Insights
Why a Padel Live Stream Doesn’t Work Without a Scoreboard
In short: Streaming is the solved half. Cameras are cheap, encoders are cheap, and every club can push a match to YouTube this afternoon. What almost no club has is the one thing that turns that video into something anyone watches: the score. A stream without a live score is a security camera pointed at a padel court. And the score cannot come from the camera — it has to come from a scoring system that already knows it. Streaming does not compete with scoring. It depends on it.
What is actually missing from most padel live streams?
Open any club’s stream mid-match. Four people are playing. The rally ends. Someone celebrates.
You have no idea whether that mattered.
You do not know the game score, the set score, who is serving, who is ahead, or whether you have joined at 1-1 in the first or 5-4 in the third. There is no way to acquire that information except by watching long enough to reconstruct it yourself — which nobody does. The viewer either already knows the players personally, or leaves within about twenty seconds.
This is the entire problem. It is not resolution, not camera angle, not bitrate, not commentary. Every one of those can be poor and the stream still works if the score is on screen. All of them can be excellent and the stream still fails without it.
The score is what converts footage into a match. It supplies the two things a viewer needs and cannot get any other way: context (what is happening) and stakes (why it matters). A break point is only a break point if the overlay says 30-40.
Why can’t the camera just read the score?
This is the question every club asks, and it has three separate answers, all of which point the same direction.
There is usually nothing to read. Most padel courts have no scoreboard at all. The score exists in four players’ heads and is disputed roughly once a set. A camera pointed at a court with no scoreboard has no source of truth available to it, no matter how good the model is.
Where there is something to read, reading it is the fragile part. A wide-angle court camera sees a small, distant display through glass, often with one side of the court in direct sun and the other in shade. Reading a dot-matrix panel reliably under those conditions is a harder computer vision problem than it sounds, and it fails exactly when the stream matters most — bright afternoons, finals on the outside courts.
And even when it works, the dependency has not gone away. A camera reading a scoreboard is still a system that requires a scoreboard. You have not removed scoring from the stack. You have added a lossy, error-prone step in the middle of it, when the scoreboard could simply have sent the number.
The remaining option is inferring the score from play itself — tracking the ball, judging in and out, deciding who won the point. As covered in Scoring as a Service, this is a different category of product. AI camera systems are genuinely good at highlights, movement analytics, and shot data. Scoring is the one thing they are asked to do that has no tolerance for being mostly right. A highlight reel that misses a good rally is a slightly worse highlight reel. A score that is wrong is not a score.
Why does the overlay have to be live?
Because the value of a live stream is the “live” part.
A score corrected afterwards, or a results sheet typed in later that evening, produces an accurate archive and a worthless broadcast. The parent watching from work, the partner at home, the player’s coach in another country — they are all watching now, and the number has to be right now. Latency of a few seconds is survivable. Latency of a point is not: the overlay updates after the next rally has already started, and the viewer is permanently one point behind the thing they are looking at.
This is the same constraint that governs the scoreboard on court, for the same reason. The players will not wait for the score, and neither will the viewer.
What does a stream actually need from a scoring system?
Not a picture of a scoreboard. A structured event feed:
| Event | What the stream does with it |
|---|---|
| Point won, by team | Updates the overlay; marks a clip boundary |
| Game / set transition | Updates the overlay; marks a chapter |
| Server changed | Shows the serve indicator |
| Match start / end | Starts and stops the recording; closes the result |
| Player and team names | Populates the lower third |
| Timestamp on every event | Indexes the video |
That last row is the one clubs consistently underestimate. A timestamped point-by-point feed is an index of the match. With it, a ninety-minute recording becomes navigable: jump to the tie-break, clip the four match points, auto-generate a highlight reel from the longest rallies, cut a thirty-second social edit without a human scrubbing through the file. Without it you have an unwatchable ninety-minute video that nobody, including the players who are in it, will ever open again.
So the score feed is doing four jobs at once: driving the overlay, indexing the video, closing the result into the bracket or league table, and feeding the ranking. One input, four outputs. The camera provides none of them.
Which half of “streaming plus scoring” is actually hard?
