What the numbers mean

Under the route there is a band with up to four numbers in it, and most of them carry a small figure beneath the value. This page explains all nineteen numbers and all nine figures: what each one is, which measurement it is drawn from, when it stays empty, and why it has the shape it has.

We are showing you the machine rather than the result. A number printed on a wall is a claim, and the only way you can tell whether ours is worth defending is by being able to check how it was arrived at.

Some of it is duller than the poster deserves. A section on why running splits are cut by distance instead of by measurements is not going to sell anybody a print — but that decision is the difference between bars that are split times and bars that merely look like them, and you cannot see which of those you were sold by looking at the sheet.

What decides how many of the nineteen you can use is your file, not this page. What your file carries maps that out source by source.

What the numbers mean

Nineteen numbers can go under the line, four of them at a time. Fifteen can draw a small figure beneath the value; four never can, and say so here rather than leaving you to wonder. This page is the whole machine: what each number is, where it came from, when it stays empty, and why each figure has the shape it has.

The nineteen

Pick up to four. The band under the route is a fixed height whatever you choose, so a fifth column would squeeze the digits below what prints legibly — and a poster you cannot read across the room is not the product. A number you picked that your file cannot answer drops out rather than printing an empty slot: choose four, get three answers, and three cells is what goes on the sheet.

On the print	Unit	Where the number comes from	When it stays empty	Its figure
DISTANCE	km, one decimal	The file's own total where it states one, or Strava's. Only when neither does do we measure along the track — and then along the whole recording, before the privacy radius trimmed the ends off it.	Never, as long as there is a route to draw.	Against your longest
ELEVATION	metres climbed, whole	The source's own climb: the barometer in your watch, or Strava's figure for the same ride. The two often disagree and neither is wrong — they measure different things. We work one out ourselves only when neither states it.	Under 25 metres of climb. A flat ride prints nothing rather than 0 HM, which reads as a measurement of nothing.	Silhouette
ELAPSED	h:mm	Start to finish, pauses and coffee included. The source's own figure, or first timestamp to last. For a file with no time at all, this is the one number you may type in yourself.	A downloaded route with no timestamps, where you did not type a time in.	No figure
MOVING	h:mm	Clock running only. FIT and TCX state it; otherwise it is worked out from the gaps between your own points.	A file with no timestamps. A time you typed in never fills this one — a route cannot tell anybody how long you stood still.	Share of elapsed
AVG PACE · AVG SPEED	min/km · km/h · min/100 m	Strava's own average when the account is connected, because you have their screen open next to you and a number one decimal off ours looks like a fault. From a file, distance over moving time. The unit follows the sport: minutes per kilometre for running and walking, per 100 metres for swimming, km/h for riding, skiing and paddling — and the label changes with it.	No distance or no time. Also on a multisport file: one average across a swim, a ride and a run is a speed nobody swam or rode.	Splits, or a curve
AVG POWER	watts, whole	A power meter, through your file or through Strava. A tilde in front of the number means Strava estimated it from speed, weight and gradient rather than measuring it.	No meter. Nothing about a route can produce a wattage, so we leave it out rather than inventing one.	Curve
AVG HR	bpm, whole	The source's own average. Where there is none — a GPX carries readings but no summary — the mean of the file's own readings. That is the file's answer worked out from the file's own numbers, which is a different thing from a guess.	No strap, no chest sensor, no wrist reading.	Zone distribution, or a curve
AVG CADENCE	rpm, whole	Same rule as heart rate: the source's own average first, otherwise the mean of the file's own readings.	No cadence sensor.	Curve
STARTED	hh:mm, local	The first timestamp in the file. Files are written in UTC, so the offset to local time is estimated from the longitude you started at — the only timezone information a GPX carries at all.	A file with no timestamps.	Dial
TEMP C	°C, a range	Your device's own thermometer, read across the whole ride. It prints as a range rather than an average: a single 17 says nothing about a ride that began at eleven degrees and ended at twenty-three, and what you wore was decided by the range.	Devices without a thermometer — and every ride that comes in through the connected account, because Strava's streams do not carry temperature.	Curve
L / R SPLIT	left/right, whole percent	A pedal meter that can tell your legs apart. There is no second route to this number.	Anything short of that meter.	Pair
BURNED KCAL	kcal, whole	Your device's own estimate, printed exactly as it was written. We do not compute one — it would be our guess about your body.	GPX, which has no room for it, and rides that come in through the connected account.	No figure
STEPS	steps, whole	Your watch's own count, doubled: a FIT file counts strides, and a stride is two steps. Printed as it was counted — we do not divide your distance by an average step length, because that would be our guess about your legs, not your day.	Anything that is not a FIT file from a run, a walk or a hike. A bike computer counts pedal turns rather than steps, and a connected Strava account cannot pass steps to us at all — not even for a walk where the Strava website shows a number.	No figure
MAX POWER	watts, whole	The peak in the source's own summary, from a file or from Strava.	No meter.	Peak on the power curve
MAX HR	bpm, whole	The peak in the source's own summary. Unlike the average, this one is never worked out from the readings — a single spike from a sensor hiccup would become the headline.	No strap, or a source that records the readings without summarising them.	Peak on the heart rate curve
MAX SPEED	km/h, one decimal	Your device's own top speed, from the file.	GPX, and rides that come in through the connected account.	Peak on the speed curve
DESCENT	metres descended, whole	The device's own figure where it states one, otherwise worked out from the elevation the file recorded. That fallback was calibrated against real recorded rides rather than guessed at.	Under 25 metres, same floor as the climb.	No figure
AEROBIC TE	score out of 5	Your watch's own verdict on what the session did to your fitness, computed inside the device from your history. The ceiling prints with it, because the score alone does not say whether it was a lot or a little.	Devices that do not compute it — and files that carry a placeholder instead of a reading, which we drop rather than print as a score.	Share of the ceiling
ANAEROBIC TE	score out of 5	The same verdict on the anaerobic side, from the same device. Along with the aerobic score, these are the only two numbers on the sheet that were not measured but judged.	Devices that do not compute it.	Share of the ceiling
COORDINATES	degrees and minutes	The middle of the track, averaged across its points — and the track has already had the privacy radius cut from both ends, so this is where you rode, not where you live.	A recording with no positions, which has no line to draw either.	No figure

