Lab reports · 2026-11-09
The break setups most likely to deliver each goal, for a breaker with a top speed of 16 or 19 mph: where to put the cue ball, where to aim, what tip and how hard. Then six pro break styles run the same way, which ball to hit for the 8, what speed buys, the break box, and how far the model can be trusted.
Simulated in the lab's event engine (pooltool physics, ball flight on) with realistic rack gaps (±0.03 in) and stroke error (±0.3° aim, ±2% speed); finals are 4,000 breaks each on fresh racks. Tables are drawn head rail on the left, so your left rail runs along the top. The rankings travel better than the exact percentages.
From the left rail, or the centre where that's better: the setup most likely to deliver each goal, then the whole search behind it. Each top speed was searched separately. On a 7-footer the 1 needs pace: the same setup makes it about 23% of the time at 19 mph and about 6% at 16.
1 ball on the break
Aiming for the 9 with this break doesn't really work: every finalist made it 3.0%–6.7% of the time.
Every setup searched for this goal
Each cell is the best tip and speed (up to 19 mph) at that spot and aim, from 80 breaks: about ±4 points of noise, so read the shape, not single cells. The outlined cell is the recipe's neighbourhood; grey ↑ cells need the cue raised over 12° to clear the rail, which the model can't play, so they aren't scored. Load it in the simulator →
1 ball on the break
Aiming for the 9 with this break doesn't really work: every finalist made it 2.4%–3.9% of the time.
Every setup searched for this goal
Each cell is the best tip and speed (up to 19 mph) at that spot and aim, from 80 breaks: about ±4 points of noise, so read the shape, not single cells. The outlined cell is the recipe's neighbourhood; grey ↑ cells need the cue raised over 12° to clear the rail, which the model can't play, so they aren't scored. Load it in the simulator →
most balls on the break
Aiming for the 8 with this break doesn't really work: every finalist made it 1.2%–2.1% of the time.
Every setup searched for this goal
Each cell is the best tip and speed (up to 19 mph) at that spot and aim, from 80 breaks: about ±4 points of noise, so read the shape, not single cells. The outlined cell is the recipe's neighbourhood; grey ↑ cells need the cue raised over 12° to clear the rail, which the model can't play, so they aren't scored. Load it in the simulator →
most balls on the break
Aiming for the 8 with this break doesn't really work: every finalist made it 0.8%–1.5% of the time.
Every setup searched for this goal
Each cell is the best tip and speed (up to 19 mph) at that spot and aim, from 80 breaks: about ±4 points of noise, so read the shape, not single cells. The outlined cell is the recipe's neighbourhood; grey ↑ cells need the cue raised over 12° to clear the rail, which the model can't play, so they aren't scored. Load it in the simulator →
the 8 on the break: the second-ball break
The centre head-ball break on the same racks: 1.2% (0.9%–1.6%). Same direction as Dr. Dave's 169 pro games (second-ball 5.9% against head-ball 0.8%), a smaller gap.
Every second-ball setup searched
Each cell is the best tip and english at that spot and cut, from 1,000 breaks on a loose rack (±0.03 in); the headline is 4,000 fresh breaks.
the 8 on the break: the second-ball break
The centre head-ball break on the same racks: 1.0% (0.8%–1.4%). Same direction as Dr. Dave's 169 pro games (second-ball 5.9% against head-ball 0.8%), a smaller gap.
Every second-ball setup searched
Each cell is the best tip and english at that spot and cut, from 1,000 breaks on a loose rack (±0.03 in); the headline is 4,000 fresh breaks.
1 ball on the break
Aiming for the 9 with this break doesn't really work: every finalist made it 4.0%–5.1% of the time.
Every setup searched for this goal
Each cell is the best tip and speed (up to 16 mph) at that spot and aim, from 80 breaks: about ±4 points of noise, so read the shape, not single cells. The outlined cell is the recipe's neighbourhood; grey ↑ cells need the cue raised over 12° to clear the rail, which the model can't play, so they aren't scored. Load it in the simulator →
1 ball on the break
Aiming for the 9 with this break doesn't really work: every finalist made it 1.1%–3.7% of the time.
