Triggers vs manual entry: when automation actually earns its keep
A trigger is not "set and forget". It is a way to be present on a route you cannot watch — and it fails differently than you do.
Manual entry wins when the route is one you are already watching. You see the book, you feel how fast it is moving, and you can abort halfway through if the second leg looks thin. Nothing automated reads a market that way.
A trigger wins on everything else — the other sixty routes that are not on your screen, the window that opens at 04:00, the alt that reprices while you are placing an order somewhere else. An arbitrage edge is a coverage problem before it is a speed problem, and a human covers one route at a time.
The trade-off is failure mode. A human hesitates and misses the fill; a trigger fires and then discovers the second leg cannot be filled at the size it assumed. That is why the threshold you set matters more than the automation itself: a trigger armed at a spread barely above round-trip fees will open one-sided positions and call them arbitrage.
The practical split most people land on: manual for the two or three routes you actually understand, triggers for breadth, and a spread threshold on the triggers wide enough that a partial fill on the second leg still leaves the position worth closing.