I'll give examples of each: market microstructure, exchange protocols and connectivity.
Probably the most well-known in the public domain is the idea of a "canary order". In some options firms this might be called a "tachyon" in reference to its virtual faster-than-light property. Others simply call it a "private fill (ack)".
In some markets, an order acknowledgement publishes sooner on the order gateway than the same event publishes on the public market data. This microstructural property allows you to build your own order book representation that is ahead of other participants who were not involved in the fill.
This is where the researcher comes in. For an options trading firm, modeling shifts in the curve conditional on large moves is important. Orders of unit size can be layered across multiple levels deliberately to monetize this microstructural property on large moves. Similarly, the trading platform can be optimized to publish the private fill between different strategies so that orders with regular intent also contribute. In other markets, the opposite effect is true, and it is someone's responsibility to understand the distribution other than the systems engineers or developers.
MPIDs and feed arbitration
Very commonly in equity options protocols, MPIDs are available for you to profile other market makers, build a signal, and follow up differently based on the firm that you have just traded against. Because of this, a well-known phenomenon is that smaller options trading firms might simply follow a large market maker like Optiver on the rolls.
Likewise, you can hide your own intent by striping your orders across multiple brokers using sponsored access.
Understanding the underlying also helps significantly with an options trading strategy, and equity markets have multiple protocols that publish redundant or overlapping information. Heterogenous systems often invite opportunity - this creates the possibility that one can subscribe to multiple feeds for the same market to build an accurate order book representation faster than other participants.
Markets have different matching architecture:
- Some have multiple, public gateways which handle incoming orders for the same set of symbols.
- Similar to this is an architecture where there's a mix of private and public gateways, and you're allowed to pay the market operator an extra fee for "high performance connectivity" (or whatever marketing term they use for it) which guarantees you a dedicated, private gateway.
- Others have multiple gateways which divide the universe into different subsets of symbols.
This is where understanding the connectivity comes in. Say, a strategy may benefit from maintaining multiple sessions and routing to the gateway with the least traffic.