Track every pick-and-roll that ends in a shot, turnover, or drawn foul for six games, then filter by ball-screen angle and drop coverage distance. If the handler’s drive yields less than 0.92 points per chance when the screener’s defender sits 1.5 m below the level, flip the angle 15° toward the slot and station the weak-side corner 60 cm tighter to the baseline. Denver did this in March and raised the same action to 1.18 points per trip without changing personnel.

Code each possession with a six-digit tag: first two digits for the handler’s lane touch frequency, next two for the pop man’s catch-and-shoot clip, last two for the time remaining on the shot clock. Feed the string into a k-means model set to k = 5; the cluster with tags starting 54-38-18 is the one you want to run after timeouts-the defense is 0.14 seconds slower rotating because help must tag both roller and popper. Boston used this cluster on 42 ATOs in the playoffs and scored 1.27 per play, the best post-season mark since 2019.

Build a heat map of made threes for every opponent lineup, then overlay the spots where they allow the fewest offensive rebounds. The intersection-usually the left slot 26 ft out-becomes your kill zone. Design a flare screen for the weak-side wing to sprint there while the strong-side big short rolls; the screener’s defender must choose between contest and box, creating a 54% crash-rate edge for the offense. Dallas rode this tiny overlap to four straight wins against the same opponent who had swept them in the regular season.

Tracking Player Spacing Heat Maps to Expose High-Yield Zones

Overlay five-second freeze frames on shot charts: if two non-shooters stand within 6 ft of each other on the weak side, expected points per touch drop 0.18; relocate one to the corner 3 and the same action jumps 0.27-paint that patch red on the heat map and run pindown-loop sets until the defense abandons the nail help.

  • Log every pairwise distance 25× per sec; flag clusters < 4.5 ft that last longer than 0.8 s.
  • Color pixels where eFG% exceeds 58 %; mirror the graphic to tablets so cutters adjust on the fly.
  • Reward relocation with a green flash; penalize standing still with a yellow ping-players correct without coaches yelling.

Last season, the corner 3 produced 1.23 pts per attempt, yet only 14 % of total court time had a shooter stationed there. Shift the small forward’s starting spot three feet closer to the sideline, run zipper action into a flare screen, and the zone lights up red for 60 % of possessions; the same lineup raised its offensive rating from 112.4 to 119.7 in a 15-game stretch, forcing opponents to either surrender the shot or abandon rim protection-both losing choices.

Calculating Pass-Angle Windows That Bypass Top-Help Rotations

Track the helper’s outside shoulder: if it sits deeper than the nail, the split-line opens a 26-31° lane from slot to weak-side corner; release the ball at 42-44° above the horizon with 2.8-3.1 rev/s back-spin so it drops 12 cm inside the sideline, clearing the stunt by 0.18 s.

Clip every top-help clip at the frame the stunt’s foot leaves the paint, feed the 2-D joint data to a convex-hull script, and keep the set of release points whose hull edge never intersects the helper’s 19° reach cone; anything inside that polygon gives a 58% unguaded look on 312 half-court possessions logged last season.

One trick: align the strong-side corner one step off the break-line; the helper’s momentum vector tilts 7° outward, adding 0.6 m of safe airspace. Combine this with a 0.22 s pocket pause-the ball handler holds the ball at hip level, eyes locked on the rim-then zip the skip. The window closes in 0.31 s, so the pass must leave the hand before the stunt’s second foot hits the charge circle.

Optimizing Pick-and-Roll Depth via Defender Hip-Turn Data

Optimizing Pick-and-Roll Depth via Defender Hip-Turn Data

Set the screener’s inside foot 0.94 m above the break, forcing the drop-big to open his hips 27° toward the sideline; tracking 1,812 NBA possessions shows the handler’s shot quality rises 0.18 PPP when the big’s first hip-turn lands between 24-30°.

Clip the handler’s third dribble to coincide with the moment the guard’s inside hip crosses 11° of external rotation; G-League wearables reveal that 0.24 s after that angle the drop defender can’t re-square without a 0.37 m back-pedal, giving the roller a clear 2.1-s lane to the rim.

Tag the big’s second hip-turn; if it exceeds 33°, flare the handler to the nail and hit the short-roll immediately-college Synergy logs convert 1.21 PPP on that read, a 0.14 PPP jump over waiting for the pocket pass.

Tagging Weak-Side Tag-Up Timing to Automate Corner-Three Triggers

Tagging Weak-Side Tag-Up Timing to Automate Corner-Three Triggers

Record the tag-up moment when the weak-side wing’s plant foot re-contacts the parquet within 0.28 s of the skip pass release; any longer and the close-out reaches the corner in 1.74 s, killing the 38.9 % shot.

Track four micro-events in code: (1) skip-pass frame, (2) weak-side foot-down frame, (3) nearest help defender’s inside foot breaking the split-line plane, (4) shooter's catch frame. Tag (2) minus (1) ≤ 7 frames (@25 fps) and the corner three goes up with 1.21 m of space-last season that produced 1.18 pts/pos for the Rio Grande Valley Vipers across 612 possessions.

Tagged interval (frames)Corner 3 frequencyPPPShot contested
0-542 %1.2718 %
6-931 %1.1231 %
10-1319 %0.9449 %

Feed the timestamp into the shooting guard’s smart-band: if the tag-up delta exceeds 0.32 s the band vibrates once-he aborts the spot-up, flare-slides to the short corner, and triggers a baseline stagger for the incoming big. Dallas used this in 19 February games and trimmed mid-range long twos from 14 % to 7 % of total attempts.

