You hit one 300-yard drive. Congratulations — it happened, it counted, it's on your board.
You are not a 300-yard driver.
That isn't a dig. It's the most useful idea in golf statistics, and once you have it, half the numbers you've ever been handed stop being confusing. A number is an answer. Different numbers answer different questions. Most arguments about golf data are two people holding two correct answers to two different questions and each insisting the other guy is wrong.
Here's how to tell which question a number is answering — including the ones we show you, where we've had to make some awkward choices out loud.
The average is the one that lies
Start with the number every launch monitor puts at the bottom of the sheet.
Somebody on this team once pulled apart a real launch monitor card, the kind you get handed at a demo day. Eight 3-iron shots. The carries:
> 227.4 · 214.4 · 80.1 · 199.5 · 227.7 · 84.8 · 217.9 · 141.4
Four strikes, three duffs, one that got away. Average those carries and you get 174.1. The card's own headline read Avg. Total 189.7.
Now look at the eight again. Not one of them landed within twelve yards of the average carry. The average describes a shot he did not hit, has never hit, and could not hit on purpose if you paid him.
That's not a freak case, it's what an average does to golf. Golf shots aren't scattered gently around a middle — they're two piles. There's the pile where you caught it and the pile where you didn't, and an average splits the difference and hands you a number sitting in the empty air between them.
So we don't use them
House rule, written into the code that builds every club number here: medians and percentile bands, never means. The comment at the top of that file says it in two words — never means.
A median is the number with half your shots above it and half below. One shank barely moves a median. One shank moves an average a long way. That's the entire argument, and it's why one bad swing can't reach into your distances and change what club you pull next Tuesday.
Where we say "Avg." and mean something else
Here's the awkward part, and you should hear it from us rather than work it out.
There is a row on the session card titled "Avg. Total." It is not an average. It's a median. The registry entry that defines that row names its estimator `median` in the same breath as it names the title "Avg." — the title kept the word golfers actually say, and the number underneath is the honest one.
We're not thrilled about it. The reason it survives is that the label right beside it always states the truth: median of N measured. Every one of those numbers arrives wearing a note saying how it was computed and how many shots it came from. So the word "Avg." is loose shorthand and the fine print is exact — but if you only read the big word, you'd have the wrong idea, and you'd have gotten it from us.
There's a related line you may see on an older frozen session card, and it exists for exactly this reason:
> This card is kept exactly as it was issued. Its numbers are means. The page above uses medians and is live — the two never merge into one figure.
Two numbers, both about the same session, different by a little, and we refuse to average them into one. That's the posture. When two things are computed differently we'd rather show you the seam than sand it flat.
Your club number is a median of twenty shots
Now the bigger one.
When the app tells you what a club does, that number is the median of your 20 best-struck shots with it. Not all your shots. Twenty.
That's two departures from what most people expect, and both are on purpose.
It's a median, not an average — for the reason above.
It's your best twenty, not everything — because "what does this club do" is a different question from "what did all my swings do." A fitter watching you hit balls does not average your shanks into your seven-iron number. He also doesn't throw the shanks away; he reads them separately, as your miss pattern, which is real and useful and a different question again. Two questions, two groups of shots. Mixing them is exactly how members end up with numbers they don't believe.
Here's the size of that gap on one real driver in this system. Take every qualifying shot ever hit with it and find the middle carry: 205. Take the median total of the 20 best-struck: 244.
Thirty-nine yards apart. Both computed correctly. Both true.
The 205 was wrong twice over as a number to club off — wrong group (every mishit dragging it down) and wrong measure (carry on a driver, when what you care about is where the ball finishes). The ruling deleted 205 from display rather than relabel it, on the grounds that a member looking at two driver numbers on one page will club off the wrong one.
"Best twenty" does not mean "longest twenty"
This is the part that keeps the number honest and it's easy to miss.
If you took your twenty longest shots, you'd be taking your twenty biggest flukes — the flyer, the one the sim's weather helped, the one you got everything out of and nothing since. Rank by distance and you overstate your gapping by half a club or more, which on the course means you're between clubs on every approach and you can't work out why.
So the ranking isn't distance. It's contact. The twenty selected are the twenty you struck best, and their distance is then read off. Best-struck, not best-outcome — because what a fitter means by "your number" is what you can repeat, and clean contact repeats. Distance doesn't.
Contact quality comes from smash factor, which needs the club markers to compute. When those aren't on a club, the ranking falls back to ball speed, and this is where it gets careful. Ranking on raw top ball speed would quietly reintroduce the exact bias we just removed, because the fastest ball speeds come from the hardest swings, not the best strikes. A heave keeps the speed and loses the carry.
