All chapters·CHAPTER 07/08Advanced~14 min

The overnight session

While you sleep, Asia and London build the range your open inherits: inventory, gaps, ONH and ONL, and what each means at 08:30.

01A market that closes for one hour a day

Think of a diner that never shuts. At noon there are three cooks, a full floor and a line out the door; at 3am there is one cook, four customers and half the menu. The coffee sold at 3am is real coffee, paid for with real money. But nobody would poll those four customers and publish the result as the neighborhood's opinion. Futures run the same shift pattern. ES closes at 16:00 CT, reopens at 17:00 CT, and then trades twenty-three hours straight, yet the crowd behind the tape changes completely as the clock turns.

Learn the shifts by name. The 17:00 CT reopen is the thinnest tape of the night: settlement flows, position patching, a few algos re-hedging; a modest order can move ES a point here that would not move it a tick at 10:00. Through the evening the Asia hours take over, roughly 18:00 to 01:00 CT: Tokyo, Singapore and Hong Kong desks hedging regional books and reacting to Asian data, real business but shallow. At 02:00 CT comes the London hand-off: European cash equities and bunds open, macro desks arrive with actual size, and the overnight tape gets its adult supervision. From about 06:00 CT the pre-open builds, US desks filtering in, the 07:30 CT data releases, positioning into the 08:30 bell. Add it up and the whole overnight session often prints only 10 to 15 percent of the day's volume while occupying 65 percent of its clock.

That ratio is the key that unlocks the whole chapter. Chapter 1 taught you that acceptance is price plus time plus volume. Overnight, the time is abundant and the volume is missing, so every level the night builds is a claim signed by a small crowd. Real, but provisional. The map below runs one full clock, from the evening reopen through the close; note that its axis is stamped in New York time, 18:00 reopen and 09:30 open, while this chapter's desk numbers use Chicago time, 17:00 and 08:30. Same moments, one label apart. Flip through the four symbols and watch how much of each day's story is already drawn before the RTH divider.

LIVE
18:00 · ETH06:30 · read08:25 · brief09:30 · open16:00 · close
10:12ES: accepted above prior VAH 5,654, trend day watch
11:05ES: overnight gap fully filled at 5,648
KEY IDEA
Overnight prices are real; overnight acceptance is provisional. A level built by the night is a claim awaiting an RTH vote, because the voters who matter were asleep when it was made.

02ONH and ONL: the first frame of the day

Before the bell, every desk marks the same two lines: the overnight high, ONH, and the overnight low, ONL. Together they are the day's first frame, the range the open inherits. A wide overnight range, say 30 points on an ES whose 14-day ATR is 60, means the night already spent real energy: news got priced, stops got run, and RTH often opens into a market that has partly made up its mind. A narrow range, 8 or 10 points of drift, means the night settled nothing; the day session will have to do its own discovery, and the first hour tends to be where it starts.

Take one night whole before naming its parts. The strip below runs a single ES session end to end and colours each stretch with the shift that traded it, so the two extremes arrive with their authors attached, and the bell shows up only after fifteen and a half of the day's twenty-three hours are already spent.

one clock, three crowdsThe day is twenty-three hours long
ASIALONDONRTH5,6685,6605,6525,644ONH 5,668ONL 5,64008:30 BELLON MID 5,654the correction82%17:0021:0002:0005:0008:3012:0016:0015h 30m before the bellRTH: 6h 30m
ONH5,668 · Asia 21:40
ONL5,640 · London 03:40
ON range28 pts
Open in range82%
Inventorynet long, stretched
The first thing to read at 08:25 is where in that range the open sits: 82 percent of the way up, so the night crowd is long from lower and in profit. Inventory net long and stretched, which is why the first half hour so often trades back toward the 5,654 midpoint before the day decides anything of its own.
8/8
One night of ES, the 17:00 CT reopen through the bell, on the clock the rest of the chapter keeps referring to. The gauge on the right is the last thing the night hands over: where the open sits inside the range it built.

Here is the refinement most retail never makes: the two extremes are not equal citizens. Ask WHICH shift built each one. An ONL drilled at 04:00 CT, when London's macro desks are fully staffed and European cash is trading, was made by the deepest pool the night has; it carries real information about where size defended. An ONL printed at 19:00 CT on a sleepy Asia wobble was made by the thin shift and may be nothing but a stop run through an empty book. Same line on the chart, very different authors. The engine behind this course logs which session built each extreme for exactly this reason: the who matters as much as the where.

