Order Blocks and Displacement

An order block is the last opposite-coloured candle before a strong move away from it. That sentence contains three settings and no observations: how strong the move has to be, how soon it has to come, and whether the market has to confirm it. On the twenty bars of lesson 32 — the same prices, now with bodies — ordinary choices give ten order blocks, or five, or four, or two, and two readers who each end up with about five are marking different bars.

Prerequisites: Lesson 32, because this lesson borrows its twenty bars and inherits its structure rules, and lesson 31, where displayed size stopped being evidence of how much anyone was doing.

The same twenty bars, now with bodies

Lesson 32 gave the highs and lows of a twenty-bar series. Here are the opens and closes for the same bars, so that the candles have colour. The opens are 100.1, 100.8, 101.0, 101.6, 102.3, 102.9, 103.4, 102.9, 103.9, 104.6, 105.1, 106.0, 104.2, 104.0, 103.5, 103.1, 102.1, 101.5, 100.8 and 100.2. The closes are 100.7, 101.5, 100.4, 102.8, 101.5, 104.1, 102.6, 102.1, 105.3, 103.7, 106.7, 104.6, 103.0, 105.7, 102.2, 104.9, 100.8, 103.2, 99.4 and 101.6. Every open and close sits inside its own bar, which you can check against lesson 32. The twenty bars average 1.455 points from high to low.

Now run the rule four times, changing only the distance the move has to cover.

Distance required	Order blocks	Bullish, at bars	Bearish, at bars
1.0 average ranges (1.46 points)	10	3, 5, 8, 10, 13, 15	11, 14, 16, 18
1.5 average ranges (2.18 points)	5	3, 8	14, 16, 18
2.0 average ranges (2.91 points)	4	3, 8	14, 16
2.5 average ranges (3.64 points)	2	8	16

Ten order blocks on twenty bars is one every other bar. Two is one every ten. Both counts come off the same prices, and neither multiplier is strange: one average range is what a routine bar does, and two and a half is a move you would look up at. The gap between those two settings is the gap between a chart covered in zones and a chart with two on it.

Watch the boundary as well. At one average range, bar 15 qualifies as a bullish order block because price reaches 105.1 within three bars where the rule wanted 105.055. It qualifies by forty-five thousandths of a point, about three per cent of the distance the rule asked for. That is not a flaw in this particular rule — every threshold has a boundary and some bars sit on it — but it is worth seeing once, because a zone that exists by three per cent of a number you chose is indistinguishable, on the chart, from one that cleared it by a mile.

Four bars marked in both directions

The count is the mild version. The sharper question is what the loose setting does during the decline, and the answer is that it marks both directions at once.

Bar	Close	Marked as	Because price then
13	Lower	Bullish order block	rose to 105.9, 1.6 above the bar’s high
14	Higher	Bearish order block	fell to 100.6, 3.3 below the bar’s low
15	Lower	Bullish order block	rose to 105.1, 1.5 above the bar’s high
16	Higher	Bearish order block	fell to 99.2, 3.8 below the bar’s low

Four consecutive bars, in the middle of a decline any reader would call a decline, and the rule has marked two zones that say buyers stepped in here and two that say sellers did, alternating. Whichever way price goes next, one of them was there first. That is what happens to a pattern whose definition looks forward from a candle and backward from a move at the same time: the candle is chosen by what followed it, so something is always in the right place afterwards.

None of which makes order blocks worthless. It sizes what the loose setting costs. At two and a half average ranges the same twenty bars give two zones, one on each side, and they sit where a reader would have put them unaided. The zones did not get better. The definition stopped marking everything.

What displacement adds, and where it sits in time

Displacement is the part of the definition the market has to supply. You choose the candle, the colour is arithmetic and the backwards search is mechanical, but the move afterwards either happened or it did not, and no amount of drawing makes it happen. That is a real distinction and it is the same one lesson 32 drew for break of structure: this concept can be wrong, which already puts it ahead of the triangle.

The catch is where in time it sits. The move that qualifies a candle as an order block happens after the candle, so you cannot mark one in advance by construction, and by the time you can mark it the move it was named for is over. Everything the concept offers forward is the claim that price will react when it comes back. This lesson has not tested that claim. It has counted how many places you would be standing while you waited.

