Applied Market Mechanics: Setup 01 · Spring and Reclaim

Class	Setup
Entry	01
Markets	Crypto, equities, ETFs, commodity futures
Timeframes	1D to 3M
Builds on	UMM Part 4, Daily vs Weekly Market Closes, Investing vs. Trading
Tools used	Tool 01 · Fibonacci Ranges

The first time this Setup appeared in public, it didn't have a name. On 14 January 2021 I posted a three-day chart of COINBASE:BTCUSD with five orange horizontal lines drawn by hand and a set of Wyckoff labels pencilled underneath: a spring in March 2020, two last points of support, a sign of strength in December. The caption said the markup wasn't over, and that the pullbacks would be deep. Nine days later I named a target, , "at the 4.618", and on 17 February I posted that it had been a long, windy road to get there.

Those lines look improvised. They weren't. Two candles reproduce every level I posted that year, to the precision I posted it: the daily close of 31 January 2019 at and the high of 26 June 2019 at . Run them through the formula in Tool 01 and lands at (I wrote "24.3"), at (I wrote "45k"), and at , away from the posted target. The arithmetic was already there. The rules weren't: the 1-anchor sits on a wick, which Tool 01 now forbids, and nothing in those posts says what would have proved the read wrong.

The same winter holds the other kind of page. On 20 January I wrote that my previous prediction had been far too optimistic and clearly hadn't aged well. In June I posted conditions for by the fourth quarter, a target that sat somewhere past on the same ladder. Bitcoin topped that November at . Both posts are still public, and both stay in this entry, because the method was still baking in the oven and a lab book that only keeps the runs that worked isn't one.

Five years later the method has rules, and this entry writes them down: the conditions that put a thesis in play, the close that triggers it, the close that ends it, the rung that pays it and the time it's allowed to take. Then it tests them against five public cases on five markets and four timeframes: four that paid and one that didn't, and the one that didn't changed the rules before they were published. One thing has to be said before any of that. These rules were written in October 2026, after every one of those cases had played out. They're a description of how I read, made precise, and the cases are evidence that the reading works. They can't prove the rules, because the rules were cut to fit a record I'd already seen. What will test them is the next case, written down before it resolves.

#Thesis

Spring and Reclaim. A range that has been tested at its floor and then closes back above its ceiling is a range that has finished accumulating, and it tends to resolve upward to the first rung of the expansion ladder, .

#Mechanism

Every range in Tool 01 is a record of a negotiation. The climax anchor marks where one side ran out of conviction, the reaction anchor marks how far the other side could push before it stopped, and everything in between is price being argued over. Part 4 of Understanding Market Mechanics described who does the arguing: a composite operator building a position quietly, inside a range, at prices the rest of the market has stopped caring about. Tool 01 maps an upside continuation draw onto Wyckoff re-accumulation, and the events Wyckoff named inside an accumulation range are the events this Setup watches for.

The first is the spring. Late in the range, price is pushed down to the floor or through it, far enough to trip the stops that sit under every obvious low and to shake out whoever was still holding on hope. If the floor gives way for real, the range was distribution after all and the read is over. If it doesn't, the selling that was supposed to follow never arrives, and price closes back inside. That failure to continue is the information. Supply was offered at the worst possible moment for buyers and it wasn't enough.

The second is the sign of strength: price closing above the range's ceiling, the level that capped every rally while the position was being built. A close there says demand has stopped being absorbed. What usually follows is a pullback that holds above the old ceiling, the last point of support, and then the markup.

The name of the first event suits it better than Wyckoff could have known. A spring pressed towards its floor stores energy in proportion to how far it's compressed, and the range works the same way. Wyckoff's law of cause and effect says the move out of a range is proportional to the work done inside it. This entry takes that law literally in one place only, the time limit: a range is given as many bars to deliver as it took to build.

The same law sets the horizon. A range anchored on the daily is built in weeks and pays, if it pays, in weeks. A range anchored on the monthly took years to form and asks for months or years of patience after the trigger. That's the line between a trade and a holding, drawn by the timeframe of the anchors rather than by intent, and it's why this Setup's cases run from 43 days to 19 quarters. The mechanism is the same at every scale. What changes is how long you're prepared to sit with it, which Investing vs. Trading covers in more depth than belongs here.

Wyckoff's vocabulary stays in this section. Everything from here to the reference card is written in Tool 01's terms, so the Setup can be run by someone who has never read a word of Wyckoff, and it uses no event that Tool 01 can't check.

#Conditions

All of these must hold before the trigger counts. Every item is checked on the anchor timeframe, on body closes, unless it says otherwise.

