16Regatta Execution and Campaign Management

16.1Venue reconnaissance and knowledge capture

A venue model is a ranked set of hypotheses about the wind, current and sea state the crew will meet, each stated so that a day on the water can confirm or refute it. The ranking follows how often each condition governs racing in the event window. The model exists before the first training day, is corrected after every day and is the reference the daily protocol of 16.2 reads from.

The wind component starts with the gradient wind and the ways the venue modifies it. Friction backs and slows the surface wind relative to the gradient wind S38; the turning is about 30 degrees over land and generally less over water S39, and because the wind shifts left over land in the northern hemisphere it often converges over the water S40. A sea breeze requires a land-sea temperature difference of 5 °C or more, blows typically near 6 m/s and rotates clockwise through the day in the northern hemisphere S42; it fills first offshore and moves towards the land as a mini-front S51, and Houghton's rule is to expect one when an offshore gradient wind is under 25 knots with thin or broken cloud S43. The reconnaissance therefore records, for each day, the gradient direction and strength, the cloud, and the time and direction of the sea-breeze arrival, so that arrival and rotation become a venue-specific curve rather than a generic expectation. Houghton and Campbell's checklist of bends, bands and shifts near shores, land and sea breezes, water temperature, topography and clouds S43 is the template for the entries.

Forecast models are chosen by resolution. Global models run at 9 km (ECMWF) S45 and about 13 km (GFS) S46; regional models run at 1.3 km (AROME) S47 and 2.5 km (HARMONIE) S48, and kilometre-scale models represent sea breezes better S49. The reconnaissance identifies the highest-resolution model that covers the venue, logs its forecast against the observed wind each day and accumulates a bias table, because a model with a known bias is still useful.

The current component uses fixed references: lobster pots, moored boats and the choppiness of the water S68, current ribbons, trap buoys, birds, fish and the fishermen at tidal rips S69, with tidal atlases and tide tables as the base S70 and the crew's own measurements at the mark positions, timed against the tide phase, as the entries that matter. Wind against current raises wave height and wind with current lowers it S76, and aligned wind and current give smoother water than opposed S51, so the sea-state hypothesis is derived from the wind and current hypotheses and is then observed in its own right. Waves touch bottom from a depth of half their wavelength, then slow, shorten, steepen and refract towards the shore S72, so the depth chart predicts where leftover waves will be steepest and at what angle they will run to a new breeze.

Knowledge capture is a dated file with one entry per observation, coded by gradient direction, time of day, cloud, tide phase and position on the course, and a separate page of hypotheses with the evidence count for and against each. On the water the crew samples head-to-wind bearings, pressure and current at fixed points before each race and compares them with the page, and a hypothesis that fails twice under the same code is rewritten.

16.2Daily tuning and environmental decision protocols

The day's set-up is a decision under uncertainty, and the protocol fixes the order in which the evidence is gathered, the point at which each setting is frozen and the size of change that justifies moving off the base. Without it the crew tunes to the last thing it saw; with it, to the expected distribution of wind over the race window.

The inputs are the forecast band from the highest-resolution model identified in 16.1, corrected by the bias table, the observed wind ashore and afloat, the sea state and the crew's logged base settings. The class tuning guide sets the jib sheet in the middle hole and the track pins in the innermost holes for winds up to about 12 knots and flat water, and holds cap shroud tension at the baseline up to about 11 to 12 knots, progressively tighter to flatten the mainsail as the wind rises S13. The crew's own logged gauge readings with the mast stepped and no sails, crew value on each wire, are the check that the rig has returned to base after any change. No published rake table by wind band exists; the rake and shroud settings for each band are the crew's own logged values crew value, and the protocol's job is to select among them.

