A Fresh Look at a Single Candlestick

8.1 Open, High, Low, and Close: Checking the Basic Data

A candlestick compresses prices within a defined observation period into four numbers. In a bar constructed from transaction prices, the open O is the first included trade, H and L are the highest and lowest trades, and C is usually the last included trade. Quote charts may instead use Bid, Ask, or midpoint, while some futures daily charts use the exchange settlement price for C. Daily boundaries depend on trading-day and session settings, and minute bars on each bar’s aggregation interval. Check the platform’s field definitions first.

Open, high, low, and close identify the bar’s start, finish, and price range. They cannot identify whether the high or low occurred first or how long price stayed in an area. If a judgment depends on intrabar sequence, inspect finer-timeframe or trade-by-trade data and verify completeness. Even individual trade records cannot reconstruct all resting orders, cancellations, and repricing before those trades.

When O, H, L, and C all come from the same transaction-price bar, first check that H is at least as high as O and C and L no higher than O and C. If C is an independent settlement price, check its platform definition separately; being outside the traded high-low range alone does not prove an error. Then check timezone, sessions, missing bars, and corporate-action handling. Inclusion of the opening auction and after-hours trading directly changes daily-bar boundaries. A bullish candle means only C > O and a bearish candle C < O. Rising or falling relative to the previous close compares C with that previous close; the two classifications may disagree.

Distinguish transaction and quote OHLC fields. H and L on a transaction-price bar establish that at least one trade occurred there. Extremes on a Bid or Ask bar establish only that a quote reached the level, not that an order executed. A midpoint is calculated from bid and ask and cannot directly be treated as executable. For breakouts and stops, identify the chart’s field type and refer to the order’s actual trigger price. Settlement C may be set by an exchange using a particular period or algorithm rather than the last trade. Calculating a body, CP, or price change with it describes relationships involving that settlement field.

Consistency checks go beyond numerical ordering. The same transaction-price bar normally satisfies H ≥ max(O,C) and L ≤ min(O,C), but timestamps, timezone, trading-day boundaries, and extended sessions must also align. A “flat bar” forward-filled by the platform after missing trades differs from real trading at an unchanged price. Identically shaped bars with different data definitions are not directly comparable.

OHLC cannot identify whether the high or low came first, how many trades occurred at a level, or time spent there. Finer timeframes recover more sequence but remain limited by coverage and aggregation boundaries. Confine single-bar conclusions to covered range, start-to-finish relationships, and final closing position.

Figure 8-1 A single candlestick: open, high, low, and close (schematic; bullish candle with C > O)

8.2 High-Low Spread, Body, and Wicks: Different Measurements

A bar’s price range is H-L, the range covered during the observation period. Body length is |C-O|, the net change retained from open to close. Upper-wick length is H-max(O,C), and lower-wick length is min(O,C)-L. All four use the same price units but answer different questions: range measures coverage, body measures the start-to-finish result, and wicks measure how far extremes lie beyond the open and close.

A large body with short wicks means open-to-close displacement accounts for much of the full range. A small body with a large range means price covered a broad area but closed near the open. Neither reconstructs the number of two-way trades or directly proves reversal. A long upper wick shows only that higher traded prices were not retained to the close; the lower wick is analogous. Whether it constitutes structural rejection depends on the boundary and subsequent swing.

In VSA material, “spread” usually means the H-L price range, not the distance between the best bid and ask in the order book. The latter is the bid-ask spread, describing the gap between displayed quotes at an instant. Both may widen under poor liquidity, but they measure different objects. When encountering a “spread” field, establish whether it means the whole bar’s range or the distance between bid and ask.