• An upside continuation draw. A Tool 01 range with Reverse off and the 1-anchor above the 0-anchor, both anchors on candle bodies, written in full (ticker · timeframe · candle open time (UTC) · body edge) before anything else is read.
• The scale is recorded. Linear or log, written next to the anchors. Every level below is read on that scale and no other.
• A spring. After both anchors are set, price reaches the band from to . Wicks may go below . A body close below is allowed only if the very next bar closes back above it.
• The range stands. No body close below since the anchors were set, and no run of consecutive closes below longer than the range took to form. Failing either means the range was wrong: strike it through and redraw. The second test was added after GMG failed; see When It Fails.

#Trigger and Invalidation

These are stated before any case is shown, so every case below is read against the same rules.

	Rule
Trigger	The first body close above after the spring.
Target	A body close at or above .
Invalidation	A body close below after the trigger.
Time limit	As many bars after the trigger as there are between the two anchors.

A case resolves at whichever of the target, the invalidation or the time limit comes first. That's its resolution date, and nothing that happens afterwards changes it. A case that reaches its target and then falls back into the range was a success, and keeping the gain was a question of expression. A case that closes below and later rallies through its target was a failure. The rule that stops a retrace from counting against a success is the same rule that stops a late recovery from rescuing a failure.

A close back below after the trigger doesn't end the case. It's the reclaim taking longer than it looked, and two of the cases below did exactly that for months.

#Case Studies

Each case is labelled with the grade of its evidence.

• Capture: a public post with an independent timestamp, made before the outcome.
• Record: a private but dated record made before the outcome. Weaker, because it can't be checked by anyone else.
• Reconstruction: a reading drawn after the fact. It can illustrate the Setup but never counts as evidence that the Setup saw anything coming.

Four of the five cases in this entry are Captures, GMG's in When It Fails among them. The Bitcoin case is a Reconstruction: its draw is public and timestamped, but every post showing it came after the case had resolved, so it shows how the Setup reads and proves nothing about foresight. It comes first because it's where the method started. Figures are rendered from the anchors exactly as they were drawn in the capture, mistakes included, and where a capture differs from what Tool 01 would draw today, the case says so. Outcome figures mark the trigger and the resolution and shade everything after it, so what came next stays visible without counting. They run to the end of each snapshot, except Bitcoin's, whose daily candles stop at the end of February 2021; years of dailies would be unreadable, and the prose carries the rest.

#Case 1: COINBASE:BTCUSD, 2020–21

Reconstruction. The draw appears in posts of 14 January 2021, 23 January 2021 and 17 February 2021.

Draw. COINBASE:BTCUSD · 1D · 2019-01-31 00:00 · body (close), as the 0-anchor; COINBASE:BTCUSD · 1D · 2019-06-26 00:00 · wick (high), as the 1-anchor. Linear. The 1-anchor is a wick; under Tool 01 it would be the body's upper edge at . The case uses the wick, because that's what was drawn.

Conditions as read then. The March 2020 crash reached , or , inside the spring band without a single daily close below ; the lowest body was , at . The range stood throughout.

Outcome. The trigger was the daily close of 3 November 2020 at . The deepest pullback came the next day, to (), and there was no close back below . The target, , closed on 16 December 2020 at , 43 bars after the trigger against a limit of 146. Resolved: target. The January posts came after resolution, while price was already well up the ladder. They're the public record of the draw, which is what makes the reconstruction checkable.

#Case 2: AMEX:NLR, 2020–23

Capture. Posts of 17 May 2023, with follow-ups on 24 October and 22 November 2023 in the same thread.

Draw. AMEX:NLR · 1M · 2012-06-01 · body (open), as the 0-anchor; AMEX:NLR · 1M · 2017-12-01 · body (open), as the 1-anchor. Log. The May 2023 post showed the chart with no levels drawn; the range first appears in the October follow-up, on a log chart, and that's the scale used here. The November follow-up shows the same anchors on linear, where falls at instead of . Both were reached.

Conditions as read then. March 2020 wicked to , at , and closed the month below at . April closed back inside at , which is exactly the one-bar recovery the spring allows. The range stood: sat at .

Outcome. The trigger was the close of May 2021 at , barely above . What followed was two years of argument at the ceiling: June 2021 closed back below it, and the monthly closes crossed in both directions until October 2022, when price pulled back to (). From November 2022 every close held above. The May 2023 post came in the middle of that last stretch. The target, , closed in September 2023 at , 28 bars after the trigger against a limit of 66. Resolved: target.

#Case 3: NYSE:BABA, 2022–25

Capture. Posts of 10 January 2024 and 27 September 2024.

Draw. NYSE:BABA · 1M · 2015-10-01 · body (open), as the 0-anchor; NYSE:BABA · 1M · 2014-11-01 · body (close), as the 1-anchor. Log. The captures show the range on 3D and weekly charts; the anchors are monthly bodies, so the anchor timeframe is the monthly, read down onto lower charts as Tool 01 allows.

