How I structure the London open → NY overlap without overtrading the first hour

[PTScalper]

Architecture Note for High-Volume Execution

cTrader's LINQ queries against the History collection are generally very fast. However, if your daily history pool becomes excessively large across thousands of trades, evaluating CountTicketsInSegment inside OnTick could introduce minor latency to your execution thread.

To optimize this for extreme low-latency environments, you can transition from a polling model to an event-driven model: establish an integer variable _currentSessionTickets that resets at the start of the session, and increment it strictly inside the OnPositionOpened event handler instead of counting collections continuously.

[LondonScalper]

Hard-stop by time, always.... In the 15–30 minutes before the pre-market macro drops... the order book thins out.

Same religion. Setup quality becomes a story you tell yourself while the book is emptying.

I hard-fence new entries in that pre-NY runway — clock first, chart second. Spreads, voids, and slippage in that window will eat an A+ structure’s expectancy even if the level is “correct.”

If you’re running any semi-auto assist, a dead-zone time fence isn’t optional; it’s part of the edge definition. I’ve lost more to “just managing this one into the bell” than to being flat for twenty minutes. Flat is a position.

[PTScalper]

Hard-stop by time, always.... In the 15–30 minutes before the pre-market macro drops... the order book thins out.

Same religion. Setup quality becomes a story you tell yourself while the book is emptying.

I hard-fence new entries in that pre-NY runway — clock first, chart second. Spreads, voids, and slippage in that window will eat an A+ structure’s expectancy even if the level is “correct.”

If you’re running any semi-auto assist, a dead-zone time fence isn’t optional; it’s part of the edge definition. I’ve lost more to “just managing this one into the bell” than to being flat for twenty minutes. Flat is a position.

Hardcoding that time fence directly into the execution script is non-negotiable. If you leave it to manual discretion, the human brain will always rationalize taking "just one more clean scalp" right into the pre-open spread widening. Slippage doesn't care how textbook the level looked on your chart—once market makers pull quotes ahead of the bell, the mathematical expectancy of the setup is inverted before the fill even registers.

When your time fence trips ahead of that pre-NY runway, do you hard-flatten all running inventory mechanically, or do you just freeze new tickets and let an aggressive trailing stop manage whatever risk is already on?

[LondonScalper]

Hardcoding that time fence directly into the execution script is non-negotiable. If you leave it to manual discretion, the human brain will always rationalize taking "just one more clean scalp"

Yes. Written rules fail at the exact moment the book thins — that’s when the script has to win the argument.

I treat the pre-open dead zone like a news blackout: no new risk, full stop. Managing an already-open ticket into that window is a separate, rarer permission; inventing a fresh entry because the level looks textbook is not. Slippage and quote-pull don’t care about your confluence checklist.

If you’re semi-auto, hardcode the fence. If you’re fully discretionary, put a visible countdown on the secondary monitor. Either way, “one more” is usually the ticket that inverts the morning’s expectancy.

[PropScalpDesk]

Session structure gets discussed as clock times. The part that actually changes results for me is what I am allowed to do in each segment of the London-to-New-York window.

Permissions beat clocks. My Frankfurt map looks close to yours: early London is markup and A+ only; mid-London is playbook setups with one symbol bias; overlap is correlation-aware and optional if the morning already paid or already hurt.

PTScalper’s line is the one I underline: clock times without permissions are a pretty schedule. Hard max tickets per segment fixed overtrading more than motivation ever did. When the count is done, the segment is done — even if a late candle looks clever.

I also refuse to invent a second style because Phase A was quiet. Quiet is data. On funded risk I keep Phase A smaller still; trailing DD does not care that I “needed to participate.”

Pine/state talk is useful for automation, but discretionary desks still need the same state machine in writing: what is allowed now, what is watch-only, what is flat.

Do you write segment permissions on a card next to the screen, or keep them as soft habits?