The selection targets the middle of the expected band with a control margin for its upper edge. The wind speed rises and falls by about 7 percent every 6 to 12 seconds, gusts carry 30 to 40 percent extra speed, and directional oscillations run with periods from about 3 minutes in heavy air to 10 minutes or more in light-medium air, with amplitudes from 5 degrees in light-medium air to 40 degrees in heavier air S50. A rig tuned to the lull is overpowered in every gust; a rig tuned to the gust is slow for half the race. The protocol therefore records the forecast median and upper edge and chooses the band whose logged settings keep the boat controllable at the upper edge while conceding least at the median. Where the forecast straddles two of the crew's bands, the sea state breaks the tie, and chop favours the more powerful setting.

Settings divide into those that can be changed afloat and those that cannot, and the second group is frozen at a fixed interval before launching crew value. After the freeze a change is made only when the observed band moves by more than the width of one tuning step, so that the base is never abandoned on the strength of a single gust or lull. The environmental decision runs the same way: Houghton's rule S43 confirms or rejects the sea-breeze hypothesis of 16.1, and the expected clockwise rotation S42 at roughly 5 degrees per hour S52 gives an expected direction at each start, checked against the head-to-wind samples.

The testable record for each day is the band chosen, the settings, the wind observed at each start and the error between the two. Over an event the errors form a distribution; a bias in one direction is a protocol fault, a wide spread with no bias is a forecast fault, and only the first is corrected by changing the protocol.

16.3Pre-launch, pre-race, and between-race routines

Routines remove decisions from the time-critical window. Every task that can be completed before its consequence arrives is completed then, in a fixed order, so that attention in the starting sequence and between races is spent on the wind and the fleet rather than on the boat.

The pre-launch routine is an equipment check ordered by the consequence of failure: rig terminals and spreader fittings, trapeze wires and hooks, halyards and their locks, rudder and daggerboard fixings, then sheets and blocks. The class rules allow one hull, one set of spars, one daggerboard, one rudder blade and one mainsail, jib and gennaker per event unless lost or damaged beyond repair S3, so the check is also the preservation of the only permitted sails and spars; a chafe point found ashore is repaired, one found afloat is a race lost. The rig is verified against the crew's logged base gauge readings crew value whenever it has been re-stepped or adjusted, and the settings frozen under 16.2 are confirmed as set.

The pre-race routine runs backwards from the start. The sequence displays the class flag at five minutes, the preparatory flag at four, gives one long sound at one minute and removes the class flag at zero, with times taken from the visual signals S1. Before the class flag the crew has the line: a transit through both ends to a fixed object ashore S51, and where possible two or three transits at either end, one on the line, one over and one behind S66. The bias is measured by comparing the line bearing with the head-to-wind bearing, since for every 5 degrees of bias one end is ahead by about one eighth of the line length in distance sailed S65, and the preparatory flag is read for the penalty regime, the I flag, the Z flag with its 20 percent scoring penalty, the U flag and the black flag S1. Head-to-wind samples at fixed intervals during the sequence are compared with the expected direction from 16.2. Tracking from the 2015 Europeans showed a minimum gap to the line of 10 m and a starboard-tack advantage of under one second in time to the line after the gun S16; the routine treats the last minute as the execution of a time-and-distance plan already made, not as a fresh decision.

The between-race routine is a fixed sequence with no coach input, because the Paris 2024 rule allowed no voice, radio or data transmission from the first warning signal to the end of the last race of the day S4. The previous race is closed with a single statement of the decision that mattered most, so that it does not occupy attention in the next race; the equipment is checked at the points most likely to have been damaged; the wind is re-sampled and compared with the day's hypotheses; a setting is changed only if the 16.2 change threshold is crossed; and the plan for the next race is stated in the crew's own protocol vocabulary before the warning signal. An observer would see the same sequence in the same order before every race, with step times that vary little.

16.4Coach-crew and shore-team information flow

Information flows in defined directions at defined times, and the boat is autonomous while racing. At Paris 2024 the notice of race prohibited voice, radio or data transmissions from the first warning signal to the end of the last race of the day, except in an emergency or with equipment provided by the organisers S4, and class rule C.5.2 limits what the boat may carry to timing devices, up to two compasses without functions beyond heading, heading memory and timing, camera and sound recording equipment, and tracking equipment where permitted by the class, the notice of race and the sailing instructions S3. The design assumption follows: everything the crew needs during racing is on the boat or in the crew's heads before the first warning signal.