An overnight profile makes the shifts visible, and the one that runs with this section is the night the rest of the chapter keeps using: lowercase letters a through h are the Asia hours, capital L, M, N are London, and y, z are the pre-open. Read it bottom to top and the night tells its story: Asia grinds ES up from the 17:00 reopen and prints the ONH at 5,668 around 21:40, London arrives and sells the book down to the ONL at 5,640 at 03:40, and the pre-open climbs back to hand the 08:30 bell a price of 5,663.

CHECK YOURSELF
Two different nights print the exact same ONL price. Night one: the low is a two-minute wobble at 19:00 CT during Asia. Night two: the low is drilled at 04:00 CT during London and defended on two retests. Which ONL deserves more respect at 08:30?
5,668e← ONH, 21:40, Asia. One thin print.
5,664defz← z: the pre-open lifts back up here
5,660cdefgyz
5,656bcdfghNyz← the night's fattest row: Asia and pre-open agree
5,652abcghLNy
5,648abhLMN← 02:00 hand-off: London takes the book lower
5,644LMN
5,640M← ONL, 03:40, London. The deep shift made this one.
One overnight session, 17:00 CT to 08:30 CT. Lowercase a-h: Asia hours. Capital L, M, N: London. y, z: pre-open. The ONH belongs to the thin shift; the ONL belongs to the deep one, and that asymmetry is tradeable information.

03Overnight inventory: who carried what into the bell

Now ask the frame a second question: WHERE inside the overnight range does the open sit? That location tells you who is carrying position into the bell. If ES ground up all night and opens at 5,663 inside a 5,640 to 5,668 range, the crowd that traded the night is sitting on longs bought lower, almost by definition: the tape spent the night lifting, so the average overnight participant owns inventory from below and is in profit at the open. Desks compress this into one phrase: overnight inventory is net long. A night that ground lower and opens near its ONL is the mirror image, inventory net short.

Why does a day trader care what somebody else's blotter looks like? Because profitable inventory gets monetized at the bell. The 08:30 open hands the night crowd their first deep pool of liquidity in fifteen hours, and the rational move for a trader long from 5,648 with the market at 5,663 is to sell some into it. So the first 30 minutes often trade AGAINST the direction the night ground out, not because the night was wrong but because it is being cashed in. This is the inventory-correction open, and misreading it is one of the most expensive habits in day trading: the new short seller sees the first half hour go red, concludes the sellers have taken over, presses, and then watches the market resume the overnight direction at 09:15 once the inventory is clean.

KEY IDEA
The first 30 minutes often pay off the night shift. Expect the open to correct against the overnight grind when inventory is one-sided, and refuse to promote that correction into a day-direction call until it takes out real structure.
THE TRAP
The trap: treating a 03:00 London low as if it were RTH support. A trader marks the overnight ONL, watches RTH sell down to it at 10:40, and fades it at full size 'because the level held twice last night'. But the crowd that defended 5,640 at 03:00 went home at the hand-off; nobody standing there now ever agreed to buy it. Until RTH itself prints time and volume at the level, an overnight low is a hypothesis. Trade it at hypothesis size, and only after the day session shows a response there, or you will keep funding the traders who waited for confirmation.

04Gaps: how loud was the night?

A gap is the night's summary statement, and the first job is to define it honestly. On a 24-hour chart ES almost never gaps, because the overnight walked price there tick by tick. The gap that matters to a profile trader is the TRUE gap: today's open measured against yesterday's RTH close. It answers the only question worth asking at 08:30: how far did the night move the market while the day-timeframe crowd was away? And there is a stronger grade above it. An open beyond yesterday's entire RTH RANGE, not just its close, is the loudest statement an overnight can make: every single price the day crowd agreed on yesterday has been rejected before they even sat down.

