THE LAB (PART 2)
Key Takeaways
- This story reflects a key shift defining the maturity and expansion of the global pickleball landscape in July 2026.
- Decisions and infrastructure investments made now are establishing the long-term foundations of the sport.
When a Rally Becomes a Chain of Decisions
How PEWPEW Analytics, DUPR, and a conversation with Scott Ross changed the way I think about pressure in pickleball
There is a tendency in sport analytics to believe the next breakthrough will arrive as a model.
A better algorithm. A cleaner dataset. A more sophisticated rating system.
But the more time you spend inside emerging sports like pickleball, the more you realise something else.
The breakthrough is usually not a model.
It is a way of seeing.
This is a continuation of a conversation I have been having with Scott Ross of PewPew Analytics over the past few weeks — a discussion that started around shot-level valuation and gradually moved somewhere more interesting.
Not toward disagreement.
But toward layering.
The first layer: explaining what happened
Scott’s work with PEWPEW Analytics sits at what I would describe as the most immediately useful level of modern sports data.
It takes shot-level video input and assigns probabilistic value to each action based on how it changes the likelihood of winning a rally.
In simple terms, it tries to answer three questions:
* why a rally was won or lost
* where value was created or lost
* and how different phases of play contributed to that outcome
This is not traditional statistics.
It is not winners, errors, or percentages in isolation.
It is a system that distributes credit and blame across a rally based on context.
A third-shot winner, for example, is not treated as an isolated event. It is weighted differently depending on the quality of the serve and return that preceded it.
As Scott put it, the aim is not to evaluate moments, but sequences.
That distinction matters.
Because most players think in moments.
Where the conversation began to shift
Where things became more interesting was not in disagreement, but in the gap between explanation and interpretation.
Scott’s framing, as he explained it, is fundamentally a probabilistic valuation system:
Each shot changes the probability of winning the rally, and PEWPEW attempts to capture that shift in a structured way.
It is, in effect, a framework for answering:
how did this rally unfold, and where did the value actually come from?
That is already a major step forward for a sport still dominated by surface-level metrics.
But it also opens a second question.
One that is less about outcome, and more about structure.
The second layer: pressure as a shifting state
Where my own thinking has been moving is slightly underneath that framework.
Not in opposition to it, but as a refinement of what it is actually describing.
Because when you watch enough doubles pickleball, something becomes clear.
Rallies are not really made up of isolated “good shots” or even clearly defined phases.
They are made up of small shifts in control.
Not possession in the football sense.
Something more fragile.
A sequence of micro-transitions where each shot does one of a few things:
* increases pressure
* maintains pressure
* relieves pressure
* or transfers it to the opponent
And those shifts accumulate over time until the rally breaks.
The winner is often just the final visible expression of that breakdown.
Not the cause of it.
Where Scott and I converge
What became clear through the conversation is that this is not a disagreement.
It is a layering problem.
Scott’s PEWPEW framework already captures much of this implicitly.
As he put it, if shot values are correctly estimated in context, then over time they necessarily reflect:
* pressure creation
* pressure absorption
* pressure transfer
* and eventual pressure collapse
Even if those terms are not explicitly used.
That is an important point.
Because it suggests that probabilistic valuation and “pressure-based” thinking may not be separate systems at all.
One may simply be a mathematical expression of the other.
Why DUPR sits in the middle of this
Another thread in the conversation, which Scott has explored in detail, is DUPR.
Not as a rating system to be criticised or defended, but as a behavioural reality.
Millions of players already engage with it.
Not because they fully trust it.
But because it gives them a number that represents them.
The issue, as Scott frames it, is not disagreement with the number itself.
It is absence of explanation.
Players do not necessarily think their rating is wrong.
They think they do not understand why it is what it is.
That gap between outcome and explanation is where both DUPR and systems like PEWPEW begin to overlap.
One provides the number.
The other attempts to explain it.
The third layer: longitudinal identity
Where this becomes more interesting is not in single matches, but across time.
Because neither ratings nor shot-level breakdowns fully answer the question most players actually care about.
If you’re following how the global game is shifting day by day, the World Pickleball Report breaks this down in our daily briefing.
Not:
* what happened today
But:
* what kind of player am I becoming
This is where Scott and I again converge, even if from slightly different directions.
He frames it as a probabilistic explanation system that can extend into longitudinal tracking.
I have been thinking about it as something closer to identity formation through data.
Not just performance measurement.
But pattern recognition over time:
* where a player consistently creates advantage
* where they consistently lose control
* and how those patterns evolve across sessions
This is not coaching advice.
It is structural self-understanding.
And that distinction matters.
Where analytics becomes interpretation
There is a temptation in sports analytics to move quickly from measurement to instruction.
If data exists, the assumption is it should tell you how to improve.
But both Scott and I have circled back to the same concern.
That may not be the first useful step.
The first useful step may simply be:
understanding what is actually happening in a way that feels consistent with reality
Not advice.
Not drills.
Not optimisation.
Just explanation.
Once that exists, improvement becomes contextual rather than generic.
The underlying problem neither system has fully solved yet
There is still a fundamental challenge sitting underneath all of this.
Pickleball is not a stable possession sport.
Control shifts constantly.
A rally can change direction in a single shot.
That makes it very difficult to separate:
* individual action
* partner influence
* opponent behaviour
* and structural momentum
Which is why both probabilistic models and pressure-based thinking are trying to solve the same problem from different angles.
How do you describe a game where control is continuously in motion?
The emerging answer
What this conversation has clarified for me is that the future of pickleball analytics is unlikely to sit in a single layer.
It will not be:
* pure ratings
* pure coaching AI
* or pure shot-level valuation
It will be a stack.
With at least three layers emerging:
Scott’s work sits clearly in the first:
probabilistic shot valuation and rally explanation
The second layer is still forming:
pressure and state transitions within rallies
And the third layer, which neither system fully solves yet:
longitudinal identity across time
How a player actually plays, not just what they did.
Closing thought
The most important insight in all of this is not technical.
It is structural.
We are not trying to describe isolated shots.
We are trying to describe a moving system of control that never fully stabilises.
Scott’s contribution to this conversation has been to formalise how value flows through that system.
My own contribution here is probably to suggest that what we are measuring is not just value.
But pressure in motion.
And somewhere between those two ideas, a new language for understanding pickleball is beginning to form.
Not finished.
Not complete.
But increasingly difficult to ignore.
Attribution note
Concepts and system architecture related to PEWPEW Analytics, probabilistic shot valuation, and rally-level expected value modelling are drawn from ongoing discussion with Scott Ross (PewPew Analytics). Interpretive framing around pressure states and micro-transitions represents independent synthesis built from those shared foundations.
📖 Read the Full July 2026 Issue
This article appears in Issue #18 of World Pickleball Magazine — download the complete edition free.
