Strokes Gained and Shot Tracking for Amateurs

Most club golfers who start tracking shots discover the same thing within about six rounds: the part of their game they complain about in the bar is not the part costing them shots. The 16-handicapper who talks endlessly about three-putts is usually losing five shots a round between 130 and 190 yards. The 8-handicapper convinced his driver is the problem is often losing more from 40 yards than from the tee. Strokes gained exists to settle that argument with arithmetic rather than memory, and the tooling to do it on a UK members’ course now costs less than a fitted driver.

This page is about making that arithmetic trustworthy. Not the theory, which takes ten minutes to understand, but the practical business of getting clean data out of a phone camera, a £400 launch monitor and a shot-tracking app that has to guess whether your ball is in the fairway at Woodhall Spa or the heather next to it.

What strokes gained actually measures, and why the baseline decides everything

Strokes gained is one subtraction. Take the expected number of strokes to hole out from where your ball starts, subtract the expected strokes from where it finishes, subtract one for the shot you just played. Positive means you gained ground on the benchmark. Negative means you lost it.

The benchmark is the whole argument. Mark Broadie’s PGA Tour tables, the ones that made the metric famous in Every Shot Counts, say a tour player needs 2.98 strokes from 160 yards in the fairway and 1.61 from ten feet on the green. Measured against that, a 15-handicap loses about 0.9 strokes a hole and learns nothing. Strokes gained for amateurs only becomes useful when the expected-strokes table is built from players of roughly your standard, which is why the handicap-banded baselines inside Shot Scope, Arccos and Clippd matter more than the headline feature list.

A workable 15-handicap baseline looks roughly like this:

Notice the shape of it. A 15-handicapper gains almost nothing by moving from 195 to 150 yards in the fairway, about a third of a stroke, and that is the fact that quietly destroys the “just advance it up there” approach to par 5s. The big drops sit inside 100 yards and inside 15 feet.

A worked hole, shot by shot

Here is a real-shaped hole for that player: 410 yards, par 4, slight dogleg right, rough that is genuinely punishing in September.

Tee shot. Drive of 218 yards into the right rough, 192 left. Start 5.05, finish 4.55, minus one. Strokes gained: −0.50.

Second. Five-iron advanced to 65 yards in the fairway. Start 4.55, finish 3.25, minus one. Strokes gained: +0.30.

Third. Wedge to 22 feet. Start 3.25, finish 2.05, minus one. Strokes gained: +0.20.

Fourth. First putt to three feet. Start 2.05, finish 1.10, minus one. Strokes gained: −0.05.

Fifth. Holed. Start 1.10, minus one. Strokes gained: +0.10.

Five shots, total strokes gained +0.05, against a baseline expectation of 5.05. A bogey that was, in fact, a par-standard hole for that player. And the only shot that cost real ground was the tee shot, which most golfers would have walked away from thinking “fine, it’s in play”. That is the value of the subtraction: it separates the shot that felt bad from the shot that was bad.

The misdiagnosis that costs club golfers a season

Take an actual round export. This is the shape of what a tracking app gives you after 18 holes:

ROUND  14 SEP 2026   Notts GC (Hollinwell)   89 gross
PLAYER 14.2 index, course handicap 16

                        SG vs 14-handicap baseline
Off the tee                       -2.31
Approach 100-200 yds              -3.08
Approach inside 100 yds           -0.44
Around the green                  +0.62
Putting                           +0.11
                                  ------
Total                             -5.10

APPROACH DETAIL
  100-124 fairway   2 shots   GIR 2/2   avg proximity 44 ft   SG -0.34
  125-149 fairway   3 shots   GIR 1/3   avg proximity 61 ft   SG -0.91
  150-174 rough     4 shots   GIR 0/4   avg proximity 93 ft   SG -1.52
  175-199 any lie   2 shots   GIR 0/2   avg proximity 78 ft   SG -0.31

PUTTING DETAIL
  First-putt avg distance 41 ft     Putts 34     3-putts 3

Three three-putts and 34 putts. Every instinct says the putting was the problem. The maths says putting was fractionally positive, because from an average first-putt distance of 41 feet, a 14-handicapper is expected to take a shade over two putts, and three-putting three times out of eighteen from that distance is normal. The putts were a symptom. The cause is 150 to 174 yards from the rough: four shots, no greens, an average of 93 feet from the hole.

This pattern repeats across amateur datasets with irritating consistency. Broadie’s amateur samples put roughly two-thirds of the scoring difference between handicap levels in the long game, with approach play the single largest slice. Shot Scope’s published amateur numbers say much the same in a different language: greens in regulation runs around 60% for scratch players, near 30% for a 10-handicap and under 15% for a 20, while putts per round differ by only two or three across that whole span.