Every setup searched for this goal
Each cell is the best tip and speed (up to 16 mph) at that spot and aim, from 80 breaks: about ±4 points of noise, so read the shape, not single cells. The outlined cell is the recipe's neighbourhood; grey ↑ cells need the cue raised over 12° to clear the rail, which the model can't play, so they aren't scored. Load it in the simulator →
most balls on the break
Aiming for the 8 with this break doesn't really work: every finalist made it 1.1%–2.0% of the time.
Every setup searched for this goal
Each cell is the best tip and speed (up to 16 mph) at that spot and aim, from 80 breaks: about ±4 points of noise, so read the shape, not single cells. The outlined cell is the recipe's neighbourhood; grey ↑ cells need the cue raised over 12° to clear the rail, which the model can't play, so they aren't scored. Load it in the simulator →
most balls on the break
Aiming for the 8 with this break doesn't really work: every finalist made it 0.7%–0.9% of the time.
Every setup searched for this goal
Each cell is the best tip and speed (up to 16 mph) at that spot and aim, from 80 breaks: about ±4 points of noise, so read the shape, not single cells. The outlined cell is the recipe's neighbourhood; grey ↑ cells need the cue raised over 12° to clear the rail, which the model can't play, so they aren't scored. Load it in the simulator →
the 8 on the break: the second-ball break
The centre head-ball break on the same racks: 1.1% (0.8%–1.5%). Same direction as Dr. Dave's 169 pro games (second-ball 5.9% against head-ball 0.8%), a smaller gap.
Every second-ball setup searched
Each cell is the best tip and english at that spot and cut, from 1,000 breaks on a loose rack (±0.03 in); the headline is 4,000 fresh breaks.
the 8 on the break: the second-ball break
The centre head-ball break on the same racks: 0.4% (0.3%–0.7%). Same direction as Dr. Dave's 169 pro games (second-ball 5.9% against head-ball 0.8%), a smaller gap.
Every second-ball setup searched
Each cell is the best tip and english at that spot and cut, from 1,000 breaks on a loose rack (±0.03 in); the headline is 4,000 fresh breaks.
The break patterns documented in The Modern Break, run as drawn on both table sizes at 16, 19 and a pro-like 24 mph (Deuel's at 16 and 19), with a tight Magic-Rack rack, 2,000 breaks each. On each diagram the dashed circle is the ghost ball at contact, the ringed ball is the one hit first, and the white arrow is where it's sent; the close-up shows the rack. Each card says how far the model can be trusted for it.
9-ball side break
Good for comparing setups · rates approximateCue ball on or near the head string close to the side rail; a full, square hit on the 1. The near wing ball heads for the breaker-side foot corner, the far wing banks cross-side, and the 1's route to the far side depends on speed.
Source: Dr. Dave, 9-ball break FAQ
Drawn: 9-ft table, cue ball 4.9 in of cloth off the left cushion, 10% below centre (draw); full on the 1.
| Table | Speed | Made a ball | Balls | The 1 | The 9 | Scratch |
|---|---|---|---|---|---|---|
| 7 ft | 16 mph | 57.9% | 0.83 | 5.6% (side 3.9%) | 3.7% | 3.5% |
| 7 ft | 19 mph | 65.0% | 1.00 | 11.3% (side 8.7%) | 4.5% | 6.4% |
| 7 ft | 24 mph | 69.0% | 1.15 | 15.8% (side 11.1%) | 6.5% | 10.3% |
| 9 ft | 16 mph | 40.3% | 0.49 | 2.1% (side 1.4%) | 2.3% | 1.4% |
| 9 ft | 19 mph | 52.1% | 0.71 | 7.0% (side 5.6%) | 2.4% | 2.4% |
| 9 ft | 24 mph | 56.8% | 0.84 | 12.3% (side 9.2%) | 3.0% | 7.0% |
Measured: 2023 US Open (Diamond 9-footers, 4-inch corners): 73% successful breaks, about 1.0 ball per break, 9 on the break 1.4%.
9-ball box-edge cut
Model caveat · it undersells the 1 in the side (about 10% against Ko's 23 of 24)With a marked break box, set up near the box's edge and cut the 1 so it works the side pocket. Measured from video, Ko sets up 1.2 in inside the box's left edge and hits the 1 about 0.6 full at about 16.6 mph; the 1 rolls into the cross side.
