In my last chost, I pointed out the areas where Puyo Spectator Assist fails to convey key information about game state. With that being said, I think PSA does scratch the surface of a notoriously opaque, yet critical component of competitive Puyo: maximum possible chain. In fact, I believe MPC is the special sauce that makes high-level Puyo play possible.
Ask an experienced Puyo player what the game is about, and you'll get a range of answers: "It's a game of chicken", "It's a race to see who can build the biggest chain the fastest", "It's a puzzle game with footsies" are all responses I've read in the past. All these refer to the task of building a chain to overwhelm the opponent, whether it be through absolute domination of board state or by creating a maximum chain situation through a well-timed attack. Flexible building to maximize one's own chains and keen observation of the opponent's maximum possible chain are all differentiating factors that make Puyo such a deep and rewarding game.
Let's take a look at what "maximum possible chain" might entail, starting off with a cut-and-dry set of board states. hinata1112 on the right clearly has a maximum possible chain of 6, while yosshi on the left, due to being buried in garbage, can only hit a 2 chain maximum despite having additional chain links placed down at the bottom of the board. PSA would be able to accurately identify both maximum chains here.
In the next game, however, a situation has emerged which makes each players' maximum chain less well-defined. At this moment, both players have cut off access to their main chain as a result of building on top of their GTR. yosshi on the left currently is actually able to fire off their main chain, but needs a very specific green-purple piece in order to do so, which is not present in the upcoming piece queue and otherwise only has a 1/8 chance of spawning. hinata1112, on the other hand, is much farther away from readying up their main chain - needing two blues and 3 yellows in very specific locations on their board.
The latter boards raise a pair of questions: at what point should a main chain detector start (and stop) using look-ahead techniques to establish max chain length; furthermore, how can we algorithmically create chaining oracles? One branch of techniques are an extension of PuyoSpectatorAssist -- running simulations using whole pieces instead of single puyo, or maybe running Monte Carlo simulations with random puyos. The second general strategy would be building an artificial intelligence to read chains. Both strategies have their own associated challenges, and thus need engineering talent to figure out.
Once the MPC has been calculated, the task becomes how to incorporate this information into an effective visual aid. One inspiration comes from the "showdown" graphical flair introduced in Puyo Puyo Champions. Perhaps each player's maximum possible chain could be represented in a tug-of-war format, with visual intensity representing how close that potential chain is from completion. Here, on top of the technical challenge of rendering the desired information on a stream overlay, graphical design chops are needed (which I clearly don't have lol).