Which tracking systems read a UK round correctly

Shot detection is the part that breaks. A strokes gained report is only as good as the app’s belief about where each ball sat, and that belief comes from GPS, a club tag and some inference. The differences between systems are practical rather than philosophical, and I have covered the full comparison, pricing and detection-accuracy notes on shot tracking apps for UK club golfers. The short version for anyone deciding this week:

Shot Scope (V5/V6, X5 watch) uses passive tags screwed into the butt of each grip, detected when the watch passes near the club. No subscription, which over four years is the difference between roughly £180 total and £600. The company is based in Edinburgh and its course mapping across UK and Irish courses is the best of the group, including the small parkland clubs that American-first apps map badly.

Arccos (Link Pro plus 14 sensors) detects shots acoustically and by GPS, no watch required, and its Caddie feature gives you a club recommendation with wind and elevation. Detection is genuinely good on full shots and genuinely unreliable on soft chips and short putts, so budget five minutes of round editing afterwards. It needs an annual membership, currently in the region of £100 to £125.

Garmin (Approach S70 or S44 with CT10 tags) wins if you already wear a Garmin. Strokes gained lives in the Garmin Golf app, the tags are automatic, and the watch is a watch rather than a golf device.

Clippd takes a different route: it ingests shot data and practice data and reduces each skill to a 0 to 100 Player Quality score, which is easier to act on than a decimal but hides the underlying expected-strokes model. Worth it if you are coached and your coach reads the same dashboard.

Whatever you choose, two habits matter more than the brand. Register the pin position or accept that putting data is noise, and edit the round the same evening while you can still remember that the eleventh was a provisional, not a lost ball.

What AI reads off your phone footage, and what it makes up

Now the swing side, where the marketing has run some distance ahead of the physics.

Single-camera 3D systems are the real advance. Sportsbox AI 3DGolf takes one iPhone video and reconstructs a 3D skeleton, giving you pelvis and thorax rotation in degrees, sway, thrust, lift and hand path frame by frame. Validation work comparing it against marker-based motion capture broadly supports the large-scale body measures: pelvis and thorax rotation land within a few degrees, sway and lift within a centimetre or two at address-to-top positions. It costs somewhere around £15 to £20 a month depending on tier.

What the same class of system reads badly is anything small, fast or thin. Lead wrist flexion and extension, clubface orientation, shaft lean at impact: these are derived from a handful of pixels moving faster than the sensor can resolve, and the error bars are wide enough that a 10-degree change in a report may be nothing at all. If wrist angles are your actual question, a direct sensor answers it properly. HackMotion straps to the lead wrist and reads flexion and extension continuously, and the numbers are blunt: tour irons tend to arrive at impact roughly flat to five degrees flexed, while mid-handicappers commonly show 10 to 25 degrees of extension. That is a measurement, not an inference from video.

The generic “AI swing analysis” bolted onto general golf apps mostly delivers pose-estimation overlays and a plain-language critique generated from them. Useful as a mirror. Not useful as a measurement, and specifically not useful for anything you would change a grip or a shaft over.

The frame-rate maths nobody puts on the box

This is worth doing once on paper, because it explains most of the disagreements between a video app and reality.

A clubhead at 100 mph is travelling 146.7 feet per second.

At 30 frames per second, the club covers nearly five feet between frames. There is no impact frame. There is a frame before impact and a frame after, and anything the software tells you about impact position has been interpolated. At 240 fps you get a genuine sample every seven inches, which is enough to see shaft lean and roughly enough to see face orientation if the light is good.

Shutter speed is the second half of it. Slow-motion capture on an iPhone only holds a short exposure when there is enough light; film in a dim indoor bay in February and the clubhead smears into a grey arc regardless of frame rate. Practical setup: outdoors or under bright lights, 240 fps, phone on a small tripod at hip height, about eight to ten feet away, lens locked at 1x, full body and club in frame at both extremes of the swing, exposure locked with a long press before you start. A £15 tripod improves your data more than a £200 app subscription does, because inconsistent camera height is the largest source of frame-to-frame variation in every single-camera system.

Getting honest numbers from a budget launch monitor

Three units dominate the £400 to £700 bracket in the UK, and each has a different relationship with the truth.

Garmin Approach R10 measures club speed and ball speed by Doppler radar and models much of the rest. Trust the speeds, treat spin as an estimate, and be aware of its well-known tendency to flatter iron carry distances. Around £400 to £500.

Rapsodo MLM2PRO actually measures spin when you use the Callaway RPT balls with their printed markers, and reports modelled spin when you do not. The two datasets are not interchangeable, so pick one ball type and stay with it. Full feature access is subscription-gated.