The morning flow runs in three parts. The coach delivers the forecast synthesis and the day's venue hypotheses from 16.1, with the model bias applied. The crew delivers the settings decision from 16.2, stated as the band chosen and the reason. The shore team delivers the equipment state, the notice-board record, including any amendment to the sailing instructions and the protest time limit in force, and the logistics for the day. Each part is a statement, not a discussion; the discussion happened the previous evening. On the water before the sequence the coach boat contributes what the crew cannot measure themselves, observed wind samples across the course area, current at the marks and the timing of any sea-breeze front, and the crew receives it verbally at the coach boat and restates the day's plan in return, so that the coach can hear whether the plan has absorbed the new observations.

During racing there is no flow, and the division of labour on the boat governs. The helm generally makes the tactical decisions and steers while the crew does more of the physical work and controls the sails S9; the crew's own protocol, in the Temporada 26-27 and Precision Communication documents, defines the channels, vocabulary, priority order and silence discipline, and this section only fixes how the external flow interfaces with it. The interface rule is that the coach's pre-race input is delivered in the crew's own vocabulary, so that the crew stores it in the form in which it will be used.

After the last race the flow reverses. The coach brings the observed record, the wind log, the video and the tracking; the crew brings their perception of each leg; the shore team brings the equipment findings. The debrief of 16.5 reconciles these, and any difference between the coach's record and the crew's account is logged as a perception error to be tested, not resolved by authority. The standard is that the crew can state the day's plan and its reasoning without the coach present, that the coach's post-race record matches the crew's account or the difference is written down, and that no piece of information the crew needed during a race is found at the debrief to have been missing.

16.5Race debrief methodology

A debrief reconciles what the crew believed during the race with what the record shows, leg by leg and decision by decision, and produces a small number of coded items that feed 16.6 and 16.10. Its value is in the reconciliation, because the crew's memory of a race is a sequence of impressions weighted by emotionwhy and the record is a sequence of positions weighted by nothing.

The record comes from three sources. Event tracking publishes per-boat metrics, gap to leader, average speed over ground, distance travelled, number of manoeuvres, leg VMG and cross-track error, which the class defines as the extra distance sailed upwind from the course middle line on either side S78 S17. Onboard loggers add the resolution the tracking lacks: the Vakaros Atlas 2 uses a 25 Hz dual-band GNSS receiver, logs at 2, 5 or 10 Hz and reports distance-to-line, time-to-line, time-to-burn, shift tracking and VMG S79; the Sailmon MAX records boat speed, heading, heel, pitch, position, augmented true wind and line bias S80; Njord builds polar targets and gain-and-loss comparisons from the log files S31. Video from the coach boat supplies what none of these show, the trim and the body positions.

The order of the debrief follows the weight of the evidence. The 49erFX study broke a race into 38 skills and regressed race time on them; VMG on the first beat had a coefficient of -2.0, almost ten times that of any other skill, distance behind the line at the start had a coefficient of 0.060, and a good start with an efficient first beat was the most important combination for a shorter race S37. The debrief therefore starts with the start, then the first beat, then each remaining leg, and the same three questions are answered in the same order for each: what was the plan, what happened, and whether the difference came from the decision, the execution or the chance draw of the wind. The third question is answered on the information available at the time of the decision, so that a wrong decision that worked is still logged as wrong and a correct decision that lost is not.

The output is at most crew value items per race, each coded to a category (start, upwind speed, shifts, manoeuvres, downwind mode, mark, rules) and to the condition code of 16.1, with a time stamp that points to the track. The coach holds the record and asks the questions; the crew answers; nothing is argued beyond the evidence. The debrief runs on the day of racing after the boat is put away, late enough for the emotion to have settled and early enough for the memory to be intact, and the crew's post-race protocol sets the length. An observer would see written entries with time stamps in a consistent format for every race of the event, and could reconstruct from them, without the crew present, what was planned, what happened and what the crew concluded.