So the fallback sets aside the top of your own ball-speed range with that club — roughly the fastest tenth — and takes its twenty from the band below. The app even says so on the page in those words: ranked by solid-strike ball speed, hardest swings set aside. When that change was tested on a pitching wedge, the trimmed band's carry came out at 91 yards against 85 for the raw-top-speed picks. Setting aside the hardest swings produced the longer number, because it produced the better-struck one.
And we tell you it runs hot
A best-twenty median is optimistic by construction. It's the middle of your good ones.
We don't hide that. There's a sentence required on every surface that prints one of these numbers, and it reads:
> Flush numbers — your best 20 strikes with each club. On the course, expect a touch less.
There's a second one on the gapping page, and it's subtler:
> Gaps compare flush numbers. Your longest clubs scatter most, so gaps at the top of the bag read a touch wide.
That one is worth a second read. The wider a club's shots scatter, the further its best-twenty median floats above its true middle — and a driver scatters far more than a wedge. So the gaps between your long clubs read a little generous compared to the gaps between your short ones. One consistent rule for every club makes the same club's number match everywhere in the app; it does not make different clubs perfectly comparable to each other. We'd rather print that sentence than let you find it out on a par five.
Below twenty you get a fraction, not a number
If a club hasn't produced twenty qualifying shots yet, the app doesn't show a distance for it. It shows X/20 and asks you to keep hitting.
Under ten measured shots for a given number, it goes further and shows nothing at all — the honest words are not enough measured shots for this metric. Between ten and twenty, if a number does appear, it must carry its own count: median of 12 measured. Never a bare figure.
That looks stingy. It's the opposite. A number built from six shots is a coin flip wearing a decimal point, and the moment we print it you will club off it. The honest form of not knowing is a count, not a padded guess.
There's a second reading on that fraction, and it's worth hearing. A club that chronically can't produce twenty clean strikes — you hit it constantly and the count barely moves — is telling you something. That may not be your swing. That may be the club.
Where medians lie too
A median isn't magic. It fails badly in one specific place, and this system has a ship-blocking rule about it because the failure is so convincing.
The failure is signed numbers — anything that can be negative one way and positive the other. Curve. Spin axis. Which side of the line you finished on.
Picture a golfer who hits half his shots with 15 degrees of slice and half with 15 degrees of hook. Take the middle of those signed values and you get approximately zero. Zero reads as dead straight. He is the least straight golfer on the property.
This isn't hypothetical, and it isn't a near miss. On that same reference card, eight spin-axis readings running from −35.1° to +22.7° summarise to 3.3°: a man spraying it both ways, printed as a man who hits it dead straight. And our own swing scorer once took the absolute value of a signed median and scored a two-way miss as perfect — the worst pattern in golf, rendered as a clean sheet. That's a real bug that really shipped, and it's why the scorer now refuses: when your miss is mixed, it declines to grade that dimension at all rather than grade it wrong.
The same warning sits in capital letters above the number in our own database: it is a diagnostic statistic, not a grade. It may not grade a shot, colour a dot, or word a shape. Anything you see about your shot shape is graded one shot at a time and reported as a mix — you curve it this way this often, that way that often — precisely so a two-way miss can never hide inside a zero.
Rule of thumb worth carrying off this page: never average a direction. Distances average fine. Sides don't.
Best and typical answer different questions
Which brings it back to the 300-yard drive.
Your best number and your typical number are both real, both yours, both worth having. They answer different questions. The best answers what am I capable of on the swing of my life. The typical answers what club do I pull. Only one of those belongs in your hands on a 165-yard approach, and it isn't the exciting one.
The app puts them on one ruler on purpose, in a sentence shaped like this:
> Best 278 · typical driver total 249 — you live 29 yards from your best.
Both numbers are yours, and the arithmetic under it is stricter than it looks. Your best is the maximum of the same group your typical is the middle of, so the gap can never come out negative — but it can come out zero, if you've only one tracked drive, and "you live 0 yards from your best" is a sentence nobody should have to read, so the app drops it and lets the two marks say it without the subtraction.
It also refuses to draw the comparison at all when the two aren't the same kind of distance. If your best is a total and your typical is a carry, those are two different rulers and subtracting one from the other produces a figure that means nothing. The app keeps the honest half of the picture instead of a matched pair that lies.
Three numbers, three questions
That's the toolkit.
Your best — what happened once. Real, yours, frame it. Don't club off it.
Your typical — the middle of your best-struck twenty, and a touch hot. The one to club off.
Your spread — how wide those shots landed. The one that says whether to trust the other two.
The 300-yard drive isn't a lie. It happened. It just isn't the answer to the question you were about to ask it.
From the glossary
Related reading: Reading a session like a coach · What counts as a record here · Reading Your Shot Data Like a Fitter · Your best with every club is not your yardage
Reading a session like a coach
Forty numbers came back from your session. You read all forty. You changed nothing.