Source: WNT event rules · 2023 US Open break tracker
Drawn: 9-ft table, cue ball 18.9 in of cloth off the left cushion (inside the break box, dashed), 15% below centre (draw); about ¾-ball on the 1.
| Table | Speed | Made a ball | Balls | The 1 | The 9 | Scratch |
|---|---|---|---|---|---|---|
| 7 ft | 16 mph | 58.0% | 0.80 | 6.8% (side 5.3%) | 3.6% | 8.8% |
| 7 ft | 19 mph | 64.5% | 0.98 | 11.7% (side 8.6%) | 4.8% | 11.9% |
| 7 ft | 24 mph | 71.3% | 1.17 | 16.1% (side 12.6%) | 6.2% | 15.5% |
| 9 ft | 16 mph | 39.6% | 0.46 | 2.6% (side 2.0%) | 1.8% | 6.0% |
| 9 ft | 19 mph | 48.3% | 0.62 | 6.0% (side 4.9%) | 3.0% | 7.4% |
| 9 ft | 24 mph | 55.1% | 0.80 | 10.7% (side 8.3%) | 3.6% | 10.7% |
Measured: Ko Ping Chung, 2023 US Open: 1 in the side on 23 of 24 tracked breaks. His break measured from video →
10-ball modern default
Good for comparing setups · rates approximateCue ball slightly off centre, a square-ish hit on the 1: second-row balls head for the side pockets, the 1 can reach the top corner, the 10 runs back toward the opposite corner. The 10-ball rack barely supports a wing-ball corner drop, so the goal is a spread with a shot on the 1.
Source: Dr. Dave, 10-ball break FAQ
Drawn: 9-ft table, cue ball 17.6 in of cloth off the left cushion, 10% below centre (draw); full on the 1.
| Table | Speed | Made a ball | Balls | The 1 | The 10 | Scratch |
|---|---|---|---|---|---|---|
| 7 ft | 16 mph | 51.9% | 0.69 | 2.6% (side 0.9%) | 2.9% | 9.6% |
| 7 ft | 19 mph | 56.9% | 0.84 | 4.2% (side 1.6%) | 3.7% | 12.5% |
| 7 ft | 24 mph | 67.0% | 1.07 | 9.1% (side 5.9%) | 4.9% | 12.9% |
| 9 ft | 16 mph | 36.4% | 0.43 | 1.5% (side 0.2%) | 1.5% | 4.4% |
| 9 ft | 19 mph | 44.9% | 0.57 | 2.6% (side 0.5%) | 3.0% | 9.3% |
| 9 ft | 24 mph | 51.1% | 0.73 | 3.6% (side 1.7%) | 4.3% | 11.5% |
8-ball lead-ball power
✓ Matches pro data within a few pointsCue ball slightly off the centre line, about half a diamond past centre from the breaker's view, hitting the 1 as square as possible: second-row balls fly to the side pockets, corner balls travel about four rails to the corners.
Source: Dr. Dave, 8-ball break FAQ · BD Aug 2016
Drawn: 7-ft table, cue ball 13.5 in of cloth off the left cushion, 10% above centre (follow); full on the 1.
| Table | Speed | Made a ball | Balls | The 8 | Scratch |
|---|---|---|---|---|---|
| 7 ft | 16 mph | 57.3% | 0.83 | 0.7% | 2.3% |
| 7 ft | 19 mph | 65.6% | 1.05 | 1.4% | 3.0% |
| 7 ft | 24 mph | 71.6% | 1.27 | 2.4% | 5.5% |
| 9 ft | 16 mph | 38.5% | 0.49 | 0.1% | 2.4% |
| 9 ft | 19 mph | 50.6% | 0.68 | 0.4% | 2.5% |
| 9 ft | 24 mph | 58.4% | 0.87 | 1.1% | 4.0% |
Measured: Top men's pros on bar tables (169 games): made a ball 80.5%, scratched 7.6%, 8 on the break 0.8%.
8-ball Deuel second-ball
Model caveat · its scratch rate runs high (about 8% against pros' 2%)Deuel “breaks from the corner, hitting the 2nd row ball squarely, maybe 15–17 mph”, with a stun stroke and no side. The 1 sits in front of the second-row ball, so from anywhere in the kitchen the fullest hit that still misses the 1 is about 0.7 of the ball. Drawn on the head string 6 in off the rail, bridging off the rail with the cue level. (Tucked into the head-rail corner, the cue must rise about 9° over the head rail and the cue ball hops: in the model that scratched half the time at 19 mph.)