Conditions as read then. October 2022 wicked to , at , and closed the same month back inside at : the cleanest spring of the five. January 2023 is worth a look on the way past. It traded up to , above , and closed at , below it. That's no trigger, and February fell to .

Outcome. Both posts came before the trigger, inside the range; they're the early entry the guidance below allows, and the Setup itself wasn't live yet. The trigger was the close of February 2025 at . April dropped to (), back below the ceiling, without a close near the floor. The target, , closed in September 2025, 7 bars after the trigger against a limit of 11. Resolved: target. Price went on to the following month and was back below by June 2026. That retrace is after resolution and doesn't change the case; it's the reason the expression section exists.

#Case 4: COMEX:HG1!, 2020–25

Capture. Post of 26 June 2024.

Draw. CAPITALCOM:COPPER · 3M · 2015-07-01 · body (close), drawn at as the 0-anchor; CAPITALCOM:COPPER · 3M · 2008-04-01 · body (close), as the 1-anchor. Log. Two corrections, both stated rather than fixed. The intended 0-anchor was the quarter's close at , and the line was placed slightly low; the case uses as drawn. Every resolution date below is identical on either value, and moves from to . The capture is a broker's CFD feed, whose prices differ from the exchange's, so the figures use COMEX copper futures, where the same two quarters close at and , within of the capture.

Conditions as read then. Copper spent four quarters closing below after the 0-anchor, from late 2015 to mid-2016, with a low of (). That was the range still forming. It never came near at , so under Tool 01 it was a test, and the Setup wasn't looking for a spring until both anchors had been tested and held. The spring came in the first quarter of 2020: a wick to and a close below at , recovered the next quarter.

Outcome. The trigger was the close of the first quarter of 2021 at . Mid-2022 closed back below and pulled back to (). The June 2024 post came after the first touch of that May. The target closed in the fourth quarter of 2025, 19 bars after the trigger against a limit of 29. Resolved: target.

#The Record

Case	Anchor TF	Scale	Spring (deepest)	Trigger	Deepest pullback	Target closed	Bars / limit
BTCUSD	1D	Linear		3 Nov 2020		16 Dec 2020	43 / 146
NLR	1M	Log		May 2021		Sep 2023	28 / 66
BABA	1M	Log		Feb 2025		Sep 2025	7 / 11
HG1!	3M	Log		Q1 2021		Q4 2025	19 / 29
GMG	1W	Linear		12 Jan 2026		missed: time limit	11 / 11

Two things in the first four rows are tempting, and neither is a rule. Three of the four pulled back hard after the trigger, and two of those three stopped within a few hundredths of ; BABA held at . And all four reached the target. Cases chosen by the person who wrote the rules can't carry either finding. The first goes into the next Setups as something to watch, and the second waits for a sample nobody picked.

#When It Fails

#TSXV:GMG, 2023–26

Capture. Posts of 8 May 2023 and 13 June 2023.

Draw. TSXV:GMG · 1W · 2021-05-17 · body (close), as the 1-anchor; TSXV:GMG · 1W · 2021-08-02 · body (open), as the 0-anchor. Linear, as in the June capture. The anchors are 11 weeks apart, which is the time limit.

Conditions as read then. The spring came in the first week of April 2023: a wick to () and a close back inside at . In June price climbed to , five cents under , and the June post drew the range that week. There was no close above the ceiling, so there was no trigger.

Outcome. August 2023 closed below the floor and stayed there. Over the next two and a half years GMG made 123 weekly closes below , down to a low of , or . It never closed past at , so under Tool 01 the range still stood, and under the rules as first written the 2023 spring was still the spring. Price climbed back through the range in December 2025, and the close of 12 January 2026 at () was the trigger. The next week traded to , three cents short of the target at , and closed below it. Then it fell back into the range, never closed at the target and never closed below the floor. The 11 weeks ran out with the week of 30 March 2026. Resolved: time limit.

Why it failed. Three things, in order of weight.

1. The range had stopped being a range. The test that armed the trigger happened in April 2023. Everything after it said the floor had given way: 123 consecutive weekly closes beneath it, eleven times as long as the range took to build, most of the way to . What triggered in January 2026 was a recovery from that collapse, read off a range that had stopped describing the market. The rules as first written asked how far price fell below the floor and never how long it stayed there.
2. The trigger closed far above the ceiling. At the target was away and the floor away. Even had it paid, it would have paid poorly for the risk.
3. A small cause gets a short clock. The range took 11 weeks to build, so it had 11 weeks to deliver. It came within three cents in the first of them and spent the other ten falling back.