16.6Separating recurring performance losses from isolated incidents

A loss is recurring when it appears under the same condition code across races more often than chance would produce; an isolated incident is a loss whose cause is not expected to recur. The distinction decides the response: a recurring loss changes training and settings, an incident changes nothing. Confusing them in either direction is expensive, because a fix applied to an incident disturbs a base that was working, and an incident explanation attached to each instance of a recurring loss hides a systematic fault behind a series of plausible one-off stories.

The mechanism is the table of coded debrief items over the event. Each item carries its category and its condition code, wind band, side of the course, leg and sea state. A category that accumulates entries under the same code is a candidate recurring loss. The candidate is tested, not accepted: the crew predicts, before the next race in that condition, that the loss will appear, and a confirmed prediction moves the candidate from watch to test. The published evidence on where losses concentrate sets the prior: the 49erFX regression found first-beat VMG the dominant skill S37, tracking analysis of Olympic-class regattas concluded that training should prioritise the upwind course and manoeuvring efficiency, more prominently in strong wind S30, and extra manoeuvres on the first and last legs were associated with shorter race times S37. A candidate in one of these categories needs less evidence to be accepted than one in a category the data treats as minor.

The shape of a loss on the track distinguishes its type. A speed loss degrades the boat's position against the fleet steadily along a leg, with no step; a decision loss appears as a step at the moment the fleet split, and the size of the step matches the shift or pressure difference the crew could have seen; an execution loss appears as a step at a manoeuvre or a mark. Speed losses that recur in one wind band point at settings or technique in that band; decision losses that recur on one leg point at the venue model or the way the crew reads the course; execution losses that recur at one manoeuvre point at the handling routine. Rule 42 belongs here too: breaches in the 49er occur in the 4 to 8 knot wind range S18, so a yellow flag in that band is an incident once and a technique fault when the same movement produces it again.

The decision rule has three levels. Watch: the item is logged and its condition code noted. Test: one variable is changed in the next race in that condition, and only one, so that the outcome is attributable. Fix: the item becomes a training priority in 16.10 and, where settings are involved, a revision of the crew's logged values. The standard is that for each category on the table the crew can state whether the loss is recurring or isolated, cite the number of instances and their condition codes, and name the test that moved it between levels.

16.7Score management across qualifying and finals

Score management applies the event's scoring rules to set the risk tolerance for each race; it does not set the plan. Under Appendix A each race is scored by finishing place in the Low Point System S1, the series score is the total of race scores excluding the worst unless the notice of race says otherwise S1, and a series tie is broken first by listing each boat's scores from best to worst and comparing at the first difference, then by the last race, then the next-to-last S1.

At Paris 2024 the Men's Skiff sailed an opening series of 12 races plus a Medal Race, with the worst score discarded, the top ten qualifying for the Medal Race and Medal Race points doubled and not discardable S4 S6. For LA28 World Sailing has published sailing dates of 16 to 28 July 2028 with the 49er qualification series starting on 25 July and the final on 28 July S7. The format approved by the World Sailing Council on 10 December 2025 provides a three-day qualification series and a finals day of two single-point races for the top ten S8, with scores adjusted so that no competitor enters the final day with a gap greater than 9 points to the competitor ahead or behind S8; a January 2026 description of the proposal put the opening series at 8 to 9 races and added that from fourth to tenth no competitor could be more than 18 points behind third W1. The final format was to be determined no later than 31 May 2026 and no World Sailing confirmation after that date was found S8, so the crew treats these figures as provisional until the LA28 notice of race is published.

The discard governs qualifying risk. With one discard the first bad race costs the score it displaces from the discard and the second costs the lesser of the two, so risk tolerance falls the moment the discard is used; with no discard every race counts and the tolerance is set low from the first start. Dellenbaugh's percentage principle, to go for many little gains rather than one big gain and to treat a first-beat flyer as a lottery ticket S51, and Walker's rule to sail conservatively toward the middle on legs that are neither oscillating nor obviously one-sided S62, are the strategic expression of a low tolerance. Each morning the crew computes its position against the cut for the final, the points to the boats immediately ahead and behind, the races remaining and the discard state, and sets the day's tolerance from those figures.