THE DATE IS NOT ONE OF THE NINETEEN. IT PRINTS ON THE LINE ABOVE THE ROUTE, ALWAYS, SO IT NEVER TAKES ONE OF THE FOUR SLOTS FROM SOMETHING THAT WAS MEASURED. A TILDE IN FRONT OF A WATTAGE MEANS STRAVA ESTIMATED IT RATHER THAN A METER MEASURING IT. ONLY THE LABELS ARE YOURS TO CHANGE — THE VALUES COME FROM YOUR FILE.

Nine shapes, and why each one

A figure is fourteen millimetres tall on a 50×70 and one column wide. It has room to answer one question, so each shape is chosen for the question its number actually raises — not for variety. All of them are drawn in ink, never in the route's colour: if the figure takes the colour of the line, two things on the sheet claim to be the important one, and the important one is the route.

Curve
The reading through the ride, left to right.
Drawn under Avg power · avg cadence · temp · avg speed on a ride · avg HR without zones
Each sample is the average of its slice of the ride, not a spot reading. Spot readings were tried first: a power meter swings a hundred watts between one second and the next, and picking single moments turned the curve into a hairball — noise in a box a few millimetres tall, not the shape of the ride.
The scale is the series' own low and high, not zero. Nobody rides from nought to their maximum, so an absolute scale would flatten almost every ride into a line near the top and answer nothing.
Silhouette
The height profile, filled down to the baseline.
Drawn under Elevation
Filled rather than stroked, because climb is a mass you accumulated — not a value that swings up and down. A line would read as a reading; the filled shape reads as the ground.
The cell does not carry its own copy of the profile. It asks the drawing engine for the same elevation series the profile band uses, so the two pictures of the same ride cannot end up disagreeing.
Splits
One bar per kilometre, in order through the run.
Drawn under Avg pace on a run or a walk
A curve of pace is hard to read: it is jagged, and the eye goes looking for a level rather than a shape. Splits answer the question you actually have — which kilometre was quick — and they are how a runner reads their own run.
The count is not rounded to something tidy. A half marathon is twenty-one splits, and eight bars claiming to be splits would be a different thing wearing the same name. If they come out too thin, pick fewer numbers: three cells make each column a third of the sheet wide instead of a quarter. That is your decision, not a rounding we do behind your back.
Share
A part of a whole: the outline is the whole, the fill is the part.
Drawn under Moving · aerobic TE · anaerobic TE
Moving time and elapsed time used to sit in two cells with a clock each, and the difference between them — how much of the day you spent standing still — was arithmetic left to the reader. As a share it is the picture.
The training effect scores use the same shape but measure against 5.0, because that is a ceiling rather than a normal. Measured against your own average instead, a 3.6 would draw as a bar filled well past halfway and mean something else entirely.
Against your longest
This ride as a share of the longest one you have.
Drawn under Distance
The track is twice the reference, so your longest ride sits at the midpoint: a ride like all the others fills half the bar, one twice as long fills it. Without that midpoint a half-filled bar is just a half-filled bar.
It needs the connected account, because a single uploaded file has no history to compare against — then distance prints as a plain total with no figure. It also disappears when this ride is the longest: there is nothing to measure it against. The bar always carries a line of text saying what it is measured against; that line is the figure's axis, and without it the bar makes a claim it cannot support.
Dial
How far into the day you set off, clockwise from midnight.
Drawn under Started
A day is a circle, and that is the whole reason for the shape. Eleven at night and one in the morning are two hours apart, not twenty-two — a bar or a curve would insist otherwise, and be wrong about it every night.
It is drawn as two pieces: the ring as a stroked outline, the elapsed part as a filled wedge. The first version put both in one path, and since a figure is filled, the whole dial came out as a solid disc with the time invisible inside it.
Pair
Two shares of one whole. Left leg filled, right leg as outline.
Drawn under L / R split
Two filled shapes would need two colours to tell them apart, and the figures are drawn in one ink so that nothing on the sheet competes with the route. Filled against empty carries the same distinction with one colour.
It is one number — your left leg's share — and one number is exactly what a chart is normally built to refuse. See the floors below.
Zone distribution
Five bars in fixed order: how much of the ride sat in each heart rate zone.
Drawn under Avg HR, once we know your maximum
Scaled from zero, not from the smallest bar. Two per cent and fifty-nine per cent have to look like thirty times the difference; scaled between the extremes, a ride spent almost entirely in one zone would look evenly spread.
Position names the zone, since there is no colour to do it, so the bars stand in fixed order from one to five and a zone you never entered keeps its place as a stub. A line under the figure says where the boundaries came from — your own Strava zones, or a maximum you gave us, or one estimated from your age.
Peak
A dot on the series the number is the peak of.
Drawn under Max power · max HR · max speed
A peak has no shape of its own — a single number in an empty box says nothing — so it borrows the curve it came off. The dot then tells you something the number cannot: that the spike came on the last climb rather than in the first ten minutes.
Which is also why the peak is never drawn alone. Without the curve behind it, a dot floating in a cell is decoration.

