Polars Explained: The Boat's Speed Map

A polar is the boat's speed map: for any true wind strength and angle it gives the boatspeed and heading you should be making. But it is worth being precise about what that map is — not a log of past runs and not a marketing curve, but the solved output of a force-balance model, corrected against your own data. Understanding how the number is generated is what lets you trust it, calibrate to it, and know exactly where it stops being true.

What a polar actually is

Strip away the picture and a polar is a lookup table. Rows are true wind angle (TWA) — the angle of the real wind to the boat's track through the water — and columns are true wind speed (TWS). Every cell holds a boatspeed. Ask "how fast at 20 knots of breeze, 130 degrees off the wind?" and you read straight to the cell, interpolating between the tabulated points.

The diagram is that table drawn in polar coordinates: one curve per wind strength, the bearing around the plot is TWA and the radius from the centre is boatspeed. The classic teardrop falls out of the physics — pinched near head-to-wind inside the no-go zone, bulging on the reaches where apparent-wind driving force peaks against still-modest resistance, and tucking back in dead downwind where the boat can never exceed the true wind's downwind component.

Two points on every curve matter more than any other, and both are defined geometrically rather than by eye. Velocity made good is the projection of the boat's velocity onto the wind axis: VMG = boatspeed × cos(TWA). Maximum upwind VMG is the point on the curve furthest up the page, which is exactly where a horizontal line is tangent to the curve. Maximum downwind VMG is the tangent point at the bottom. Those two tangent points are your beating and running angles — the numbers a tactician reads first — and their geometry is the reason the upwind and downwind targets behave so differently, which we come back to below.

Where polars come from: inside the VPP

Polars begin as predictions and mature into measurements. The prediction comes from a Velocity Prediction Program, and it is worth knowing what it actually solves, because that defines every assumption baked into your target.

A VPP is two things bolted together: a boat model (a set of force equations) and a solution algorithm. At each TWA/TWS point it enforces two equilibrium conditions simultaneously. First, longitudinal balance: the sail driving force equals total resistance (F_drive = F_resistance). Second, transverse balance: the sail heeling moment equals the hull righting moment (M_heel = M_right). The solver takes initial guesses for boatspeed, heel angle and sail trim (typically a flat parameter that de-powers the sail and a reef/twist parameter), computes the force residuals, and iterates the independent variables until both equations close to tolerance. The converged boatspeed is the polar number for that cell.

The two force models it balances are where the real engineering lives:

• Hydrodynamic (resistance) side. Upright bare-hull resistance is dominated by two components: viscous drag (skin friction over the wetted surface, scaling roughly with the square of speed) and residuary resistance (wave-making, which climbs steeply as the hull approaches its wave-speed regime). In the industry-standard approach this comes from regression equations fitted to the Delft Systematic Yacht Hull Series — tank data across a matrix of hull forms — plus corrections for heel and for the added resistance of leeway. On top of the bare hull, the keel and rudders each generate lift to resist leeway, and that lift comes with induced drag that grows with the square of the side force, which is why a boat pinched too high (large leeway, high side force) is dragging a much larger induced-drag penalty than the helm can feel.
• Aerodynamic (driving) side. The rig is modelled as a set of lift and drag coefficients versus apparent wind angle for each sail type, commonly validated against wind-tunnel results, then de-powered by the flat and reef parameters as wind builds so the solver can keep heel within the righting-moment budget. This is why a polar has a de-powering knee: below it the boat is fully powered and speed rises with wind; above it the sails are being progressively flattened and twisted to shed heeling moment, and boatspeed gains flatten out.

A subtlety that trips people up: a VPP normally references TWS to a standard 10 m height, whereas your masthead unit sits well above that and reads into a stronger, veered layer of the wind gradient. Vertical wind shear can present a markedly higher speed and a different direction aloft than at deck level, so the raw masthead TWS is not the TWS the polar was built against without a height correction. If that correction is wrong, the whole percentage-of-polar reads high or low by a fixed bias.

Prediction is only the starting line. A campaign refines the polar against reality using logged instrument data. In software such as Expedition or a modern instrument processor, the boat records TWS, TWA and boatspeed continuously; the analyst bins the data by wind angle and strength, discards moments the boat was mid-manoeuvre or badly sailed, and keeps the best genuine performance in each bin. Those numbers overwrite the theoretical cells. Over a season the polar stops describing a computer model and starts describing what this boat does in this crew's hands — a far more trustworthy target, and the only version worth steering to.

The VMG geometry, and why the targets are asymmetric

Here is the part that changes how a boat is sailed, and it is not symmetrical — and the asymmetry is a direct consequence of the tangent geometry above.

Upwind, chase a target speed. The polar hands you the boatspeed that yields best VMG for the current breeze. Crucially, near that tangent point the curve is locally flat: because VMG = boatspeed × cos(TWA) and you are sitting at the maximum, the first derivative is zero, so a degree or two of TWA either side costs almost nothing in VMG. That is exactly why the best-VMG angle is a poor thing to steer to — it is insensitive — while boatspeed is the sensitive variable that actually discriminates a good groove from a slow one. So the helm and mainsheet trimmer work together to sit on the target speed: under the number, the driver bears away a touch and the main comes on to build; on the number, the helm edges back up to convert speed into height. A clean target speed is also something a driver can hold by feel through waves, where TWA off the masthead is too laggy and damped to steer to directly. As a rough sense of scale, a well-sorted keelboat's upwind tangent sits around 40-45° TWA and shifts only a few degrees across the usable wind range — narrow enough that speed, not angle, is the lever.