When O, H, L, and C come from the same valid trade sequence, with H at least as high as O and C and L no higher than either, all four lengths should be nonnegative. H-L measures full coverage; |C-O| measures only open-to-close distance. The upper wick extends from the higher of O and C to H, and the lower from L to the lower of O and C. Bullish or bearish direction does not change the body-length formula. If O = C, body length is zero even when range is large. If H = L, all four transaction fields are normally equal and range, body, and wicks are zero. If an independent settlement C lies outside traded H and L, do not insert these mixed fields directly into ordinary wick formulas.

Wicks show distances between extremes and the open and close, not their causes. A long upper wick establishes that the highest trade exceeded both endpoints and the final trade was below the high. It cannot reveal whether H or L came first or whether price met new selling, profit taking, or thin quotes above. Lower wicks share these limits. A wick at a clear boundary followed by further movement in the opposite direction strengthens an interpretation of rejection.

Bar range and order-book spread may both widen in low liquidity, but have no fixed ratio. The quote spread can change without trading, whereas bar range comes from the selected field’s extremes across the entire period. Execution concerns live Bid/Ask and depth; price-volume analysis concerns the period’s traded range. Similar names do not make them interchangeable.

8.3 Closing Position: Where Price Finishes within the Bar

For normal transaction-price OHLC satisfying L ≤ C ≤ H and H > L, this course defines the 0–1 normalized closing position as CP=(C-L)/(H-L). The numerator measures the close’s distance above the low; the denominator normalizes that distance by the bar’s full range. CP near 1 means a close near the high, near 0 a close near the low, and 0.5 the midpoint. Bars with different absolute ranges can thus be compared on the same scale.

Some platforms use CLV=((C-L)-(H-C))/(H-L)=2CP-1, linearly mapping the same position to -1 through 1: 1 at the high, 0 at the midpoint, and -1 at the low. CLV adds no new price information. Choose a definition and use it consistently; CP of 0.5 is not CLV of 0.5. If C is an independent settlement price that may lie outside traded H and L, first check field definitions as in Section 8.1 rather than applying these range interpretations directly.

Normalization provides comparability while discarding price units, absolute range, and intrabar sequence. Two bars with identical CP can have completely different ranges, bodies, volumes, and ordering of highs and lows. CP also omits O and cannot replace body direction. A high CP can accompany a large bullish body, a small body after a gap down, or even a bar entirely below the previous close.

When H = L, the denominator is zero and both CP and CLV are mathematically undefined. Mark them “no range / not calculable,” rather than forcing 0, 0.5, or 1. No range may result from unchanged actual trade prices, a trading halt, price limits, sparse data, or platform filling. Missing H, L, or C must also remain missing rather than become zero. If software uses a conventional display value, exclude or separately classify it before calculating means or quantiles.

Trading significance still depends on location and subsequent action. A high CP after a probe at support establishes only that the bar closed away from its low. If contemporaneous trade data also show buyers actually absorbing sell orders, later action does not erase that historical fact. A subsequent break below L that holds invalidates the expectation that “the absorption was sufficient to keep support effective.” A low CP at resistance uses the mirror test. C, H, and L on an unfinished bar still change, so current CP is provisional. Intrabar strategies may execute predefined conditions; closing strategies wait for the aggregation period to end.

Figure 8-3 Four different candlesticks with the same closing position (schematic)

8.4 Relative Range: Comparing a Bar with Earlier Volatility

“Wide” and “narrow” require a baseline. The simplest comparison uses several neighboring completed bars; alternatives include the same session period, similar events, or the recent range distribution. Keep the selected baseline consistent to determine whether this bar’s coverage is normal expansion, contraction, or an anomaly. Absolute range also depends on price level. Across long periods or instruments, percentage range or a consistently defined volatility scale may be used.

Extremely wide bars can raise the mean, while the median is more robust to outliers; together they can describe recent normal conditions. A window that is too short mistakes one event for a new normal; one that is too long can conceal a changed volatility regime. “Relatively wide” should therefore name its comparison rather than become a permanent label. Natural volatility differences between session periods should be treated separately, like volume.