FlightScope Mevo needs the metallic dots stuck to the ball for spin. No dots, no spin, and the carry number inherits that.

The usable habit with all of them is to stop reading individual shots. Hit eight balls with one club, discard the longest and the shortest, and record the median carry. Then do it again next week and compare medians. Single-shot numbers off a budget monitor carry enough error that a 6-yard difference means nothing, while a 6-yard shift in the median across two sessions is probably real.

Gapping: the cheapest strokes on your list

Back to the player losing 1.5 strokes from 150 to 174 in the rough. Here is what eight-ball median testing showed:

Look at the hole between the 5-iron at 172 and the hybrid at 196. That is 24 yards with nothing in it, and it sits precisely where his approach losses live. From 185 yards he is choosing between a 5-iron he has to hit perfectly and a hybrid he has to hold back, and in the rough neither works. A 7-wood or a 3-hybrid at about 184 carry fills it for the price of one lesson. Meanwhile the 8-iron through 9-iron through pitching wedge are separated by nine and eight yards, which is inside his own shot-to-shot spread and therefore not a real distinction.

Half of amateur gapping problems are this: three clubs covering 25 yards at the bottom of the bag, and a hole you could park a bus in at the top.

Stop counting putts

Putts per round is a broken statistic because it rewards missing greens. Hit eighteen greens and you will putt 34 times; miss twelve, chip to five feet repeatedly and you will putt 30 and shoot worse. The number that actually predicts scoring for amateurs is first-putt distance, and the lever on it is approach proximity.

Work the arithmetic. If you cut average proximity from 150 yards in the fairway from 61 feet to 44 feet, you save roughly 0.15 strokes on each of those shots. Four or five of them a round makes about 0.7 strokes, every round, without touching your putting stroke. Compare that to becoming a genuinely better putter from 8 feet, which for most players is worth a fraction of it and takes considerably longer to achieve.

Where putting practice does pay is lag. From 40 feet, the target is not holing out, it is finishing inside four feet. A 15-handicapper who leaves the first putt inside four feet from long range eliminates most of their three-putts, and the drill is dull: two balls, five distances from 30 to 60 feet, score one point for inside four feet, and repeat until you hit 8 out of 10.

UK corrections: winter, WHS and the course map

Cold air and soft turf will quietly corrupt a whole winter of tracking. The commonly used figure is about two yards lost per 10°F of temperature drop on a mid-iron, so a 25°C August afternoon against an 8°C January morning costs roughly six yards on a 7-iron before you account for wet fairways killing rollout and two extra layers restricting your turn. If your app has a temperature or elevation adjustment, switch it on. If it does not, keep summer and winter gapping tables separately rather than pretending one number holds year-round.

Handicap inputs matter too. Under the World Handicap System your index is the average of the best 8 of your last 20 scores, so it lags your current form by weeks and moves in steps. When you set a handicap band in your tracking app, set it to your index rather than the score you shot last Saturday, and resist the temptation to nudge it down after one good round: an optimistic baseline makes every strokes gained figure look worse and hides which part is actually failing.

Course mapping is the third UK-specific trap. Green outlines and fairway boundaries on crowdsourced maps are frequently wrong at smaller clubs, and a fairway edge drawn 8 yards off will misclassify your lie and therefore your expected strokes. Play your first three tracked rounds at your home course, check the app’s lie calls against what you remember, and report the bad holes. Both Shot Scope and Arccos fix submitted map errors, usually within a fortnight.

Four weeks that move the number

Given the report above, here is what the player should actually do. Not everything at once, which is how tracking data usually gets abandoned.

Week one. Full eight-ball gapping session, median carries only, and order whatever club fills the 172 to 196 hole. Nothing else.

Week two. Proximity ladder from 150 yards. Ten balls, and instead of counting greens, measure dispersion width: hit all ten, pace out how wide the pattern is left to right. Most 14-handicaps discover their 150-yard pattern is 30 yards wide, which is the real reason to aim at the middle of the green rather than the flag.

Week three. Tee shots, scored on a single criterion: in play or not. Ten drives, and if fewer than six finish where you could hit a normal second shot, drop to the 3-wood for a month. The 13-yard loss costs about 0.1 strokes per hole against a baseline; a penalty costs a full stroke plus position.

Week four. Back on the course, tracker on, and compare the approach block against the 14 September round. One number, one comparison, no browsing the other twelve dashboards.

Come back to the full report every fifth round rather than every round. Single-round strokes gained has enough noise in it that a two-stroke swing in any category means very little, and five rounds is roughly where the approach numbers stabilise enough to act on. The players who improve fastest with this stuff are the ones who look at it least often and change the fewest things between looks.