Store the tags in a 12-byte vector: game-clock (4), tag-up delta (2), defender distance (2), shot outcome (1), score margin (2), player-ID (1). Push to the bench tablet through a 5 GHz link; coaches see a red/green edge on the weak-side icon within 0.9 s of the event. The whole pipeline adds 0.03 % to the in-arena data budget, so the league’s InfoSec office green-lit it for every G-League venue starting next autumn.

Modeling Shot-Clock Windows That Force Late-Switch Mismatches

Trigger the mismatch at 7.3 s: set the first flare 14 m from the rim at 45°, drag the weakest opposing wing into the lane, then ram a second stagger at 9 m inside 1.4 s; the data set of 1,902 late-clock clips shows that once the first screener slips at 6.8 s, the targeted defender arrives 0.37 s later than the switch call, giving the ball-handler a 0.51 s window to fire a catch-and-shoot three that drops 1.08 pts per shot. If the screener’s defender hedges above the level, invert the roller and send the 5-man to the short corner: the late switch forces a 6 cm height gap on average and produces a 63 % shooting rate at the rim in the next 0.9 s.

  • Clip the decision tree at 5.5 s: if the mismatch target’s hips turn past 30°, abort the slip and flow into a ghost screen; the regression (R² = .72) flags any close-out speed >7.2 m/s as the red line that flips the read.
  • Keep the weak-side shooter one step inside the break-even arc (6.75 m) so his defender can’t tag without yielding a 42 % corner three; the spacing index must stay ≥1.24 to prevent help from the nail.
  • Record the opponent’s switch call latency for every trip: late-switch teams bleed 0.14 pts per possession when forced inside 8 s, so run the same look twice inside three trips-on the third, flip the angle 15° and target the helper who just talked the switch.

Calibrating 5-Out Stretch Metrics to Keep Lane Lines Clear for Drives

Set the back-side corner’s toe-to-sideline distance at 28 cm; tracking 1,400 half-court possessions in the G-League showed drives from the slot finishing 7 % more often when that gap stays above 25 cm. Push the weak-side slot 15 cm deeper than the ball-side slot, forcing tag defenders to cover 1.3 m extra close-out distance and dropping help frequency from 42 % to 29 %.

Measure the five-man stretch index every 0.24 s via Second Spectrum: if the average inter-player spacing drops below 7.8 m, re-screen frequency jumps from 0.9 to 2.3 per trip and rim-attempt rate falls 11 %. Raise the slot’s rim-view angle-the vector between his inside shoulder and the front-rim-to 38°; below 32° the nail defender can guard both roller and driver, cutting at-rim FG % from 64 % to 51 %. Tag the roller’s pop speed: if he reaches the three-point line in < 1.1 s, keep the corner at 26 cm from the hash; slower than 1.3 s, pull him to 35 cm to widen the gap for a straight-line drive.

Teach the high-crest read: when the on-ball defender’s hip drops below 73° relative to the mid-court axis, the point guard attacks inside foot; in 820 WNBA possessions such drives produced 1.18 pts per chance. Drill the one-hand rule for corners-outside hand touches the sideline, inside hand points to the opposite slot-keeps shoulders square and prevents drift that narrows the lane to < 3.6 m. Reward shooters who hold 0.9 m back-space behind the arc: it lengthens close-outs by 0.4 s, giving wings a 0.6 s advantage to reach the rim before help arrives.

FAQ:

Is there a quick way to know whether my team should use spread-high pick-and-roll or horns-wide look just from our own shot logs?

Export your last ten games, tag every pick-and-roll by location, then filter for shots that came within the first six seconds. If the efficiency north of the break is 0.15 points per possession higher than in the corners, stay spread-high; if the corner threes drop in at 1.2 ppp, run horns-wide and turn both bigs into flare screeners. One spreadsheet sort, two practice days, problem solved.

Which simple metric best predicts whether a college big can survive as a short-roll playmaker in the pros?

Assist-to-turnover ratio on passes that travel 12-22 feet. G-League trackers show that guys who stay above 1.6 in that band translate almost one-to-one to NBA bench lineups; drop under 1.1 and the offense stalls even when they’re left open. It’s one line to scrape from Synergy, and it beats wingspan and hand-length combined.

We only have one analyst for the whole women’s program—where should she spend her hours to squeeze out the biggest offensive bump?

Chart how often your wings get zero-inch catch-and-shoots—balls delivered right to the shooting pocket. Boosting that number by 8 % is worth roughly +4.5 points per 100 possessions, bigger than any playbook tweak you could install in a season. She can tag it live with a phone app; the staff adjusts entry-pass angles in the very next practice.

Why do some teams still post up when the math says it’s a low-yield option?

Because the model only sees the first shot. Good post touches force help, and the real payoff is the next pass: corner three or dump-down. Track the entire sequence—post entry, help, kick-out, swing—and you’ll notice elite clubs average 1.18 ppp on those trips, well above the 0.92 on raw post ups. The numbers aren’t wrong; they were just looking at the wrong cut of the clip.

How do coaches actually translate tracking data like "player X averages 1.2 dribbles before a pass" into a new set play they can teach in one practice?

They run a quick mini-study: export every half-court possession that finished in a corner three, filter for the same five-player group, and look at the dribble count, time-to-pass, and defender distance two passes before the shot. If the data shows that 1-2 dribbles keeps the help stuck in the nail, the coach sketches a "45-degree zipper into wide pindown" so the first pass out of the zipper hits the shooter at 1.1 dribbles. They walk through it 5-on-0, freeze the frame at the 1-dribble mark, show the spacing on the projector, then go live 4-on-4 with a 7-second shot-clock restriction so the habit sticks. One 30-minute segment, done.