How the numbers are made

Every projected time on this site ends with a short phrase saying where it came from. This page explains each one, and how the pieces combine.

The sources, most trusted first

1. inReach fix. A GPS position with a time, from a runner's satellite tracker (Christianne and David carry one). A fix is a fact about where they were at that moment. The device reports about every ten minutes and the feed can lag a few more.
2. Report, entered by hand. Someone tapped In now or Out now, or typed a time, on this site. As good as the person's clock and attention.
3. Report, from ultralive. The race's own check-in system for the 100 only, radioed in by ham operators. The time is when the runner is said to have come through, and it is often written down in a batch a while later.
4. Plan. The runner's own ultraPacer or crew pacing plan: intended arrival times, including their planned stops. Christianne, David, Kristen and Whitney have one.
5. Earlier results. A runner's previous Superior finishes at any distance (and unfinished 100s, section by section), from the race's published results, turned into a pace level for this year.
6. Official results. The timing company's finish list (Athlinks, which the race's ChronoTrack results page shows), checked every two minutes: a finish there counts even when the tracker and the crew missed it. It carries no splits and no drops.
7. History. Twelve years of 100-mile splits, about 2,500 runners, plus five years of 50 and marathon finishes. This is where the fade, the night slowdown, the section difficulty and every ± come from.

The phrases

from tracker

The projection starts at the runner's newest inReach fix on the course. Their pace comes from their own arrivals so far, measured section by section against the history; the stretch since the last station counts once it is more than a mile. Ahead of them, each section's time is the history model at their measured level. The row says how far out that fix was from the station it is projecting and when it was taken, so a number built on a two-hour-old position reads as one. from tracker & plan means their plan is blended in as well: each section ahead is the plan's time (scaled by how they have been running it) weighted against the model's, 70/30 in the plan's favour, dropping to 30/70 if they have been wandering far from it.

from last report

No fresh tracker fix; the projection starts at the runner's last aid-station report, ultralive or entered here, and the row names that station and the time it holds. It starts at the out time when one was recorded (the runner is on the trail from then), otherwise at the in time, in which case the history's typical stop rides inside the next section. Pace and the plan blend work as above. Because a hand-recorded time can be minutes to an hour off, the ± from a report is wider than from a fix.

from plan

Nothing observed yet. The number is the runner's plan, exactly as written, with the history's range around it. The first arrival switches them to one of the phrases above.

from earlier results

Nothing observed yet and no plan, but the runner has finished Superior before. Their earlier finishes are run backwards through the history model into a pace level (recent results and this year's distance count more), and the projection is the model at that level from the gun.

no data yet

Nothing observed, no plan, no earlier results. Rather than invent a time, the page shows a window: from the arrival of a fast runner (the field's 10th percentile) to the station's cutoff.

The ±

Every projection carries a range: the 80% band of the error made by the same kind of projection, from the same station, across every finisher in the history — four runs in five land inside it. The How sure switch in the bar at the top of the dashboard, a station page or a splits page changes it for every page on this phone: 50% (the 25th–75th percentiles), 80% (the 10th–90th percentiles), 90% (the 5th–95th percentiles), 98% (the 1st–99th percentiles). Point at any bracketed range, or tap it, to see all of them for that one projection at once. What the range is for is not missing a runner, so read the earlier end of it as the time to be standing at the station by. At 50% you would miss them half the time, which is why 80% is what the site shows unless you change it; 98% will almost never leave you short, at the price of a long wait. The figures below are for the 50% range, the tightest of them. For the next station that is about ±6–11% of the section time, roughly ±10–17 minutes; for the finish it is hours early in the race and shrinks as they go. Starting from a tracker fix widens it a little (the tracking interval and GPS scatter), and it narrows as the runner closes on the station, never below a few minutes. A runner steadier than the field earns a narrower band. The band is honest, not conservative: the truth lands inside it about half the time.

Time of day, or from now

The switch at the top of the dashboard, the map, a station page or a splits page changes every time on the page between the clock and how long from now: in 1:20 for something ahead, 1:20 ago for something past, and a projection's brackets become the two ends of the wait. The choice is remembered on this phone. A splits row's race elapsed — the classic 6h45 split — is in the row's details, beside the paces.

The words on the rows

• moving, stopped: what the inReach shows from its last two fixes (a runner standing at an aid station reads "stopped"; so does someone sitting on a rock). stale 47m: the newest fix is older than the tracker's interval allows, so the device may have lost the sky or been switched off. no fix: the feed holds nothing for the window.
• feed unreachable / tracker feed unreachable: Garmin's site could not be read; everything from reports and check-ins still shows.
• in / out: arrival at and departure from a station; "(12m here)" is the stop. passed, no time recorded: a later station has a time for them, so they went through this one, but nobody caught the time.
• no data yet: nothing observed, and no plan or earlier results to project from; the window shown runs from the earliest plausible arrival to the cutoff. no report yet: the race has started and nothing has been seen of them.
• overdue: the projected arrival has passed with no word; they are somewhere before that station. due now: within a minute or two of it.
• † on the splits page: a time taken from a station report (ultralive or entered by hand) rather than the tracker. !: a cutoff at risk on the projection. ● after a station: crew can be there.
• after the cutoff, PAST cutoff: the projected arrival is later than the station closes; the margin shown is how much.
• from a report vs tracker on a recorded time: whose word it is; see recorded times.

Recorded times

A time in the past with no ~ is recorded, not projected. tracker means interpolated between two inReach fixes either side of the station; from a report means an in or out that was entered here or came from ultralive. † on the splits page marks the latter. When both exist for a station, the tracker's arrival is used and a stated out is taken as fact.

What the model knows about the course

Section difficulty comes mostly from how twelve years of runners actually ran each section, not from grade; the climb per mile is in the model but carries little weight, because the roots and rock matter as much as the hills. Runners slow with the miles they have run (about 1.7× by mile 50), and about 7% more in the dark. A 50-miler at Sugarloaf has run 21 miles, not 70, and is projected that way; each distance is calibrated so the model's median finish matches that race's real median.

Cutoffs, crew, drop bags

All from the race's aid station chart dated 8 August 2026, transcribed by hand. The chart says it is subject to change; the race's own materials win over anything here.