Pine Script plotcandle: Synthetic OHLC and Missing Bars

Validate all four OHLC values

For a price candle, require four available values, high at least as large as open and close, and low no greater than open and close. plotcandle() itself uses the largest supplied OHLC value as the high and the smallest as the low. That normalization can conceal a swapped column or a broken calculation. A strict guard makes the error visible as an omitted candle instead.

Run a six-case reproduction in a separate pane

This complete v6 indicator repeats six hypothetical cases across chart bars. The values are test coordinates, not market observations. Add it to a chart and open the Data Window. Use Case number to locate each row in the next table; the cycle is based on bar_index, so it is not tied to specific timestamps. Reject invalid ranges is on by default.

//@version=6
indicator("plotcandle input cases", overlay = false)

rejectInvalid = input.bool(true, "Reject invalid ranges")
caseNumber = bar_index % 6
float o = caseNumber == 1 ? 102.0 : caseNumber == 2 ? na : 100.0
float h = caseNumber == 1 ? 105.0 : caseNumber == 3 ? 99.0 : caseNumber == 4 ? 101.0 : 104.0
float l = caseNumber == 1 ? 99.0 : caseNumber == 3 ? 103.0 : caseNumber == 4 ? 99.0 : caseNumber == 5 ? na : 98.0
float c = caseNumber == 1 or caseNumber == 4 ? 100.0 : 102.0

hasOHLC = not na(o) and not na(h) and not na(l) and not na(c)
rangeIsValid = hasOHLC and h >= math.max(o, c) and l <= math.min(o, c) and h >= l
canDraw = hasOHLC and (not rejectInvalid or rangeIsValid)
bodyColor = c > o ? color.teal : c < o ? color.orange : color.gray

plotcandle(canDraw ? o : na, h, l, c, title = "Test candles", color = bodyColor, wickcolor = color.gray, bordercolor = bodyColor)
plot(caseNumber, "Case number", display = display.data_window)
plot(o, "Input open", display = display.data_window)
plot(h, "Input high", display = display.data_window)
plot(l, "Input low", display = display.data_window)
plot(c, "Input close", display = display.data_window)
plot(rangeIsValid ? 1 : 0, "Valid range: 1=yes", display = display.data_window)

Compare each bar with the expected output

These are expected results derived from the inputs and documented plotting rules, not a claim of a live TradingView test. With the guard on, cases 2, 3, and 5 must be empty. Turn the guard off to isolate the built-in normalization in case 3. Missing-value cases stay empty in either mode.

• For case 3, the raw Input high readout remains 99. The displayed upper extreme is max(100, 99, 103, 102) = 103. Normalizing a drawing does not repair its source variables.
• Replacing na with nz(value, 0) would invent a zero price and could stretch the scale. Preserve the gap until a documented fallback is appropriate.

Calculate synthetic OHLC without treating it as a trade

The next indicator averages each OHLC component with the preceding chart bar. It deliberately uses arithmetic rather than filling missing inputs: the first loaded bar has no prior values and stays empty. Use a standard Candles chart when comparing this output with market OHLC. On a Heikin-Ashi chart, the built-in inputs are already synthetic. This two-bar average is not the Heikin-Ashi formula.

• The synthetic close is (102 + 100) / 2 = 101, while the current chart close is 100. The synthetic candle is a doji even though the current standard candle closes below its open.
• The synthetic high 104.5 is an average of highs, not the highest price of the combined two-bar interval; that interval reached 105.
• Drawing a close at 101 neither submits an order nor proves an execution at 101. If this calculation becomes a signal in a strategy, evaluate fills separately under the strategy emulator rules and real trading costs.
• With Wait for chart bar close on, the live bar stays empty until confirmation. Turning it off lets the current synthetic candle change as the chart OHLC updates.

//@version=6
indicator("Two-bar synthetic OHLC", overlay = true, behind_chart = false)

waitForClose = input.bool(true, "Wait for chart bar close")
o = (open + open[1]) / 2.0
h = (high + high[1]) / 2.0
l = (low + low[1]) / 2.0
c = (close + close[1]) / 2.0
hasOHLC = not na(o) and not na(h) and not na(l) and not na(c)
rangeIsValid = hasOHLC and h >= math.max(o, c) and l <= math.min(o, c) and h >= l
canDraw = rangeIsValid and (not waitForClose or barstate.isconfirmed)
directionColor = c > o ? color.teal : c < o ? color.orange : color.gray

plotcandle(canDraw ? o : na, h, l, c, title = "Synthetic means", color = color.new(directionColor, 65), wickcolor = directionColor, bordercolor = directionColor)
plot(close, "Original chart close", display = display.data_window)
plot(c, "Synthetic close", display = display.data_window)

Separate candle colors, pane placement, and visibility

color controls the body, wickcolor controls the wick, and bordercolor controls the outline. The examples explicitly set all three. The first script uses a separate pane because its fixed values could be far from the symbol price. The second uses overlay=true and behind_chart=false to place its candles over the main chart. The native candles still exist underneath.

• A plotcandle call draws on chart-bar positions. It does not accept custom time coordinates or create a new chart timeframe.
• Candles use plot counts, not label IDs. The script limit is 64 plot counts; a candle call can consume up to seven when its color arguments are series.

Give the coding agent inputs and expected results

Send Pineify the complete failing script, the selected chart type, and one mismatching row from the six-case table. Specify whether malformed OHLC should be rejected, which bars may be omitted, and whether the live bar should wait for confirmation. Ask the agent to preserve the input readouts and run a syntax check. Then compare the resulting candle with the same row in TradingView; compilation alone cannot verify chart placement or price semantics.