The decisions that are not obvious

Most of the work in these figures went into things you would only notice if we had got them wrong. Here they are anyway — a poster is a claim, and a claim you cannot inspect is just a nice-looking assertion.

Splits are cut by distance, not by measurements

A kilometre run slowly contains more measurements than a kilometre run quickly — the watch is recording all the while. Cut the series into equal numbers of measurements and the slow kilometres come out wider than the fast ones, which means the bars are no longer split times. They just look like them.

So the cuts follow distance. Where the file carries its own running distance — most do — we cut exactly where each kilometre fell. Where it does not, the cuts are worked out from speed, which is close but not the same measurement, and that difference is why we prefer the file's own.

Pace and speed are one series read in opposite directions

There is only one measurement underneath: speed. On a ride the label says AVG SPEED in km/h, and higher is better, so the figure is drawn as it stands. On a run the label says AVG PACE in minutes per kilometre, and lower is better, so the figure is turned upside down and the fastest kilometre becomes the tallest bar.

Get that backwards and the poster contradicts itself: the tallest bar would sit where the running was slowest, under a word saying the opposite. It is the one trap in the whole set, and it is why the same number wears two different labels depending on what you were doing.

A figure is pieces, not one line

Every figure comes back as a list of parts, each with its own treatment: stroked outline, or filled area. That sounds like plumbing, and it is the reason the half-filled bars work at all.

The first version put the track and the fill in a single path. Since a figure is drawn filled, a bar that was meant to be half full came out solid black — the ratio, which was the entire point, was invisible. An outline and an area cannot be the same element.

Two floors that reject single-number figures

A chart normally needs at least two points; drawing one is meaningless. That rule sits in two places in the drawing code, and both had to agree.

The dial and the pair each have exactly one number — minutes since midnight, and your left leg's share — so they fell through both floors and drew nothing at all. Nothing failed; the cells were simply empty. Both floors now name the shapes that are allowed to have a single number, in one list rather than a check per figure, so the next one does not have to discover the problem again.

Zones are a calculation, not a measurement

Your watch measured your heart rate. Nothing in any file records which zone you were in — the zones are that heart rate divided up by boundaries that belong to you, and 150 beats is zone two for one person and zone four for another. Strava does exactly the same thing, and says so on the screen.

So the figure needs a maximum. If your account is connected and you have set your own zones there, we use those — anything we compute would be a worse answer than the numbers you set yourself. Otherwise we ask for your maximum and suggest one from your own hardest rides rather than the highest single reading, which is often a sensor hiccup. Failing that, 220 minus your age, which is a statement about an average body: on our own files it lands sixteen beats below what has actually been recorded, and every zone boundary moves with it. When that is the number in play, the figure says the zones were estimated.

Your ceiling is not the same in every sport

Maximum heart rate on a bike sits below the running figure — you are seated and using less muscle — and in water the gap is wider again. Use the running number on a ride and every zone boundary shifts up, and the ride looks easier than it was.

The adjustment sits on the maximum, not on the percentages. The percentages describe effort and are the same whatever you are doing; it is the body's ceiling that moves.

A heart rate series can be broken, and we would rather draw the curve

One real run in our own archive carries over two thousand readings and only twenty-six different values, more than a thousand of them the same number. That is a sensor holding its last reading for twenty minutes, not a heart that sat still.

A zone distribution computed from that looks precise and confident and is fiction. So below fifteen distinct values we draw the curve instead: it shows what the file recorded without claiming what it meant.

You can switch a figure off and keep the number

Every number that can draw one gets its figure without you having to ask. Pick four and you can get four figures in a band where each column is a quarter of the sheet wide, which is not always what you want — so each one has a switch.

Turning it off leaves the value where it was. You are declining a drawing, not a number; dropping the number is a different decision, made in the list beside it.

Everything on this page is read out of the code that draws the print — not written from memory. If a number here disagrees with your poster, the poster is right and this page is a bug.

What your file carries Six steps to a print

Every figure answers one question

And says which one it is.