Features
One product for the whole race — before, during and after
SailTac is a small server that runs on the boat's Windows or Mac computer. It reads your NMEA instruments and serves a fast interface to every device aboard — offshore, round the cans, or on a practice day. Here is what it does, roughly in the order you'll meet it.
01 · Plan
Weather routing that admits what it doesn't know
Every routing tool draws a confident line across the chart. SailTac draws the same line, then colors each leg by how much the forecast models actually agree with each other there. Green means they concur. Red means the routing is a guess, and you should keep your options open.
In the routing engine
- Import your polar from your ORC certificate by reference, sail number or boat name
- Isochrone optimal routing over your boat's own polar, with land avoidance
- Multi-leg courses that route mark to mark, chaining each leg's ETA into the next leg's forecast
- Mark rounding-side constraints — leave to port or starboard
- Plan a start up to 16 days ahead and route that moment's forecast
- Expedition-style leg table: ETA, distance, tack split, average boat speed and the sail called for each leg
- Animated wind overlay with a time scrubber — play the whole race forward
- Point it at the sailing instructions and it reads the course out of them — marks, rounding sides and the sequence
- Save plans, reload them, and overlay the race you actually sailed on top
- A printable pre-race brief for any saved plan — the timeline and the helm plan, to hand round the night before
- NOAA and CHS ENC charts, cached for the whole course in one click before you leave the dock
- Bring your own charts anywhere else — import unencrypted S-57 cells, a .000 base cell or a zipped exchange set
- Tactical reference, not an approved ECDIS: the charts are recoloured, and encrypted S-63 cells aren't supported
- Import the GRIB files you already get by Iridium — GRIB 1 or 2, cropped to your course, needing no connection at all
- Imported GRIBs stay separate: never blended with the fetched models, never scored in the accuracy scorecard
The ten models
Each model has a different resolution and a different horizon. SailTac tracks both: a 48-hour high-resolution model is used for the hours it actually covers and dropped after that, rather than being blended in as calm air and quietly dragging your forecast down. Models with no data where you're sailing aren't offered at all, so you never tick one that stays silent.
| Model | Source | Grid | Horizon |
|---|---|---|---|
| Global | |||
| Six independent forecasts, worldwide, long enough to cover an entire race. | |||
| ECMWF | ECMWF IFS (European) | 25 km | 360 h |
| AIFS | ECMWF AIFS (AI forecast) | 25 km | 360 h |
| GFS | NOAA GFS (US global) | 13 km | 384 h |
| ICON | DWD ICON (German) | 11 km | 180 h |
| UKMO | UK Met Office global | 10 km | 168 h |
| GEM | ECCC GEM (Canadian global) | 15 km | 240 h |
| Regional | |||
| Higher resolution over North America. They sharpen the first day or two, then hand back to the globals — and SailTac only offers the ones with data where you're sailing. | |||
| HRRR | NOAA HRRR rapid-refresh | 3 km | 48 h |
| NBM | NOAA National Blend | 2.5 km | 264 h |
| NAM | NOAA NAM mesoscale | 12 km | 60 h |
| HRDPS | ECCC HRDPS (Canada) | 2.5 km | 48 h |
NBM is a post-processed blend of other NOAA models rather than an independent forecast — it's there for its resolution over a long horizon, and SailTac's ensemble knows not to count it as a second opinion.
02 · Race
Instruments for the whole crew, not just the nav station
The interface is served over the boat's Wi-Fi, so the tactician on the rail, the trimmer at the winch and the navigator below are all looking at the same live data on their own devices, each with the layout they want.
On the water
- Live displays on every phone, tablet and laptop on the boat's Wi-Fi
- Custom dashboards built from 49 widgets, dragged and resized, saved per crew member
- Day, night and high-contrast themes — red-on-black for the dark watches
- One-touch race start that freezes the forecast for later scoring
- 1 Hz logging of position, wind, speed, heading, helm and sail
- Automatic mark rounding and course advance while you race
- Lift and header tracking with a tack-relative reference that resets on each tack
- Target heel as a band per point of sail, with live error against it — the numbers come off your tuning guide, never invented for you
- Magnetic or true heading, with the World Magnetic Model built in
- Crash-resume: kill the power mid-race and it picks the same race back up
The screens you'll use
Racing
The numbers that matter on the rail, big enough to read wet.