The 9-point compression S8 changes the value of a qualifying lead: a leader gains nothing by extending beyond the gap and everything by holding order, and a chasing boat loses nothing by being far back rather than just outside it. In each of the two single-point races S8 the maximum swing between two boats that both finish is the fleet size less one, which for the ten boats of the final S8 equals the maximum permitted gap, so across the two races a boat can swing twice that, and a disqualification scored behind the fleet swings more, so no position is safe entering the final day, while before the last race a gap of more than 9 points to the boat behind is safe where both boats finish, and the tie-break rules of Appendix A decide equal totals S1. Finals-day strategy therefore concentrates on the boats within reach, and the race is sailed against them with the covering and consolidation logic of the tactics chapters rather than against the fleet.

16.8Adaptation after damage, penalties, or poor results

Adaptation is a bounded response chosen from a prepared set, and the size of the response is proportional to the evidence of 16.6, not to the emotion of the moment. Damage, penalties and poor results each have a decision tree written before the event, because the moment they occur is the moment the crew is least able to construct one.

Damage is constrained by the equipment rules. The class rules allow one hull, one set of spars, one daggerboard, one rudder blade and one mainsail, jib and gennaker per event unless lost or damaged beyond repair S3, so the default is repair and replacement follows the event's procedure. The shore team holds repair materials for the likely failures, the pre-launch check of 16.3 finds the precursors, and rule 28 lets a boat correct a course error before finishing S1, so a handling failure at a mark does not end the race. A repair changes the boat, so the rig is re-checked against the crew's logged gauge readings crew value and the settings against the base before the next race.

Penalties follow the event's regime. Paris 2024 and the 2025 World Championship replaced the two-turns penalty of rule 44.1 and the Appendix P penalties with a one-turn penalty for the 49er, and the event documents state which penalty applies at each regatta S4 S5; under Appendix P an observer who sees a rule 42 breach penalises the boat as soon as reasonably possible with a sound signal, a yellow flag and a hail of the sail number S2. After a contact incident the choice is between the penalty, a bounded and known cost, and a hearing, a binary outcome on evidence the crew cannot fully assess afloat. The rule is to take the turn when fault is in doubt and the place lost is small relative to the series, and to protest when the other boat's breach is clear and the place lost is large. A protesting boat hails "Protest" at the first reasonable opportunity S1; the default protest time limit is two hours after the last boat finishes S1, cut to 60 minutes at the 2025 World Championship S5, and the shore team files within that limit with the video and the tracking. A rule 42 flag, most likely in the 4 to 8 knot band S18, triggers a technique change for the rest of that band's racing under whichever penalty regime the event documents specify S4 S5.

A poor result is handled by 16.6 before anything is changed. One poor race is an incident until the table says otherwise, and the largest error available to the crew is to abandon a working base after a single race: the change threshold of 16.2 holds. When the table already shows a recurring loss in the day's condition, one variable is changed for the next race and the outcome is attributed to it. The starting flags scale the line margin the same way: a Z-flag penalty of 20 percent is a known cost, a U-flag disqualification is voided by a restart and a black-flag one is not S1, so the margin is set by the flag flying, not by the previous start. The standard is a written decision tree for each of damage, penalty and poor result, and a post-event record showing that each response taken during the event was one of the prepared branches.

16.9Travel, recovery, logistics, and equipment preservation

A campaign is a chain of events connected by transitions, and logistics is the management of the lead times and failure modes of those transitionswhy. Each transition, road travel, container shipping, measurement, a training block, a regatta, has a date by which it must be complete for the next one to start, and the critical path is the sequence of those dates with the least slackwhy. The plan is written backwards from the event's measurement date, because measurement is the point at which the boat and sails become fixed for the event.