Whether a gap fills is not folklore, it is a measurable function of two variables. First, size in ATR terms: small gaps are usually just overnight drift with no conviction behind them and fill at high rates; gaps that are a large fraction of an ATR were driven by real news or real repositioning and fill far less often on the same day. Second, inventory: a gap up that the overnight crowd ground out themselves is carried by profitable longs who will sell the open, helping the fill; a gap up that ran AGAINST a net-short overnight crowd is a squeeze, and squeezed shorts buying their way out at the bell push price away from the fill. The product measures fill rates per gap-size bucket and inventory state per symbol, because 'gaps usually fill' is exactly the kind of sentence that costs money when left unquantified.

WORKED EXAMPLE · Two gaps, two statements
Yesterday RTH range5,632 - 5,662
Yesterday RTH close5,655
14-day ATR60 pts
Night A opens5,663
Night A true gap5,663 - 5,655 = +8 (13% ATR)
Night B opens5,690
Night B true gap5,690 - 5,655 = +35 (58% ATR)
Night A cleared yesterday's close by 8 points but sits just 1 point above yesterday's high of 5,662: a whisper, 13 percent of an ATR, the kind of gap that fills before lunch more often than not. Night B opened 28 points beyond yesterday's entire range, 58 percent of an ATR: the overnight has rejected every price the day crowd printed yesterday. Same direction, both 'gap ups' in a headline, and they belong to different statistical universes. Bucket B's play is to look for acceptance and go with; treating it like bucket A and shorting for the fill is how a gap trader has one great month and then one terminal week.

05One full night in numbers

Now run the whole framework over the night from the profile in section 02, the way a desk would at 08:25. The facts first, then the read.

CHECK YOURSELF
A different night: ES grinds DOWN from 17:00 onward, London drills the ONL of 5,610 at 03:15, and the 08:30 open prints 5,612 inside an overnight range of 5,610 to 5,638. Yesterday's RTH close was 5,636. Name the inventory, and the open's first risk.
WORKED EXAMPLE · ES, the 08:25 read
Overnight range5,640 - 5,668 (28 pts)
ONH built byAsia, 21:40
ONL built byLondon, 03:40
Yesterday RTH close5,655
Open (08:30 CT)5,663
Open location in ON range(5,663-5,640)/28 ≈ 82%
True gap+8 = 13% ATR
Inventorynet long, and stretched
The read: the pre-open climbed 23 points off London's 5,640 low and the open sits in the top fifth of the range, so overnight inventory is not just long, it is stretched long. Expect the correction: night longs selling the bell can plausibly push 8 points to the 5,655 close, filling the small gap, and the ON midpoint at 5,654 sits right there as a companion magnet. The map for the first hour: ONH 5,668 overhead, gap fill 5,655 and midpoint 5,654 below, London's ONL 5,640 as the floor that actually meant something. What invalidates the fade: RTH acceptance ABOVE 5,668, letters building and volume printing above the ONH, which says fresh day-timeframe buyers took over from the night crowd instead of paying them out. And keep the trade its actual size: this is an 8-to-14-point inventory correction thesis, not a short-the-day thesis. The night ground UP; only structure breaking, not the open's reflex, is allowed to argue the day goes down.

06A 3am alert with a name tag

Everything in this chapter assumes somebody was awake to see it, and for most traders nobody was. Chapter 5 left you with a list of magnets the market comes back for, naked POCs above all, and the market does not check your time zone before visiting one. A naked POC from last Tuesday getting its first touch at 03:00 CT is a real event at a real level; the only overnight-specific question is who touched it. That is why the alert engine watches all three overnight windows and stamps every overnight alert with its session: a [London] tag on that naked-POC test tells you the deep shift did it and the level's response is worth grading, while an [Asia] tag on the same line warns you the witness pool was thin. The event is never anonymous, so your 08:25 read starts with the night already labeled.

IN THE PRODUCTAlerts fire around the overnight clock and each one carries its session stamp, [Asia] or [London], next to the level it touched, so the night you slept through reads like a labeled log instead of a mystery.
See session-tagged alerts

07Next: the player bigger than the day

You can now read the whole clock: the RTH auction from chapters 1 through 6, and the fifteen hours that frame it. One question remains, and it is the one that separates a signal reader from a trader. The overnight crowd cashes out in half an hour; day-timeframe traders are flat by the close. So whose orders keep value migrating higher for two straight weeks? Chapter 8 introduces the longer-timeframe player, teaches you to find their footprints on the profiles you can already read, and then folds all eight chapters into a morning routine that fits between 08:00 and the bell.