Start
RRS 26 sequence, line bias, pin and boat pings, time to gun.
Course
Leg sequencer with auto-advance and overstood layline warnings.
Wind
True and apparent wind, shifts, and the lift/header picture.
Navigation
Range and bearing to the active mark, ETA, VMC, depth.
Performance
Boat speed against target, by point of sail and wind band.
Route
Weather routing, the leg table, saved plans and the wind overlay.
Races
Every race you've logged — analyze, report, recap, results.
Helm
Who's driving, and how each of them does in what conditions.
Dashboard
Build your own screens from 49 widgets, saved per crew member.
Sails
The crossover chart, live: the right sail lights up.
Trim
Target heel band for the point of sail you're on, and how far off you are.
Calibration
The guided reciprocal run that tells you what your paddlewheel reads wrong.
Water
The current this boat actually measured, a vector every fifteen minutes.
Polars
Import from your ORC certificate, a CSV or a .pol file — or edit the table by hand.
Sail record
Every sail's lifetime: hours, load, time over its rating, service state.
Forecast record
Which models have actually been right, in your water, over seasons.
Underneath all of it
Is the paddlewheel telling the truth?
Every percentage of polar in every report is computed from boat speed. If the transducer reads 4% slow, so does all of it — and the error hides as a permanent quarter-knot of adverse tide that no water ever produced. SailTac measures it two ways and tells you the number to type into your instrument system.
Solved while you sailed
A speed error scales with boat speed and is the same on every course; a current is the same vector whatever the boat is doing. Change of heading separates them, so a race with enough spread in it solves both at once — no special procedure, no slack-water day.
A guided reciprocal run
Sail a heading, turn 180°, sail it back. The pair cancels the current exactly, which is why this one works in tide rather than needing a still day to be honest. The timer counts accepted seconds, so it stalls when the helm wanders off the heading.
What the tide actually did
Checking the paddlewheel produces a real current vector every fifteen minutes as a by-product. Where the harmonics already exist that is a curiosity — and on water nobody publishes predictions for, it is the only current record there is.
The correction is reported as a finding, never as time lost: reading 4% slow doesn't mean the boat lost 4% of the race, it means the measurements were off by that much. And it tells you to enter it in the instrument system's own calibration rather than in SailTac — that is where it also fixes the apparent and true wind derived from boat speed, and the polar you build next season.
03 · Analyze
The argument on the dock, settled with evidence
Did we lose that race in sail selection, helming, maneuvers or routing? Was the staysail worth flying? Which model should we have trusted? Who should drive the night watches? SailTac answers all four from the log — and tells you when it can't.
Rather than take our word for any of that — read one. 8 pages from 2026 Port Huron → Mackinac, Cove Island course, printed from the shipping build against the real log. No email required.
Every figure has a basis
Maneuver loss is measured from the real speed drop through each logged tack, not from a rule of thumb. Routing loss reports “not measurable” if you didn't save a plan to compare against, rather than inventing a figure.
Estimates that overlap say so
The loss attributions are independent overlapping estimates, and one can exceed the total — fixing your staysail use and your helming together would push you past polar. The report states this instead of adding them up into a satisfying, wrong number.
Compressed demos aren't scored
Run a sped-up demo race and the model scorecard refuses to grade it, because the log times don't line up with a real forecast. It says so and waits for a real race.