What it changed. The second test under the range stands exists because of GMG. Tool 01 judges a range's failure by distance, a close past . This adds time, on the same principle as the time limit: a range is allowed as many consecutive closes below its floor as it took to build, and one more strikes it through. Under that rule GMG's range fails in the week of 23 October 2023, its twelfth consecutive close below , and the 2026 trigger never happens. It changes none of the other four cases; the longest run below the floor among them is copper's four quarters, against a limit of 29. A first attempt, requiring the spring to be the latest test of the floor, didn't survive checking: December 2025 wicked to and closed exactly on , which is a fresh spring by this Setup's own definition, so that rule would have let the trigger stand. GMG stays scored as what it was under the rules it met: a failure, kept in ink, with the rule that came out of it written underneath.

#The Targets Nobody Should Have Set

The June 2021 post asked for by the fourth quarter, past on a range whose markup had already run to nearly . It missed by more than two-thirds. Ethereum got the same treatment twice: thirty to forty thousand within twelve months, in June 2021, and a drawn road to in June 2022. Its highest price since has been , in August 2025. None of those reads had a target rule, and each reached for a rung the range couldn't justify. This Setup's conservative target exists because of posts like those.

#The Wick that Wasn't a Trigger

BABA in January 2023 traded through and closed below it. On a lower timeframe it looked like a breakout. On the anchor timeframe it was a wick, and February fell more than .

#The Stop that Would Have Been Too Tight

NLR closed back below repeatedly between June 2021 and October 2022, and copper did it once in 2022. An invalidation at would have stopped both out, NLR several times over, before either reached its target. That's why invalidation sits at the floor.

#The Move with no Range

In November 2023 I put NYSE:SMR on a watchlist with a note to wait for a base. No range was drawn, so there was nothing for this Setup to test, and SMR ran from to without one. The Setup only speaks where a range exists. It will miss moves like that, and missing them is part of the price of having rules at all.

#Expression and Risk

Everything above is the experiment. This section is guidance, and it's written in terms of what you can do rather than what you should. How you act on a triggered Setup depends on things no rule can see: how well you know the market, how much of its behaviour you'd recognise without a chart, and how much confidence you're bringing to it. Position sizing and the structure of risk are covered in Trading Big Bags and A Small Edge, and they aren't repeated here.

The arithmetic at the trigger. Measured in RL units, an entry at with invalidation at and a target at risks to make . The breakeven hit rate for that trade is , which is again. The ratio keeps turning up. Four successes and one failure are no estimate of a hit rate. A trigger close well above the ceiling makes the arithmetic worse: BABA's trigger closed at and GMG's at .

You can enter earlier. An entry in the spring band, before the trigger, carries far less distance to the invalidation and far more to the target. It also carries none of the confirmation: the range hasn't proven anything yet. The BABA posts are this entry, made inside the range more than a year before the trigger.

You can re-enter after a shakeout. A body close below after the trigger ends the case, and closing the position there is the conservative response. If the next bar closes back inside, that's a spring by this Setup's own definition, and the thesis can be back in play. Re-entering then is a new case, scored on its own, and the first one stays recorded as invalidated.

The timeframe sets the horizon. A daily range asks for weeks; a monthly or quarterly range asks for months or years, through stretches like NLR's two years at the ceiling. Decide which you're doing before the trigger, because the two call for different responses to the same pullback.

The target ends the case. The market carries on. Bitcoin ran to after its target, NLR to . BABA reached and fell back into its range. What you do above is outside this Setup, and Tool 01's ladder is the map for it.

#Reference Card

• Thesis: a range tested at its floor that closes back above its ceiling tends to resolve to .
• Uses: Tool 01, upside continuation draw (Reverse off), anchors on candle bodies.
• Write: ticker · timeframe · candle open time (UTC) · body edge for both anchors, and the scale, before reading.
• Spring: after both anchors, price reaches to ; a body close below counts only if the next bar closes back above it.
• Range stands: no body close below , and no run of closes below longer than the range took to form; otherwise strike through and redraw.
• Trigger: the first body close above after the spring.
• Target: a body close at or above .
• Invalidation: a body close below after the trigger.
• Time limit: as many bars after the trigger as there are between the anchors.
• Resolution: whichever comes first; nothing after it changes the case.
• Closes: body closes on the anchor timeframe only.
• Horizon: set by the anchor timeframe; daily ranges pay in weeks, monthly and quarterly ranges in months or years.

---

Applied Market Mechanics is a record of how I read markets, published purely as education. Nothing in it is financial advice or a recommendation to buy or sell anything, and I'm not a financial adviser. Every example has already resolved. I don't disclose positions, so nothing in an entry should be read as a statement of what I hold. Markets take money from people who follow a method exactly as often as from people who don't; size accordingly.