The best routing starts with evidence.

Our ambition is to set the state of the art in sailing weather routing: the best decisions, with assumptions you can inspect and results you can reproduce.

Where ORC RaceFlow 2.0 fits

ORC RaceFlow 2.0 combines the ORC Velocity Prediction Program and certificate database with PredictWind’s routing engine. Its Weather-Routed Scoring (WRS) simulates a fleet’s passage through forecast conditions to produce race-specific time correction factors. RaceFlow 2.0 is a scoring platform, not the name of a new ORC routing algorithm.

Plan the passage. Inspect the inputs.

Offwind gives sailors an independent routing workspace. Use an ORC certificate polar or your own boat-performance table, compare forecasts and departure times, and inspect currents, waves and depth constraints. A route is useful when you understand the weather and boat assumptions behind it.

Plan a route

Check the scoring arithmetic yourself.

Our independent WRS checker compares published predicted elapsed times with published Time-on-Time and Time-on-Distance factors. Open a RaceFlow export on your device, inspect missing inputs and timing ambiguities, and download the report. Your export stays in your browser.

Open the RaceFlow 2.0 WRS checker →

Behind the rival? The fastest average route may not be your best move.

Being behind your own forecast is different from being behind a competitor. If both boats lose the same ten minutes, their relative position barely changes. To catch a rival, you need an attainable advantage in corrected finishing time. A different weather exposure can create that opening, even when it costs time on average.

Staying alongside a similarly performing rival preserves a deficit. A branch that is slower on average but more variable can win back ground in some outcomes, and can throw away a lead in others. If the rival covers, the opening disappears.

This is an objective choice: fastest expected finish, chance of beating a particular rival, and best series result are different questions. Published research on match-racing tactics and on deadline-dependent sailing under uncertain wind makes the same distinction. Neither shows a defect in ORC WRS.

Interactive example · illustrative weather

When is a slower route a better bet?

Two boats face the same range of weather outcomes. Staying nearby shares the rival’s weather. The alternate branch costs time on average but meets a different weather exposure. These inputs illustrate a mechanism; they are not fitted to a race.

Projected deficit if you stay: 10 min Negative means you are ahead on corrected time. Alternate branch’s mean cost: 5 min Branch weather swing: ±25 min Rival covers the alternate branch

Stay with the rival

Wins in 0% of weighted outcomes; ties in 0%.

Mean time remaining: 190.0 min
Mean corrected deficit: 10.0 min
Finish spread: 0.0 min
Loss in the worst outcomes: 10.0 min

Take the alternate branch

Wins in 50% of weighted outcomes; ties in 0%.

Mean time remaining: 195.0 min
Mean corrected deficit: 15.0 min
Finish spread: 25.0 min
Loss in the worst outcomes: 40.0 min

Fastest on average: Stay with the rival. Most wins: Take the alternate branch.

Each choice is made before the weather is revealed. Both boats share the weather, and covering means the rival follows the same detour. A share of outcomes is not a calibrated chance of winning. Safety constraints, other boats, series points and later maneuvers are outside this example.

The scoring hypothesis worth testing

A forecast-derived rating can reward or penalize a boat type if the issued forecast misprices its performance relative to other boats. The question is whether those relative errors are predictable before racing, large enough to matter, and repeatable on unseen events. A legal boat configuration or course policy that exploits a persistent mismatch would be an edge. Finding a favorable weather realization after the finish is not evidence that the edge was available beforehand.

Once a positive Time-on-Time factor is fixed, a faster elapsed time is still a faster corrected time for the same boat. What can change is the spread of your finish relative to a rival. Sharing weather can protect a lead; separating can create both recovery and larger losses. Lower correlation alone does not guarantee a better bet: the expected detour cost, the spread of outcomes, the rival’s response and the observations available all matter.

What our existing evidence shows

The public ORC sample we checked agrees with its published scoring arithmetic. It contains neither verified forecast errors nor a tactical win-rate benchmark.

Our archived race experiment with one rolling forecast policy was faster in some cases, slower in others, and slower on average. It does not show a general advantage, and the savings it recorded are realized differences against a baseline, not chances of overtaking another boat.

A diagnostic on archived routing decisions

We re-evaluated the first recorded decision in a set of scored race cases, holding each candidate route fixed while replaying the available forecasts. For this diagnostic, a virtual rival follows the deterministic baseline route, and we add a hypothetical time debt to our boat. This tests the objective on real routing outputs, not the behavior of a real trailing yacht.

Under a hypothetical debt, the option that is slower on average sometimes has better odds against the virtual rival than the fastest-on-average option. That is a diagnostic, not a measured win rate, and it is not a calibrated chance of winning. A short local wind-verification campaign was too small to settle which forecast is better, and it does not establish a routing or racing advantage. The next test is independent tactical replays.

From a tactical idea to a measurable edge

Candidate tactics need to be judged against independent race outcomes, using only information that was available at the time, before we make claims about them. Until that is done, we make no claim about a tactical edge.

A standard we intend to prove.

The reviewed public RaceFlow sample agrees with the published scoring arithmetic. That result validates the narrow arithmetic check; it does not establish superior route accuracy or fairer race results. Our research work covers parts of this question, and it is separate from the public planning workspace.

To establish a routing advantage, comparisons need matched conditions and independent outcome evidence. We have not yet published that benchmark. “Best” and “state of the art” describe our ambition, not a measured victory over ORC or PredictWind.