Expanding range means only that price covered more ground. If the body expands too, the close is near the directional end, and continuation follows, much of the intrabar move was retained. If range expands but the body is small or the close retreats, price visited farther levels and returned. Illiquid gaps, announcements, and price limits can also cause large ranges and must be interpreted with activity and market mechanics.

Relative-range comparisons first require a scale. Adjacent bars reveal sudden expansion; historical bars from the same session period remove natural differences between the open and midday; the recent distribution places the current range within normal conditions. Across price levels, a percentage scale such as (H-L)/reference price can be used, but the reference must be a predefined positive value. Zero or negative prices cannot use this ratio unchanged; define another positive scale or use absolute range. Across instruments, consider minimum tick size and session differences rather than comparing absolute points alone.

Time also affects interpretation. The same range formed quickly versus slowly over a full day shows different speeds of range expansion. Directional speed still requires net displacement divided by elapsed time. Fast coverage can contain back-and-forth movement, so also inspect body, close, and net displacement across successive bars. Increased volatility can enlarge normal excursions near support and resistance; do not retain a fixed “wide” label from the old regime. The baseline window should reflect recent conditions without mixing clearly different volatility regimes.

Unusually wide ranges have several possible causes. News repricing may produce high-volume jumps, a thin book may let small trades cross many levels, and faulty adjustments may fabricate a range. Wide range supports improved progress only when data are reliable, the baseline is comparable, the body shows clear net displacement, and subsequent price continues in the same direction. A wide bar quickly reclaimed in full by the next move more strongly indicates expanded volatility or boundary rejection.

8.5 Progress, Overlap, and Reactions across Successive Bars

Successive bars provide relative relationships missing from one bar. Upward progress examines higher highs and closes; downward progress examines lower lows and closes. Incremental extreme distance measures only the current bar’s extension beyond the prior reference extreme and cannot be replaced by its full range. A bar making a new high but closing inside the previous bar’s range achieved some progress without retaining it fully.

Overlap is the intersection of two bars’ price ranges. Increasing intersection generally suggests greater overlap in progress, but OHLC cannot prove repeated trading at every price in that intersection; net displacement and finer data are still needed. Less overlap with successive same-direction extreme extensions indicates smoother movement. An inside bar lies wholly within the previous range; an outside bar crosses both the previous high and low. These are geometric relationships only. Direction requires the close, location, and subsequent departure from the combined range.

A reaction is movement opposite to a specified advance. Compare how much it reclaims, how long it lasts, whether it damages the advance’s origin or key support, and whether the original direction resumes afterward. Across bars, shorter progress, more overlap, and larger reactions show weakening rhythm. Only a break and hold through key structure in the opposite direction further supports directional change.

The intersection of two bars can be written [max(L_previous,L_current), min(H_previous,H_current)]. Overlap exists only if the upper endpoint is no lower than the lower endpoint. A longer intersection means the periods covered more similar areas, but does not establish their trading sequence within it. To compare unequal ranges, divide the intersection length by a predefined reference range. The measure is undefined when that denominator is zero or required fields are missing.

Separate extremes from closes during continuing progress. Rising highs without rising closes show repeated probes of high prices without retention. Rising closes with shorter incremental high extensions show ongoing net progress while approaching resistance. Calculate the sequence’s net displacement from its start to the current endpoint rather than adding every bar’s full range, which repeatedly counts overlap.

A retracement ratio requires explicit start and end points for its reference advance and only information confirmed at the time. A swing extreme may not yet be confirmed when it occurs; confirmation is later. Reviews must not use endpoints established afterward as if known earlier. A reaction holding the breakout area with limited opposing volume and distance leaves the preceding advance effective. A return to the advance’s starting point with sustained trading below damages the original directional structure.

8.6 Gaps, Unfinished Bars, and Observation Time

A gap is a price interval between adjacent observation periods not covered by trading in that chart. An upward gap may have the current low above the previous high; a downward gap is the reverse. The looser term “opening gap” compares only the current open with the previous close. Name them separately. Gaps can arise from information during a closure, call auctions, insufficient liquidity, or price limits. Appearance alone cannot establish necessarily strong demand or supply.