After the finish
- Automated debrief attributing time lost to sail selection, helming, maneuvers, staysail use and routing
- Per-forecast-model accuracy scorecard, ranked on true wind speed and direction error
- Polar plot with your sailed scatter over the target curves
- Pinch / on / foot angle discipline against the polar's optimal VMG angle
- Wind-shift strip with tack markers and a shift-play verdict
- Sail-change timeline and per-sail wind and angle envelopes
- Staysail effectiveness versus the same primary sail without it
- Helm performance by point of sail, wind band, and day versus night
- VMG to the wind axis and to the mark, plus along-track current
- Boat-speed calibration solved from the race itself — how far off the paddlewheel reads, and by how much that moves every percentage of polar in the report
- The current the boat actually sailed through, solved per fifteen-minute window rather than taken as boat speed minus SOG
- A lifetime forecast record — model skill accumulated race after race
- A lifetime record for every sail: hours, load, exposure and service state
- Thresholds instead of estimates: a sail flown under ten minutes gets its hours and no score, a point of sail under 5% of the race gets no verdict, nothing under five knots of breeze is graded on percentage of polar, and a difference under 0.3 kn is called noise rather than a result
- Multi-page printable PDF report, and a 4:5 recap card for socials
04 · Keep
Two records that get more valuable every season
Most of what SailTac does answers a question about one race. These two answer questions no single race can: which forecast to believe here, and what your sails have actually been through.
Forecast record
Which models have actually earned your trust, in your water.
Every race adds to it. Each model is scored against the wind you really sailed through — hour by hour, like for like — and the record accumulates across seasons until it can say something a single race never could.
It grades honestly. Error is measured against persistence, so a model only scores well by beating the assumption that nothing changes. Bias is signed, so you can see a model that always blows ten degrees right. And nothing is called a winner until a paired test says the gap is real, with the race — not the hour — as the independent unit. A single breezy afternoon can never crown a model.
Where the honest answer is that no model saw it, it says so. On a windward-leeward course the whole track sits inside one forecast cell, so the record reports that every model was flat and shows the gradient your boat measured instead.
Sail record
A lifetime history for every sail in the wardrobe.
Hours flown, load-equivalent hours weighted by the sail's own area, wind and angle envelopes, time spent over its rated maximum, normalised speed against the polar, and how long since its last service.
Load is measured against each sail's own rating, because absolute load-hours aren't comparable across cloth weights — an A5 is built for twenty knots and accrues load by design, while an A1 in eight knots may be nearer its own limit. Without that correction the numbers read as one sail abused and another pampered when both were doing their jobs.
It prints as one document for your loft: a summary table covering the whole wardrobe, with opt-in detail per sail and CSV alongside. Sails the boat owns but no log has measured get their hours and load and no speed figure at all, rather than borrowing a number from a sister sail.
Not only distance racing
Three kinds of sailing, the same loop around all of them
A session is one of three things, and SailTac knows which it is looking at. Mark rounding, the report's shape and what counts as a maneuver all change to suit — because the same analysis run over a Tuesday-night windward-leeward and a three-day offshore race would be wrong in at least one of them.
Offshore and distance
Long courses, watch systems and a forecast that matters for days. Routing leg to leg, sail changes across a whole wardrobe, and helm analytics that separate day from night because the same driver is a different driver at 3am.
Windward-leeward
Buoy courses are a first-class race type, not a long race with the legs shortened. Tighter mark-rounding geometry, beat-by-beat shift play against the wind's oscillation, and a report that leads with what actually decides a W/L — the starts, the shifts and the tack count.
Practice
A training day is its own mode. Drills instead of a course, maneuvers priced per repetition, and how many seconds the boat took to get back to the speed it went into the turn with — which is the number that improves with practice and never appears in a race result.
Reliability
Because it has to keep working at 3am, 90 miles out
The classic failure mode of marine software is shipping something that breaks offshore. Everything below exists because a real race needed it.
Reconnects itself
Instrument sources reconnect after a power blip with a half-open watchdog and exponential backoff. Your NMEA configuration is saved to disk, so a reboot mid-race comes back on its own.
Append-only logging
Race samples are appended and flushed every 15 seconds. A hard crash costs you seconds, not the race.
Crash-resume
Kill the app mid-race and it resumes the same race on restart, still logging.
Health panel
Every source with its data age, whether data is flowing, log growth, memory and whether the disk is writable. No guessing.
Tested for length
A 72-hour, 259,000-sample race is in the automated test suite — analysis, export and all. Multi-day races were the design target, not an edge case.
Demo mode that can't lie
The built-in simulator suspends real instruments while it runs, banners the whole app in purple, and is always off after a restart. Simulated data can never be mistaken for yours.
See it on your own instruments
Download it free, run the built-in demo on the kitchen table, then plug it into the boat. The trial is 30 days and every feature is in it.