mark boats at a different height, or only two marks reporting, could explain part of it.

What to do: ask the boat's techs to calibrate B15's wind, and repeat these tests on the next sailing day. The numbers in this report are as measured, with no correction applied.

Race Team and Development did not sail the same job, so read this as a list of things to try, not a verdict. Race Team's upwind was mostly the run back to the start box, and Development free-sailed.

All figures are at 10–12.5 km/h unless marked. Wind was similar: the boat read 10.4–10.5 km/h in the Race Team blocks and 11.4 in Development's, so Development had up to 1 km/h more, which could account for some of its speed.

• Wind bands: 2.5 km/h, on each boat's own wind reading corrected by its calibration offset for the day. Offsets (km/h, + = the sensor reads high), from moments boats sailed within 150 m of each other: B15 +0.0.
• Bands are never merged. A band is shown only with at least 5 of our periods and 3 other boats with 5 each; for manoeuvres and roundings, 2 of ours and 3 other boats with 2 each.
• Racing is split by leg (leg 1 = reaching, the final leg left out). Practice is split by wind angle: upwind at or below 75°, downwind at or above 110°.
• Samples: 6 s straight-line periods from the app (upwind, downwind), each the mean of its ~12 half-second readings; 2 s chunks of the steady part of leg 1 (reaching: 10 s after the start to 5 s before mark 1).
• Period filters: a period is left out if it starts within 10 s of one of the boat's tacks, gybes, bear-aways or mark roundings (the boat is still accelerating), or if it is under 80 % of the boat's own VMG target (bad air, not trying). The floor also takes out genuinely slow sailing, so the table below shows what it removed race by race.
• Light day. Periods need at least 8 km/h of boat speed and tacks and gybes at least 8 km/h at the crossing, because the boats sailed and turned in displacement. The VMG floor applies to foiling periods only (both hulls at least 500 mm clear): the targets assume foiling, and against them most displacement sailing sits under 80 %. Foiling and displacement periods are compared separately where it matters.
• Every comparison is a median over that boat's samples, never a single sample. Fleet = the median of the other boats' medians, so each boat counts once. Best boat = the other boat with the best median: VMG (km/h) upwind and downwind, VMC to mark 1 on the reach, VMG through the manoeuvre on tacks and gybes, metres below target on bear-aways, exit speed on roundings, metres at mark 1 on starts. VMG % is against each team's own polar, so it is shown for us but never used to rank boats.
• Targets: SPEED v2 cheatsheet, LAB2 27m with the Big jib and LARW rudders: the sheet's row at the band's lower wind speed (the 30 km/h row for 30-32.5), as written, never interpolated. Clew targets are converted to mm with the sign turned round (- = to windward, as in the data); jib track is converted from sheet angle to jib lead % (0 % = fully inboard). Not compared: jib sheet load (the sheet's units are not kgf yet).
• Phases included: practice, racing. Races left out: none. Crew on board: 4. Notes: 27.5 m wing, LAB2, LARW. Race Team 14:45-15:30 and 16:22-16:40; Development Team 15:40-16:15 (free sailing). Two course files; the second created 15:43 local..
• Manoeuvres are free manoeuvres only: anything within 10 s of the start, a mark or the finish belongs to the Mark roundings tab, and anything after the finish is left out. Practice tacks and gybes with an app quality score under 25 are left out (the boat was not trying); racing ones are all kept.
• VMG through the manoeuvre = ground made upwind (tacks) or downwind (gybes) over the app's 25 s window, in km/h - the measure that compares boats directly. Metres lost is the app's ghost-boat figure; each ghost holds that boat's OWN pre-manoeuvre VMG, so a boat that was quicker going in shows more loss for the same turn. For bear-aways it is distance sailed below the boat's target speed over 15 s. Touchdown = lowest hull under 100 mm.

What the filters removed