Skip to content
Fortiq
Empirical Primary Study · Fortiq Research · 2026-11-21

Fortnite Drop Physics & Glider Cut Mechanics: 23,930 Evaluated Landings

This primary empirical study measures every phase of a Fortnite Battle Bus drop directly from Unreal Engine replication flags across 246 cleanly parsed 100-player server replays (23,930 first-drop landings) plus 14 local 43.00/43.10 client replays cross-checked against surveyed M1 terrain rasters.

23,930
Evaluated server landings
97.8 %
Glider-cut prevalence
15.24 m
Final reopen height AGL
54.0 m/s
Crossbones Isle bus speed

1. Dataset & Exact Engine State Extraction

From an archive of 583 server-replay headers (builds 41.10–42.20, Hera_V2_Terrain), a stratified sample of 250 replays was split deterministically 70% train / 30% untouched holdout per build. 246 replays parsed with zero extractor errors (175 train / 17,053 drops; 71 untouched holdout / 6,877 drops). For Crossbones Isle (Delilah_Terrain), 9 local 43.10 (CL-60118918) and 5 local 43.00 (CL-59596594) replays were extracted at 0 ms sampling and checked directly against shipped M1 terrain rasters.

Extraction reads exact Unreal Engine replication flags (isParachuteOpen, isParachuteForcedOpen, skydivingState, isSkydivingFromBus, replicatedMovementMode, movementBase), accounting for 110 ms client replication delay. 97.8% (16,678 / 17,053) of train landings execute a post-forced glider cut; 90.5% (15,089 / 16,678) execute a final reopen prior to touchdown.

2. Measured Flight Parameters (Builds 41.10 – 43.10)

All skydiving, gliding, open/close/reopen transition, cooldown, redive, and final-reopen mechanics are invariant from build 41.10 through 43.10. Only Battle Bus horizontal speed varies by island (75.0 m/s on Hera_V2_Terrain vs 54.0 m/s on Crossbones Isle).

Parameter Server Train (17,053 drops) 43.10 Cross-Check (9 matches) Calibrated M2 Profile
Battle Bus speed & Z75.0 m/s (Hera), 800.0 m Z54.0 m/s (Crossbones Isle), 800.0 m Z54.0 m/s · 800.0 m
Bus-exit launch (0–1.75 s)7 × 0.25 s, Δz = 35.26 mMatches ego accel ramp (0–1.75 s)7 measured 0.25 s slices
First auto-deploy AGL100.0 m AGL over surfacep50 101.45 m (ground 99.5–103.0 m)100.0 m AGL
Open deceleration3 × 0.25 s (0.75 s), Δz = 15.00 mΔz = 14.22 m to 5.00 m/s0.75 s (3 slices)
Steady full-forward glide18.00 m/s horiz · 5.00 m/s down (IQR 0.00)18.00 / 5.00 m/s (IQR 0.00)18.0 / 5.0 m/s
Open → post-forced cutp05 2.17 s · p50 2.61 s (n = 8,235)min 2.06 s · p50 2.38 s (n = 9)2.10 s total cooldown
Close + redive accel0.25 s close + 1.00 s accel, Δz = 29.64 mΔz = 28.21 m to 40.00 m/s1.25 s transition
Terminal end-redive18.00 m/s horiz · 40.00 m/s down (IQR 0.00)18.00 / 40.00 m/s (n = 2,731)18.0 / 40.0 m/s
Final reopen AGLFast direct p50 14.36 m (all p50 14.98 m)Ego p50 16.86 m · Lobby p50 18.57 m15.24 m AGL (+0.88 m)
Final reopen decel4 × 0.20 s (0.80 s), Δz = 11.97 mΔz = 12.71 m to 5.00 m/s0.80 s (4 slices)

3. Measured Steady Dive Polar (262,497 Samples)

Time-weighted medians and interquartile ranges of vertical descent speed across horizontal velocity bins in the server train dataset.

v_h bin (m/s) Samples (n) v_h p50 (m/s) v_down p50 (m/s) v_down IQR (m/s)
[0, 2)1,7350.1760.000.05
[2, 5)2,8573.8059.380.52
[5, 8)7,8576.8457.990.88
[8, 11)17,1449.6955.891.34
[11, 14)25,33712.5852.851.89
[14, 16)18,98315.0449.381.63
[16, 17.5)16,96016.7746.321.48
[17.5, 18.5)171,62418.0031.25 (p50) / 41.76 (p90)12.41

4. Untouched Holdout Validation & Cursed Cavern Regression

Evaluated with zero refitting on the untouched 71-match server holdout, the M2 phase model achieves a 0.50 s median absolute arrival error (p90: 1.28 s) on fast clean 1-cut drops.

Cohort Evaluated Median |err| p75 |err| p90 |err| Median Bias
Server Holdout · Fast Clean 1-Cut (71 matches)1,130 / 1,1400.50 s0.88 s1.28 s+0.15 s
Server Holdout · All Direct 1-Cut (71 matches)2,775 / 2,7890.65 s1.23 s2.11 s-0.31 s
Server Holdout · Top-10 Fastest/Match (71 matches)647 / 7100.57 s0.97 s1.33 s+0.46 s
43.10 Local Direct Ego (CL-60118918, dir >= 0.90)7 / 70.64 s–0.91 s-0.34 s

Regression Benchmark: Cursed Cavern (Moosinsel Chest, 59.21 m Z)

Crossbones Isle scenario (58.81 m surveyed M1 terrain, 54.0 m/s bus at 800.0 m Z): legacy rough estimate lands at 36.7 s; M2 single-glide lands at 34.74 s (jump 8.92 s); M2 Cut + End-Redive + Catch touches down at 31.00 s (jump 10.52 s, cut at 26.33 s / 135.78 m AGL, final catch at 29.74 s / 15.24 m AGL) — 3.73 s faster with verified 200/200-cell terrain clearance.

Primary source: Fortiq Dropmap Benchmark Record (specs/160-professional-dropmap/benchmark.md, 2026-11-21). Aggregated evidence receipt SHA-256: ff0a3f7d1b96ef4fb0489fd1f779a71b9c6f4b3c7d9e163777473d710c9a3c04.

Drop Physics Research FAQ

How often do competitive Fortnite players cut their glider on initial drop?

Across 17,053 training landings from 175 server replays, 97.8% (16,678 / 17,053) executed a post-forced glider cut, and 90.5% (15,089 / 16,678) triggered a final reopen before touchdown.

At what height AGL does final glider reopen happen in Fortnite?

The calibrated M2 model places final reopen at 15.24 m AGL (14.36 m measured direct-redive median plus 0.88 m character capsule offset), followed by 0.80 s deceleration to 5.00 m/s descent.

What is the Battle Bus speed on Crossbones Isle in Patch 43.10?

On Crossbones Isle (Delilah_Terrain, 43.00 and 43.10), the Battle Bus travels at 54.0 m/s at 800.0 m world Z, compared to 75.0 m/s on Hera_V2_Terrain. Skydiving, gliding (18.00 / 5.00 m/s), and terminal end-redive (18.00 / 40.00 m/s) velocities are invariant across builds 41.10–43.10.

How much faster is glider cutting compared to continuous gliding?

In the verified Cursed Cavern regression scenario (Moosinsel chest spawn at 59.21 m Z over 58.81 m M1 terrain), M2 Cut + End-Redive + Catch touches down at 31.00 s bus clock vs 34.74 s for single-glide — saving 3.73 seconds.

Apply These Measurements on the 43.10 Map

Calculate your exact jump vector in guest mode with no account wall, or explore all 12 Crossbones Isle POIs.