Is "17 inches past" really the best pace?
- Putting
- Physics
- Distance control
"Hit every putt hard enough to finish 17 inches (about 43 cm) past the hole." Known from Dave Pelz's research, this is probably the single most quoted rule in putting.
Is it true? I tested it with the same physics model that drives the Putt Break Calculator.
How it was tested
"Past distance" means how far beyond the hole the ball would stop if the hole were not there. Sweeping the pace, I computed the make probability including stroke dispersion (7% in distance, 0.8° in direction), and found the past distance that maximises it.
Two things are compared:
- Optimum: the pace with the highest make probability for that putt
- Fixed 17 inches: the pace that finishes exactly 43 cm past
If the gap is small, 17 inches is good enough in practice.
On a flat green
Stimp 10 feet, no slope.
| Distance | Optimal past | Speed at the hole | Optimum | Fixed 17 in | Gap |
|---|---|---|---|---|---|
| 1 m | 25 cm (9.8 in) | 0.60 m/s | 100.0% | 99.9% | 0.0 pt |
| 2 m | 31 cm (12.2 in) | 0.65 m/s | 92.4% | 91.6% | 0.8 pt |
| 3 m | 41 cm (16.1 in) | 0.73 m/s | 74.4% | 74.4% | 0.1 pt |
| 4 m | 50 cm (19.7 in) | 0.80 m/s | 58.6% | 58.4% | 0.2 pt |
| 5 m | 59 cm (23.2 in) | 0.86 m/s | 46.9% | 45.9% | 1.0 pt |
| 6 m | 63 cm (24.8 in) | 0.88 m/s | 38.6% | 37.6% | 0.9 pt |
| 8 m | 69 cm (27.2 in) | 0.92 m/s | 28.8% | 27.2% | 1.6 pt |
| 10 m | 76 cm (29.9 in) | 0.96 m/s | 20.9% | 19.7% | 1.3 pt |
The optimum is not a constant. It runs from 10 inches at 1 m to 30 inches at 10 m. Seventeen inches happens to be right at around 3 m.
And yet fixing it at 17 inches costs at most 1.6 percentage points. On an 8 m putt, 28.8% becomes 27.2% — one or two putts in a hundred. Nothing you would ever feel on the course.
As a single number covering every length, it is remarkably good.
What slope changes
A 4 m putt on stimp 10, varying only the up/down slope.
| Slope | Optimal past | Speed at the hole | Optimum | Fixed 17 in | Gap |
|---|---|---|---|---|---|
| 3° uphill | 35 cm (13.8 in) | 0.94 m/s | 45.1% | 44.4% | 0.7 pt |
| 2° uphill | 40 cm (15.7 in) | 0.91 m/s | 48.7% | 48.6% | 0.1 pt |
| 1° uphill | 42 cm (16.5 in) | 0.84 m/s | 53.9% | 53.9% | 0.0 pt |
| Flat | 50 cm (19.7 in) | 0.80 m/s | 58.6% | 58.4% | 0.2 pt |
| 1° downhill | 51 cm (20.1 in) | 0.68 m/s | 73.7% | 73.2% | 0.5 pt |
| 2° downhill | 63 cm (24.8 in) | 0.58 m/s | 88.2% | 86.0% | 2.3 pt |
| 3° downhill | 162 cm (63.8 in) | 0.52 m/s | 100.0% | 89.2% | 10.7 pt |
Uphill, 17 inches is close to perfect. The optimum itself moves to 14–17 inches and fixing it costs under 0.7 of a point.
Downhill is another matter. The optimum grows sharply with slope, and at 3° down the best pace finishes 63 inches (1.6 m) past. Holding to 17 inches drops the make probability by 10.7 points.
Why the rule breaks downhill
Look at the "speed at the hole" column.
At the optimal pace, the ball reaches the hole at 0.5–1.0 m/s in every case. Whatever the slope, putts that go in arrive at roughly the same speed. That is the real physics: capture depends on speed, not on past distance.
On a flat green the two track each other neatly, which is why "17 inches past" works as a restatement of it.
Downhill they come apart. A worked example:
3° downhill, 4 m putt
Speed at the hole 0.87 m/s ← slow enough to drop
Past distance 639 cm ← would run 6 m past if the hole were covered
It arrives slowly and then keeps being pushed along by the slope. Past distance stops representing arrival speed at all, so the 43 cm restatement no longer means anything.
Downhill putts have another peculiarity. At 3° down the acceptable aim window was ±3.1°, against ±0.6° on the flat. Gravity keeps adding speed along the line while friction bleeds away the sideways component, so aiming errors partly correct themselves. That is why the downhill probabilities come out so high.
Slope has a hard limit
Past 4° the calculation broke down. That is not a bug in the model.
On a stimp 10 green the rolling friction coefficient is μ = 0.0552, so the steepest slope on which a ball can stay at rest is
arctan(0.0552) = 3.16°
Beyond 3.16° a ball can never come to rest. Real greens have the same limit — you cannot cut a hole on a slope steeper than that, or the ball starts moving in the wind. The 4° uphill result, where the ball fails to reach the hole and then rolls back down forever, is correct physics.
Does skill level matter?
Varying the dispersion on a 4 m flat putt:
| Skill (distance / direction dispersion) | Optimal past | Optimum | Fixed 17 in | Gap |
|---|---|---|---|---|
| Strong (4% / 0.8°) | 30 cm (11.8 in) | 62.7% | 61.5% | 1.3 pt |
| Average (8% / 1.5°) | 53 cm (20.9 in) | 33.2% | 33.0% | 0.2 pt |
| Higher handicap (14% / 2.5°) | 69 cm (27.2 in) | 17.3% | 16.5% | 0.8 pt |
More dispersion pushes the optimum further past, but fixing 17 inches costs under 1.3 points at every level. It is not a rule only good players can use.
One important caveat
Everything above measures only whether the putt goes in. It says nothing about what is left when you miss — the three-putt risk is not in these numbers.
The "optimal" 1.6 m past on that 3° downhill leaves you a 1.6 m putt coming back, sometimes downhill again. Maximising the chance of holing out can easily cost you strokes.
Seen this way, holding to 17 inches downhill is the sensible play. You give up 10 points of make probability and get back a 43 cm tap-in when you miss. Pelz's rule has lasted because it is decent at both — holing out and avoiding three-putts.
Summary
- Flat and uphill: 17 inches is near-optimal. The cost is around a point; you would never notice
- The true optimum does vary with length (10 in at 1 m, 30 in at 10 m) — but it does not matter
- Downhill it stops working, because past distance and arrival speed come apart
- What actually matters is arrival speed of 0.5–1.0 m/s. Past distance is just a proxy for it
- Chasing make probability alone adds three-putts. Easing off downhill is the right instinct
To run the same numbers for your own green speed, slope and distance, open the Putt Break Calculator and look at the recommended pace and aim. The "map of what goes in" shows which combinations of pace and line actually drop.
Conditions: this site's physics model — rolling friction calibrated from the Stimpmeter reading, hop, skid and true roll, with capture judged at the lip. Probabilities integrate stroke dispersion, assumed log-normal in distance and normal in direction. The model and how it is validated are described on the about page.