Downwind, chase a target angle. Now the logic flips, and it flips because the downwind curve is steep, not flat. Running boatspeed swings violently with every puff, lull and wave caught — surf a wave on a planing hull and speed can jump three or four knots at essentially the same TWA — so speed is a useless reference. A lot of what looks like "planing" downwind is really surfing on a wave face, a transient the polar cannot capture cell-by-cell. The optimum VMG angle, by contrast, is comparatively steady across pressure. So the crew steers to a target TWA — soaking low in pressure, heating up in the lulls — and lets boatspeed do whatever the waves allow. Getting this backwards, hunting a speed number downwind, is one of the most common speed killers on a fast boat, because you end up steering to a number that is noise.

Underpinning both is the live instrument readout of percentage of polar: boatspeed divided by the polar target for the current TWA/TWS. Ninety-eight per cent and climbing means the boat is well set up; a sudden drop to ninety means something just changed — a header, a wind-shadow, a wave train, a trim that has drifted — and it is a prompt to diagnose now. After racing, the same numbers feed the boat-speed debrief, where a persistent gap in a particular TWS band points to the mode, sail or setup that needs work rather than a bad day.

Failure modes: where the curve stops being true

The single biggest way polars mislead is bad calibration, and it is worth being specific about the mechanism. A polar is only as honest as the TWA and TWS feeding it, and both are derived quantities: true wind is computed by vector-subtracting boat velocity from measured apparent wind, so an error in boatspeed or in masthead alignment propagates straight into TWA. The classic trap is the leeway convention. Designers quote TWA including leeway — that is, relative to the boat's actual track through the water — because that is what the physics balance uses. Many instrument systems, by default, report TWA relative to the centreline (heading). Those two differ by the leeway angle, several degrees upwind, and if they are not reconciled the upwind targets will never line up no matter how the boat is sailed. The standard field check is to sail the same true angle on port and starboard and confirm the computed true wind direction agrees on both tacks; a split means the wind calibration or the upwash/heel corrections are off. Boatspeed must be calibrated against known ground truth (allowing for current) first, because it feeds everything downstream. Calibrate first, trust the polar second — never the reverse.

Beyond calibration, the model's own assumptions set hard edges. A VPP solves for flat water and steady gradient wind; real chop adds resistance and pitching the model never saw, so achievable upwind speed in a seaway can sit meaningfully below a flat-water polar even when the boat is sailed perfectly. Current shifts the ground track but not the water-referenced polar, and confusing the two corrupts the analysis. Dirty air, shear that differs from the assumed profile, and the transient surfing gains downwind all live outside the steady-state solution. None of this makes the polar wrong; it makes it a steady-state benchmark that real conditions modulate in both directions.

Good looks like a crew who knows the two or three target numbers for the current breeze cold, holds within a per cent or two without narrating the instruments, and reacts the instant the percentage sags. Bad looks like sailing to a stale factory polar that has never been checked against the boat, chasing a speed number downwind, or steering on gut in conditions where gut is demonstrably unreliable. A well-used polar is not a straitjacket — a great helm will deliberately sail low and fast in big waves to link surfs, or high and slow to hold a lane off the start — but they always know the target they are trading against and by how much.

On a Grand Prix one-design like the Melges 40

One-design racing sharpens the polar's value to a fine point. In a strict box like the Melges 40 — a Botin-designed, epoxy-infused carbon racer with a foam-cored hull, an electrically canting keel that swings to 45° either side, a large forward canard, twin rudders and a retractable sprit — every hull is nominally identical, so the polar is genuinely shared across the fleet. Speed differences are almost entirely down to how well each crew sails to the map, which makes disciplined target sailing the primary lever you have.

The design also explains why the force-balance sitting behind the polar is unusually favourable here. Public figures put the sail plan at roughly 72 m² mainsail, 49 m² jib and a 200 m² gennaker on around 3,250 kg displacement with 1,200 kg of ballast — an exceptionally high sail-area-to-displacement number. The canting keel does the transverse-balance work: swung to windward it multiplies righting moment (reported as materially greater than a comparable fixed-keel 40) so the M_heel = M_right condition closes at a much higher driving force before the sails need de-powering. Because the keel is supplying righting moment rather than lateral resistance, the forward canard and twin rudders carry the side-force and leeway task, which is why owner-drivers can hold the bow up and push it hard and deep downwind. Published polar reporting has the boat around 17 knots at roughly 20 knots TWS running, sailing near 143-145° TWA in 11-13 knots — the classic case where a fixed speed target is meaningless and steering to a target TWA is the only sane reference, exactly as the tangent geometry predicts.

Treat every one of those numbers as indicative, not gospel. Any specific figure you actually sail to — sail areas, displacement, righting-moment gain, target speeds and the downwind angles — must be verified against the current class rules and the boat's own measured, calibrated polar and documentation rather than taken from general reporting. The right polar for your programme is the class-supplied baseline, refined with your own logged data, guarded by instruments you have calibrated and cross-checked tack to tack.

The takeaway

A polar converts a boat's design potential into a live instruction the crew can act on: the speed to hold upwind, the angle to hold downwind, and an honest percentage that flags trouble the moment it appears. Underneath the curve is a solved force balance — driving force against resistance, heeling moment against righting moment — with real assumptions about water state, wind gradient and calibration that define where it stops being true. Build it from a sound VPP baseline, refine it with your own logged data, guard it with disciplined calibration, and it becomes the objective spine of everything from VMG targets to the boat-speed debrief — the difference between hoping you are fast and knowing it.