H, L, and C of an unfinished bar may keep changing; the displayed “close” is actually only the latest field value. Using an unfinished bar to calculate wicks, CP, range, or volume can mistake unsettled data for final facts. A real-time judgment is valid only at its observation time. Rules requiring closing confirmation must await the actual end of that timeframe. Longer-timeframe bars are especially likely to contain much unfinished time.

Intrabar observation also raises information-timing issues. If complete OHLC becomes available only after the bar ends, one cannot assume its final candle shape was known when an intrabar extreme occurred. Backtests and reviews must trigger decisions from data visible then. The first bar after a gap also requires separating overnight displacement from post-open trading results, rather than combining two different mechanisms.

A true range gap requires completely disjoint traded ranges in adjacent bars. An open away from the previous close that fills back during the current bar is only an opening gap. Blank areas on quote charts may result from jumping quotes, while missing trade data can create false gaps. Stock splits, dividend adjustments, and continuous-futures roll adjustments alter historical prices. Verify raw single-market data rather than interpreting an adjustment seam as supply-and-demand imbalance.

Whether a gap is “filled” also requires specified fields and boundaries. A traded price touching the empty area differs from a quote briefly entering it. Partial filling, complete coverage, and sustained trading beyond the gap have different implications. A trend-breakout gap that holds afterward may support acceptance of a new area; a quick return to the old range means the breakout did not persist. A gap guarantees neither continuation nor permanent support or resistance.

Unfinished bars can support real-time observation, but every derived measure is provisional. H can only rise and L only fall; current C changes with the latest trade, updating range, wicks, CP/CLV, and volume. Closing-confirmation rules wait for the current aggregation period to finish; trade-by-trade, order-book, or intrabar-price rules execute on contemporaneous information. Both must avoid using a candle completed later in backtests.

8.7 How Structural Location Qualifies a Single Bar

A bar’s shape belongs within horizontal areas, trend stage, and the current swing. A downward probe and recovery near support initially shows only that lower prices were not retained. If preceding downward progress was already shortening and demand subsequently crosses and holds a local high, buying support becomes more plausible. The same lower wick in the range middle is usually only an intrabar retreat because it tests no important boundary.

A wide bar closing high at resistance, followed by sustained outside trading, shows buyers moving price out of the old range. If the next movement quickly returns inside, the bar becomes only part of a failed breakout. A high-volume narrow bar late in a trend may indicate opposing restraint; in the middle of consolidation, it may be ordinary two-way turnover. Compare earlier responses at that location, current activity, persistence outside, and the direction of subsequent resumption.

Analyze range, body, and closing position first, then whether the bar is near support, crossing a previous high, or in the range middle. Absorbed supply or demand failing to persist are provisional interpretations using location and require the next movement’s test. This uses the convenience of compressed candle information without inventing intrabar sequence or participant identity from OHLC.

Location also identifies contrary evidence. If a lower wick at support is interpreted as rejection of low prices, subsequent action should hold near its low and demand should cross local resistance. A fresh break and new swings below invalidate that explanation. A retreat from highs at resistance requires supply to push price lower; if demand regains and holds the high, it cannot still be called supply control. Without a nearby boundary in the range middle, the same shape generally offers no clear direction or risk conditions.

Trend stage changes the meaning of similar candles. A low-volume narrow bar mid-uptrend may be a normal pause. Repeated high-volume wide bars closing low at long-term resistance in a mature advance suggest buyers increasingly struggle to retain high prices, though expanding opposing swings are still needed for a turning-point interpretation. A high-volume lower wick late in a decline is initially only candidate stopping action, not accumulation. One bar flags anomalies, sequences show persistence, and swings and structure determine whether the change affects a trade.