Streaming is a solved, commoditised problem. The encoder is a small Linux box. The transport is RTMP. The destination is YouTube, and it is free. Court cameras are inexpensive and getting cheaper every year. Nothing about putting padel video on the internet is difficult in 2026, which is precisely why so many products do it.
Scoring is not commoditised, and the industry knows it. It is the piece that every booking platform, federation and ranking system needs and none of them has solved, because it is not primarily a software problem — it requires hardware on every court, in the players’ hands, that four tired people at 4-4 in the third will actually use.
Which sets the order of operations. A company with scoring can add streaming in a quarter. A company with streaming cannot add scoring at all — not without going and building the on-court hardware layer they skipped. This is why streaming products keep arriving at scoring from the wrong side, and why so many of them stall there.
For a club, the same logic inverts a common purchasing decision. Streaming looks like the glamorous purchase and scoring looks like the boring one. It is the other way round: scoring is the asset, and streaming is a feature you unlock once you own it.
What goes wrong when a club buys streaming first?
The usual sequence:
The club installs cameras on the show court. The first week is exciting. Members watch their own matches back. Then the numbers fall, for three reasons that all trace to the same root.
No context, no retention. Nobody outside the four players can follow the match, so nobody outside the four players watches. The audience never grows past the people who were there.
No index, no clips. The club has hours of footage and no way to cut it, so the social content that was supposed to justify the spend never gets made — or gets made by hand, once, by whoever had a free Sunday.
One court, no coverage. The stream covers the show court, which means it covers a small fraction of matches played. Members whose matches are never on it stop thinking of it as something the club offers.
None of these are fixed by a better camera. All three are fixed by the score feed, which is upstream of every one of them.
Does this apply to club streaming as well as tournaments?
More so, in fact.
Tournaments have a referee, a draw, and someone whose job is the result. They can — badly, expensively, manually — put a human on the overlay for the final. Most tournaments do something like this, and it is why finals streams look acceptable and everything else does not.
Club play has none of that. There is no referee, no operator, and no budget for one. Whatever appears on the overlay has to be produced by the players themselves as a side effect of doing something they were going to do anyway: pressing a button to move the score on the scoreboard they are already using. That is the only mechanism that scales past the show court to every court in the club, every night of the week.
Which is the same conclusion the scoring argument reaches from the other direction. Coverage is the unit that matters, not the flagship installation.
What does the minimum working system look like?
Three parts, and no more:
A scoreboard on every court that the players use because it is the fastest way to keep the score — dual-sided, visible from both ends, audible on every change, with the button in the player’s pocket rather than on the glass.
A feed from that scoreboard, in real time, carrying point-by-point events with timestamps.
An encoder that consumes the feed and burns the overlay into the video on the way out to YouTube — score, names, serve indicator, updating within a point.
That is the whole architecture. LedsCount runs it on a Raspberry Pi 5 with a node-canvas overlay pushing RTMP, which is a deliberately unremarkable set of parts. The interesting component is not the encoder and not the camera. It is the fact that the score arriving at the encoder is already correct, because four players on court agreed on it in real time and confirmed it on a display they can both see and hear.
Everything downstream of that is straightforward. Nothing downstream of it is possible without it.
FAQ
Can AI cameras generate a live score overlay for padel? They can generate an overlay. Whether the number in it is correct point after point, set after set, with no operator watching, is the open question — and the tolerance for error in a score is zero, unlike in highlights or analytics where these systems perform well.
Do I need a scoreboard on every court to stream? To stream one court, no. To have a streaming offer that members actually use, yes — because coverage is what makes it a club service rather than a show-court novelty.
Can the score be added to the video afterwards? For an archive, yes. For a live stream, no: the audience is watching now, and a score that arrives later is not a score anyone saw.
What about YouTube’s own overlays or a graphics operator? Both work and both require a person for the duration of every match. That is affordable for a final and impossible for a Tuesday evening across eight courts.
Does the streaming hardware have to come from the scoreboard vendor? No — but the score feed does have to come from somewhere, and an open feed from the scoring system is what lets any encoder, platform or broadcaster use it.