Rocket League Spacing Isn’t One Fixed Distance.
Pacifist Labs started with 10,882 Rocket League replay files and used 10,488 deduplicated 2026 matches for the main rank analysis. Better players generally played closer together, but at elite level total teammate distance was basically a coin flip. The more interesting pattern was where the space sat and how it changed with the play.
Two players can be the same distance apart and still have completely different protection.
Second-man spacing is normally talked about as a distance. Stay close enough to support. Do not get too close. Leave a few car lengths. The problem is that one straight-line number can describe very different positions.
A second man can be 3,000 Unreal Units from first man because they are safely behind the play, or because most of that gap sits sideways. The total distance is the same. The geometry is not.
Better players were closer overall. Within elite matches, being closer did not separate the winner.
Average teammate spacing generally compressed as the level increased. But when I compared the two teams inside the same matches, the simple idea that the closer team should win stopped working at the top.
How often the team with more total distance won
| Rank | Higher distance won | Winner difference | q-value |
|---|---|---|---|
| Platinum 1 | 54.2% | +51 uu | 0.000516 |
| Diamond 1 | 53.1% | +41 uu | 0.000438 |
| Champion 1 | 52.0% | +13 uu | 0.0833 |
| Grand Champion 1 | 49.4% | −6 uu | 0.607 |
| Supersonic Legend | 50.1% | −11 uu | 0.297 |
| Professional | 48.1% | −13 uu | 0.170 |
The raw result changed when I measured how much space sat behind first man.
Instead of total teammate distance, I measured longitudinal depth. This is the front-to-back separation behind first man along the team’s attacking direction.
In the raw paired comparison, the team with more longitudinal depth won more often at every rank tested. The individual rank comparisons below were statistically significant after false-discovery-rate correction.
Total distance cannot tell us where the protection sits
Depth-oriented
Second man keeps more of the separation behind first man. If the next interaction goes badly, another layer can still exist behind the play.
Lateral-heavy
The players can have the same total gap while more of the separation sits sideways. One forward action can still threaten both players.
Longitudinal depth by rank
| Rank | Winner advantage | Higher depth won | q-value |
|---|---|---|---|
| Platinum 1 | +166 uu | 59.9% | 1.78 × 10−15 |
| Diamond 1 | +187 uu | 61.7% | 4.04 × 10−35 |
| Champion 1 | +151 uu | 58.5% | 2.54 × 10−24 |
| Grand Champion 1 | +154 uu | 60.6% | 3.80 × 10−23 |
| Supersonic Legend | +180 uu | 63.2% | 1.33 × 10−41 |
| Professional | +198 uu | 64.3% | 4.96 × 10−46 |
That is an association, not proof that sitting further back causes wins.
The first obvious alternative explanation is that winning teams may simply spend more time attacking. So I tested that next.
What if winners only looked deeper because they attacked more?
Attacking teams naturally have more pitch behind first man. So I repeated the comparison using only matches where the winning and losing teams spent almost the same percentage of the game in the attacking half.
The two teams had to be within two percentage points of each other for attacking-half time. That left 1,484 matches.
95% confidence interval 58.9% to 63.8%.
Paired Wilcoxon p = 7.33 × 10−24.
The result also remained when the teams were close in attacking time and within one shot of each other.
That stricter subset contained 796 matches. The higher-depth team won 61.2%, with a 95% confidence interval of 57.7% to 64.5% and Wilcoxon p = 8.79 × 10−12.
It still does not mean further back is always better. It means we need to look at the situations more closely.
The amount of useful depth changed massively depending on where the ball was.
I split Professional play into defence, midfield and attack and compared winning and losing teams with the ball in the same general area of the pitch.
Longitudinal depth inside each third
| Ball area | Winner depth | Loser depth | Difference | Higher depth won |
|---|---|---|---|---|
| Defensive third | 861 uu | 830 uu | +31 uu | 54.5% |
| Midfield | 1,932 uu | 1,845 uu | +88 uu | 58.1% |
| Attacking third | 2,568 uu | 2,444 uu | +124 uu | 58.5% |
The winning team showed slightly more longitudinal depth in all three areas, with the biggest observed difference in attack. I have not yet run a separate significance test for each individual third, so these three rows should be treated as descriptive rather than three independently confirmed effects.
The typical gap expands as the play moves upfield
| Ball area | Central 50% longitudinal depth | Median depth | Median total spacing |
|---|---|---|---|
| Defensive third | 731 to 986 uu | 851 uu | 2,253 uu |
| Midfield | 1,699 to 2,147 uu | 1,919 uu | 2,867 uu |
| Attacking third | 2,242 to 2,817 uu | 2,516 uu | 3,311 uu |
Almost identical total spacing. Very different geometry.
The fresh 2026 player export contains 34 Zen matches and 35 Nass matches. Their average total second-man spacing was almost identical, but Zen placed much more of that separation behind first man.
Second-man total spacing
Their total spacing was almost identical.
The bars use a metric-specific scale so the differences remain visible. Exact values are shown beside them.