The imbalance the move is supposed to leave

The story usually adds that displacement leaves fair value gaps behind: price moves so fast that the bar before a candle and the bar after it do not overlap at all, leaving a band of prices the middle bar passed through without either neighbour trading in it, and that band is unfinished business. That is a definition you can run too, and it is the cleanest one in the family because it has no free parameter at all. On these twenty bars there are four such gaps: three bullish, at bars 4, 8 and 9, and one bearish, at bar 12.

Set the threshold at one and a half average ranges and the rule marks five order blocks. Two of them are followed by a gap within three bars. The other three displaced by more than two points each and left no gap at all, because the bars overlapped on the way. So the imbalance is a real thing that sometimes accompanies a fast move; it is not what a fast move is, and a definition that treats the two as the same object is counting one event twice.

The fourth setting, inherited

The usual remedy for a chart covered in zones is a filter: only take order blocks whose move also breaks structure. It works, and it costs a fourth setting, because break of structure is lesson 32’s rule and arrives carrying its own swing lookback.

Apply it both ways. With the sensitive one-bar swing rule and a threshold of one and a half average ranges, five order blocks survive: bars 3, 8, 14, 16 and 18. With the two-bar swing rule and a threshold of one average range, four survive: bars 8, 14, 15 and 16. Two readers, both with ordinary settings, both ending up with about five zones on twenty bars, agreeing on three of them and disagreeing on the rest. And with the three-bar swing rule nothing survives at all, because lesson 32 already showed that rule finds no structure events in twenty bars with which to confirm anything.

Four settings in series, then, one more than the sentence appeared to contain: the body rule that decides colour, the distance, the window, and the swing lookback underneath the confirmation. The number of order blocks on a chart is not a fact about the market. It is a fact about those four numbers, and the usual presentation names none of them.

Problems

1. Find out how many your own settings produce. Take sixty consecutive bars of your instrument and compute the average bar range over them. Then mark every bullish and bearish order block at a threshold of one average range within three bars, and mark them again at two and a half average ranges within three bars. Write both counts down. The ratio between them is what the word strong is worth on your chart, and most people have never seen it for their own.
2. Count the bars that get marked both ways. On the same sixty bars, at the looser of your two settings, write down for every marked bar which direction it was marked in. Then count the bars that carry a mark in one direction with a mark in the other direction within two bars either side. On the twenty above that count was four in a row. Whatever yours is, it is the number of places on your chart where the pattern was not able to be wrong.
3. Test the return, with the definition fixed first. Write down your threshold, your window and your swing lookback before looking at anything. Then take thirty consecutive order blocks that meet all three, and record two things for each: whether price returned to the zone at all, and, if it did, whether it left the zone in the predicted direction before it left in the other. Two counts, not one — a zone price never revisits is not a win, and counting it as one is the most common way this measurement goes wrong. How to Collect a Base Rate is how the count is kept honest.

Sources. Albert S. Kyle, “Continuous Auctions and Insider Trading” (Econometrica, 1985), for the result this lesson leans on twice: the informed trader’s optimal strategy is to spread the position over time so that the order flow gives away as little as possible, which is the opposite of leaving one identifiable candle behind. Robert Almgren and Neil Chriss, “Optimal Execution of Portfolio Transactions” (Journal of Risk, 2000), for what a large order looks like when it is actually executed: a schedule spread across a session that trades price impact against timing risk, rather than a decision taken at a level. Ryan Sullivan, Allan Timmermann and Halbert White, “Data-Snooping, Technical Trading Rule Performance, and the Bootstrap” (Journal of Finance, 1999), for the cost of a rule with free parameters: the performance of the best setting has to be judged against the size of the universe of settings you were free to search, and the four dials above define exactly such a universe.

The next lesson takes the last of the objects that get drawn on top of price and asks it the same question. A divergence is two series disagreeing — price making a higher high while an oscillator makes a lower one — and it needs a swing definition and an indicator setting before it exists at all. Lesson 34 is about how many numbers that comes to, what the disagreement is being measured over, and what a divergence claims that is capable of failing.

Educational only. Trading involves substantial risk of loss. Not financial advice. Past performance does not guarantee future results.