Source: Dr. Dave, 8-ball break FAQ
Drawn: 7-ft table, cue ball 3.6 in of cloth off the left cushion, centre ball (stun); about ¾-ball on the 9.
| Table | Speed | Made a ball | Balls | The 8 | Scratch |
|---|---|---|---|---|---|
| 7 ft | 16 mph | 64.8% | 0.97 | 1.4% | 8.8% |
| 7 ft | 19 mph | 71.1% | 1.16 | 2.5% | 8.3% |
| 9 ft | 16 mph | 46.1% | 0.59 | 0.7% | 10.3% |
| 9 ft | 19 mph | 54.1% | 0.75 | 1.3% | 7.0% |
Measured: Top pros' second-ball breaks (bar tables): made a ball 82.4%, scratched 2.0%, 8 on the break 5.9%.
8-ball McAuliffe thin cut
Model caveat · its scratch rate runs high (about 20% against pros' 2%)Cut the second-row ball as thin as half-ball; it takes a decent amount of draw and possibly a bit of outside english to keep the cue ball out of the rack. McAuliffe says it makes the 8 “a LOT”.
Source: Dr. Dave, 8-ball break FAQ
Drawn: 7-ft table, cue ball 3.6 in of cloth off the left cushion, 30% draw, 10% outside (left from the left rail); about ¼-ball on the 9.
| Table | Speed | Made a ball | Balls | The 8 | Scratch |
|---|---|---|---|---|---|
| 7 ft | 16 mph | 62.0% | 0.90 | 1.5% | 16.4% |
| 7 ft | 19 mph | 71.4% | 1.16 | 1.9% | 20.1% |
| 7 ft | 24 mph | 82.5% | 1.55 | 4.9% | 31.3% |
| 9 ft | 16 mph | 44.4% | 0.57 | 0.7% | 10.0% |
| 9 ft | 19 mph | 54.3% | 0.76 | 1.5% | 11.9% |
| 9 ft | 24 mph | 65.5% | 1.03 | 2.4% | 19.1% |
Measured: Top pros' second-ball breaks (bar tables): made a ball 82.4%, scratched 2.0%, 8 on the break 5.9%.
From near the left rail the cue ball can run down the side of the rack and meet a ball in the 3rd, 4th or even 5th row first: the back corner ball sticks out furthest toward the rail, so a shallow shot reaches it before anything else. For the 8, the 3rd row is best: a ¾-ball hit on the near-side third-row ball gives two to four times the head-ball break's chance, on both tables at both speeds, and edges out the second-ball break. Going all the way to the back corner gives most of that back. The racks are random (no pattern racking), and thin hits from the rail are where the model over-scratches.
the 8 on the break: third-row ball from the rail
Best third-row setup at 16 mph: 2.3%. The centre head-ball break on the same racks: 1.2% at 19, 0.7% at 16.
the 8 on the break: third-row ball from the rail
Best third-row setup at 16 mph: 0.9%. The centre head-ball break on the same racks: 0.7% at 19, 0.2% at 16.
9-foot table, Magic Rack: the best chance of the 8 on the break at each speed, for the rail second-ball break and the centre head-ball break. From 16 to 19 mph the 8 drops about once in 100 breaks either way, with no clear winner; the second-ball break pulls ahead from about 21 mph and leads clearly by 25.
The best setup from near the rail against the best from inside a WNT break box, where the best spot is the box's edge to about 3 inches in, cutting the 1. At 19 mph the box costs about half the chance at the 1 on a 7-footer and a third on a 9-footer. At 16 mph the rail's advantage mostly disappears: for the 1, speed matters more than where you break from.
Ko Ping Chung's break, measured from video → Seven US Open breaks tracked frame by frame, and what the model does with the same break.
The model under pro-like conditions against measured pro statistics: Dr. Dave's 169 top-pro 8-ball games on bar tables, and the 2023 US Open's tracked 9-ball breaks on Diamond 9-footers. Thin connectors mean close; thick mean a real gap.
Close: the 8-ball lead-ball break on bar tables. Known misses: second-ball scratches (model 8–30% against pros' 2%), the 9 on a 9-footer (3–6% against 1.4%), and the 1 in the side on the box cut. No physics setting closes these; the likely cause is how the engine resolves a frozen rack, one pair of balls at a time. That fix is in progress, and these numbers will be re-run on it.