Suppose P4 buys three times in A just after a positive factor bulletin. A few ticks later, the same seat buys the index. You still do not know what signal P4 received, or whether the first trades were good. But you have something worth tracking.
That example is the reason the tape has names on it. Continual Exchange is partly a valuation game and partly a game about deciding whose actions deserve attention.
What appears on screen
The book publishes aggregate quantity at every price level. It does not publish the owner of a resting order. If there are 12 lots offered at 1004, you cannot tell whether they belong to one seat or four.
A completed trade is more detailed. The print includes the instrument, price, quantity, aggressor side, aggressor pseudonym, and resting pseudonym. Competitor labels such as P1 and P4 last for one match. House bots use disclosed H labels.
Why a fully anonymous tape felt too thin
Without attribution, a bot can measure total buying and selling, but it cannot separate one persistent participant from several unrelated trades. That removes a useful inference problem. A seat that repeatedly crosses the spread after factor news should look different from a seat sending occasional noise orders.
It also makes replays harder to study. You can see that the market moved, but not that one participant accumulated early, another supplied liquidity throughout the move, and a third arrived late.
Why the book does not show owners
Showing resting ownership would reveal too much, too cheaply. A bot could identify a seat it considers informed and avoid that seat's quotes before any trade occurs. Cancellation patterns would become a nearly free signaling channel. Providing liquidity would disclose the strategy before it had taken any risk.
The current rule is simpler: an unfilled quote keeps its owner private. Once a trade transfers inventory, both participants appear on the tape.
What a bot can infer
Attribution does not turn the tape into ground truth. P4 might have a good signal. It might be overreacting to a noisy bulletin. The index purchase might be a second expression of the same view, or it might hedge exposure from the stock book.
The next trades help distinguish those stories. Does P4 keep paying the spread? Does activity appear in B? Does the market move toward the prices P4 accepted? A bot can maintain a per-seat belief and update it as the match unfolds.
This matters because signal quality is hidden. A seat with a weak private signal is not limited to blindly following it. It can put more weight on public news, cross-book relationships, and participants whose trades were followed by confirming news or market movement.
Why pseudonyms reset
The labels exist to make one match legible. They are not meant to create a permanent opponent profile. Seats and pseudonyms are randomized between matches, so a strategy has to infer behavior again from the current tape.
That keeps attention on the market in front of the bot. Public replays can preserve the story of a match without publishing private signals or a durable identity next to every action.
The design still has to earn its keep
Attribution is worthwhile only if reading it can improve decisions. We test that claim by comparing a flow-reading bot with an otherwise identical version that receives the same books and news without seat labels, across the same match seeds and seat rotations.
If attribution never improves decisions under those controlled comparisons, the tape is carrying detail that does not change the game. That makes the design choice measurable rather than a matter of taste.
The book tells a bot where liquidity is available. The tape adds a record of who chose to trade. Keeping those two views different gives passive orders some privacy and active decisions a history.