Equipment preservation is governed by the class rules. Hulls, hull appendages, rigs and sails are manufacture controlled S3; one hull, one set of spars, one daggerboard, one rudder blade and one mainsail, jib and gennaker are allowed per event unless lost or damaged beyond repair S3; North 3Di mainsails, North jibs and MacDiarmid gennakers are prescribed at the Olympics and in gold-fleet championship racing S3; the minimum hull weight including wings is 94.0 kg under rule C.7.3 S3. Three consequences follow. The event sail set is measured, stored and used only for the event, training sails carry the hours, and the switch is scheduled so that the crew has sailed the event set enough to know its shape. The hull is kept dry and unrepaired between measurements, because absorbed water and repair material both add weight to a hull that is competitive at the minimum. The rig is re-checked against the crew's own logged base gauge readings, crew value on each wire with the mast stepped and no sails, after every transport, so that the base of 16.2 is the same base at every venue.

Travel is designed around the boat's dimensions, 4.99 m length and 2.9 m beam S9 S11, which set the trailer and container options, and around the mast at 8.4 m S11. The load plan is written once and repeated, with a fixed position and protection for hull, wings, foils, spars and sail tubes, so that each unload is a check against a known state. Spares travel with the boat in an inventory ordered by the product of failure probability, consequence and replacement lead time; an item with a long lead time and a race-ending consequence is carried even if it rarely fails.

Recovery, in the logistics sense, is scheduled time. The plan reserves blocks between travel, training and racing whose content belongs to the crew's physical and medical staff and is outside this book; logistics treats those blocks as fixed constraints and is judged by whether they survive travel delays and equipment problems. A buffer consumed by a late container has failed the crew even if the boat arrives, so buffers are sized from the recorded delays of previous transitions, not from the nominal schedule.

The standard is a written plan with a date and a named owner for every transition, a spares inventory in which every component permitted to be replaced has a replacement on site or a known lead time, a load plan that produces the same unload check at every venue, and a record of actual against planned dates that sizes the buffers for the next block.

16.10Post-event review and training-priority setting

The post-event review converts the event record into a ranked list of training priorities, each with a test, and updates the crew's models of the venue, the boat and the process. It runs after the debrief tables of 16.5 and the classification of 16.6 are complete, and it separates the outcome, the score, from the process, because a good score can conceal a poor process and a poor score can result from a good process in an unlucky draw.

The ranking weighs each recurring loss by its frequency and its cost in places, and checks the result against the published evidence. The 49erFX study found first-beat VMG carrying a coefficient almost ten times that of any other skill, distance behind the line at the start a measurable cost, and a good start with an efficient first beat the most important combination S37; tracking analysis of Olympic-class regattas concluded that training should prioritise the upwind course and manoeuvring efficiency, more prominently in strong wind S30; and extra manoeuvres on the first and last legs were associated with shorter race times S37. A recurring loss in these categories ranks above one of equal frequency elsewhere. The review uses the tracking leg metrics, leg VMG, cross-track error and number of manoeuvres S78, as the measures the next event will be judged by.

The models are updated in three places. The venue file of 16.1 receives the confirmed and refuted hypotheses with their evidence counts, so that the next visit starts from a corrected model. The crew's performance model receives the logged targets and settings from the event, and because no published 49er polar or rake table by wind band exists, these logged values are the reference and are revised only from the crew's own data. The process record receives the errors of the tuning protocol of 16.2, the timing of the routines of 16.3 and any information found missing in 16.4, each of which is a process fault with a process fix.

Training priorities are limited to crew value per block, each stated as a measurable test that the next event or the next two-boat session can score. A priority without a test is a wish, and a block with too many priorities trains none of them. Each priority names the condition code in which the loss occurred, so that the training block is scheduled where that condition can be found, and the priority is closed only when the test is passed in that condition, not when the training hours have been spent. The equipment review of 16.9 contributes any priority that is a settings revision rather than a skill.

The standard is a single document produced within a fixed interval after the event, containing the ranked priorities with their tests and condition codes, the updated venue and performance models, the process faults and their fixes, and the dates of the next block in which each priority will be trained and the event at which it will be re-measured. An observer comparing this document with the next event's debrief tables can see, for each priority, whether the loss recurred.