Where Zen and Nass differed
| Metric | Zen | Nass | Difference | q-value |
|---|---|---|---|---|
| Total second-man spacing | 3,094 uu | 3,115 uu | −21 uu | 0.881 |
| Longitudinal depth | 817 uu | 458 uu | +360 uu | 0.000894 |
| Lateral separation | 1,823 uu | 1,924 uu | −101 uu | 0.493 |
| Behind first man | 69.1% | 62.0% | +7.1 pp | 0.000894 |
| Behind ball as second man | 74.2% | 67.5% | +6.7 pp | 0.000894 |
| Attacking-half time | 38.6% | 40.1% | −1.5 pp | 0.474 |
No. Nass actually spent slightly more time in the attacking half.
After accounting for attacking-half time, Zen was still estimated to maintain about 355 uu more longitudinal depth. That does not prove Zen won because he was deeper. It shows that the difference between their spacing styles cannot simply be explained by Zen attacking more.
String Theory is a moving relationship, not a tape measure.
Think of a piece of string between you and your teammate. Too much slack and you start occupying the same space. Stretch it too far and you disconnect from the play.
The research does not prove one exact pad count is optimal. The pad ranges below are my existing coaching cues for applying the idea in game.
Building an attack
I normally aim for around two boost pads centrally behind first man. Close enough to continue the play without driving underneath it.
Play collapsing back
As the play becomes less secure and starts moving towards our goal, I normally stretch towards the three-pad end of the range.
Deep defence
Once first man reaches the trigger pad, the small pad between midfield big boost and back big boost, second man’s cue becomes getting behind the goal line.
Counterattack
Compress quickly when the situation demands it. On a counter, being immediately available can matter more than preserving the normal gap.
Being connected to an attack does not mean sitting underneath first man.
When the ball was in the attacking third, winning Professional teams averaged 2,568 uu of longitudinal depth. That is not a magic target. It shows that substantial space behind first man can be completely compatible with successful attacking play.
Stay connected, keep enough depth to protect the next interaction, and change the gap with the game.
The result is useful, but there is more context to separate.
The research is observational. A relationship with winning does not prove that artificially adding depth will cause a win.
Matching attacking-half time removes one obvious explanation, but it does not make every game situation identical.
The three Professional pitch-area comparisons are descriptive until the separate per-area significance tests are run.
The next useful split is counterattacks versus sustained attacks, because those situations can demand very different support depth.
Secure possession versus contested possession should also be separated more clearly in future work.
Named player samples are examples of different styles, not proof that one professional’s positioning caused their results.
What was tested and how.
Deduplicate matches
Duplicate uploads of the same Rocket League match were collapsed before the main analysis.
Compare both teams
Winner and loser came from the same replay, keeping the comparison inside the same match.
Measure the geometry
Total spacing, longitudinal depth and lateral separation were treated as different measurements rather than one number.
Correct repeated tests
Paired tests used Benjamini-Hochberg false-discovery-rate correction where multiple comparisons were displayed together.
Dataset
The complete V3 export contained 10,882 source replay files. Match-GUID deduplication reduced this to 10,698 unique matches. The primary 2026 rank analysis used 10,488 matches.
- Platinum 1 · 1,131 matches
- Diamond 1 · 2,050 matches
- Champion 1 · 2,033 matches
- Grand Champion 1 · 1,478 matches
- Supersonic Legend, recognised professionals excluded · 2,026 matches
- Professional · 1,770 matches
Professional game-state analysis
The separate String Theory analysis exported 1,632,584 one-second states. The primary version required active play, stable first-man and second-man roles, both teammates behind the ball, second man behind first man, first man within 1,000 uu of the ball, and stable-role ages of at least one second. Sensitivity checks using 600, 1,400 and 1,800 uu involvement limits produced the same central moving-spacing pattern.
Statistical testing
The main winner-versus-loser comparisons used paired Wilcoxon signed-rank tests. Benjamini-Hochberg false-discovery-rate correction was used for groups of displayed tests. A corrected q-value below 0.05 was treated as statistically significant.
The attack-time check directly restricted matches to teams within two percentage points of each other for attacking-half occupancy before repeating the paired comparison.
Zen and Nass
The fresh 2026 named-player export contained 34 Zen matches and 35 Nass matches. The six displayed profile comparisons use match-level player observations and Benjamini-Hochberg correction across those six tests. The attack-time adjustment is an exploratory check asking whether the difference in longitudinal depth remains after accounting for attacking-half occupancy.
What longitudinal depth means
Longitudinal depth is second man’s front-to-back separation behind first man along the team’s attacking axis. Lateral separation is the sideways component. Total spacing is the ordinary straight-line distance between the teammates.
Stop looking for one perfect gap.
Better players generally play closer overall, but simply being closer does not separate elite winners from losers. Longitudinal depth shows a much stronger raw relationship with winning, and that relationship remains when the simplest attacking-time explanation is removed.
At the same time, Professional positioning shows that useful spacing is not fixed. The front-to-back gap is much smaller in defence than it is during attacking play. The better question is not “How many units should I be from my teammate?” It is “What does this situation require, and am I still protecting the play?”
Protect the play.
Adapt the gap.
The aim of Pacifist Labs is not to force every result into a coaching rule. It is to test the idea, show what the data supports, and make the explanation more precise when the answer gets more complicated.
The goal isn’t to prove Pacifist right. The goal is to get Rocket League right.Learn the framework, then measure your own replays.
The Digital Edition teaches the complete Pacifist System. Pacifist Score measures the positioning, role and decision-making behaviours appearing in your own Rocket League 2v2 matches.