A session runs in four phases: preparation before the open, observation through the first half hour, execution when price reaches a marked level with confirmation, and review after the close. Most of the day is spent in the second phase, and most days produce two or three trades at most.
This article puts everything else in the library into sequence. If the individual pieces have made sense but you are unsure how they combine into an actual day, this is that.
Pre-market: about twenty minutes
I am not looking for a trade here. I am building a map and forming an expectation.
Mark the levels
Yesterday's POC, VAH and VAL. The composite POC from the last two to three weeks. Any thin areas from the previous session. Options levels if I am using them, adjusted for the ES basis as described in options flow for futures traders.
Six or seven lines total. If I have twenty lines on the chart, I have not decided anything, I have just drawn everything.
Read the overnight
Was it balanced or directional? Where is price now relative to yesterday's value area? Those two answers set the expectation.
Write one sentence
Before the open, in a notebook: what I expect and what would change my mind. Something like "opening inside yesterday's value, expecting rotation between VAH and VAL, wrong if we accept outside either edge".
This is the highest value twenty seconds of my day. An expectation written down before the open can be checked. One held in your head at eleven becomes whatever the last twenty minutes made you feel.
The open: watch, do not trade
The first ten to fifteen minutes are the noisiest part of the session. Ranges are wide, direction reverses repeatedly, and today's developing profile has almost no data in it.
I do not trade this window. Not as a rule handed down from somewhere, but because the confirmation model cannot run properly yet: today's levels are not formed, the state is not established, and price has not tested anything.
What I am doing instead is answering the state question from balance vs imbalance. Are we rotating or going somewhere? Is the open holding inside yesterday's value or outside it? Is delta oscillating or trending?
By around thirty minutes in there is usually enough to call it.
The body of the session: waiting properly
This is most of the day, and doing it well is the actual skill.
Price is somewhere between my levels. Nothing is required of me. The job is to keep the state read current and be ready when price reaches somewhere I care about.
When it does, the confirmation model runs, and it is the same four questions every time, set out in order flow confirmation models:
- Does this level have a reason to exist, and does the state support the direction?
- How did price arrive, slowly or aggressively?
- Is anyone defending it, meaning absorption, exhaustion or a delta shift?
- Where am I wrong, and can I afford that distance?
Any one failing ends it. Most of the time something fails, which is why most approaches to most levels do not become trades.
A full ES session chart marked up as it would be pre-market: yesterday's POC, VAH and VAL extended, composite POC, one thin area shaded. Then annotate the two or three points during the day where price reached a level, and mark which produced a trade and which did not.
Managing a position
Once in, three things and no more.
The stop stays where the reasoning breaks. Not moved wider because price is approaching it. If the invalidation was correct at entry it is still correct now, and the only thing that has changed is that being wrong has become uncomfortable.
Take partials at structure. The POC for a rotation trade, the near edge of a thin area for a breakout. Profile levels give clean targets, so there is little point inventing an arbitrary ratio.
Exit early if the premise dies. This is the one people resist. If I am long from a value area low on absorption, and price returns and that absorption is gone, the reason for the trade no longer exists. I do not need to wait for the stop to confirm what I can already see. The stop is for when I am not watching or when it moves too fast.
The afternoon
Different character, and worth treating differently.
Lunchtime through early afternoon is usually the thinnest, most directionless part of the session. Levels matter less because participation is low, and footprints get noisy because small orders look dramatic against light volume.
Then the last hour. Positioning gets adjusted, volume returns, and if there is heavy same-day options gamma the market may either pin to a strike or accelerate away from one, which is covered in 0DTE options and intraday pinning.
I trade the last hour less than I used to. Moves are real but they are frequently positioning driven rather than auction driven, and my framework is built for the auction.
After the close: fifteen minutes
Non-negotiable and easy to skip.
Compare the day to the expectation. Pull out the sentence written before the open. Was it right? If not, what was the first piece of evidence that it was wrong, and how long did I take to accept it? That lag is the number that matters most, and it is the one that shrinks with practice.
Review every trade against the four checks. Not by whether it made money. A losing trade where all four passed is a good trade. A winning trade where I skipped step three is a warning, and it is the more dangerous of the two because the profit hides it.
Note the levels for tomorrow. Today's completed profile becomes tomorrow's reference, which means the review and the preparation are the same task.
What a normal day actually looks like
Two or three trades. Sometimes one. Reasonably often none.
I mention it because the gap between that and what people expect is where most of the damage happens. A trader who believes a good day means six trades will manufacture four of them, and manufactured trades are how accounts go backwards on days that offered nothing.
Days with no setup are not failed days. They are days the market did not offer anything that passed the checks, and the correct response is to have taken nothing. That is a result, not an absence of one.
Building up to this
Do not start with all of it. The sequence I would suggest:
- Levels and the state call only. No trading. Two weeks.
- Add the confirmation model, in replay rather than live. Two weeks.
- One setup live, smallest size, one trade a day maximum. A month.
- Add the review habit properly from the first live trade, not later.
- Add options positioning once the rest is boring.
Roughly two months before trading live in any size, which sounds slow and is considerably faster than the alternative of learning it with money while under pressure.
Frequently asked questions
How many trades should I take per day with order flow?
Two or three at most on a typical day, sometimes one, reasonably often none. If a process requiring context, arrival, confirmation and a defined invalidation is producing many trades a day, the criteria are too loose to be filtering anything.
Should I trade the market open?
I do not trade the first ten to fifteen minutes. The developing profile has almost no data, the day's state is not established, and price has not tested anything, so the confirmation checks cannot run properly. I use that window to read the state instead.
When should I exit a trade early?
When the reason for the trade stops being true. If you are long from absorption at a level and price returns to find the absorption gone, the premise is dead regardless of where your stop sits. The stop exists for when you are not watching or when things move too quickly to react.
Is the last hour worth trading?
It has real volume and real moves, but they are often positioning driven rather than auction driven, particularly when heavy same-day options gamma is in play. If your framework is built on the auction, be aware you may be trading a different mechanism in that window.
What should I review after the session?
Compare the day to the expectation you wrote before the open, and review each trade against your checks rather than against its outcome. A losing trade where every check passed is fine. A winning trade where you skipped a check is the more dangerous one, because the profit conceals it.
How long before I can trade this live?
Roughly two months of structured practice: two weeks marking levels and calling the state without trading, two weeks running the confirmation model in replay, then a month live at minimum size with a one trade per day cap. Slower than most people want and faster than learning it under financial pressure.