This is a compendium of frequently asked questions on Pine. Answers often give code examples or link to the best sources on the subject.
Do not make the mistake of assuming this is strictly beginner’s material; some of the questions and answers explore advanced techniques.
Reusing this code: You are welcome to reuse this code in your open source scripts only. No permission is required from PineCoders. Credits are appreciated.
close variable holds both the price at the close of historical bars and the current price when an indicator is running on the realtime bar. If the script is a strategy running on the realtime bar, by default it runs only at the bar’s close. If the
calc_on_every_tick parameter of the
strategy() declaration statement is set to
true, the strategy will behave as an indicator and run on every price change of the realtime bar.
To access the close of the previous bar’s close in Pine, use
close. In Pine, brackets are used as the history-referencing operator.
greenCandle = close > open
Once you have defined the
greenCandle variable, if you wanted a boolean variable to be
true when the last three candles were green ones, you could write:
threeGreenCandles = greenCandle and greenCandle and greenCandle
Note that the variable name
3GreenCandleswould have caused a compilation error. It is not legal in Pine as it begins with a digit.
If you need to define up and down candles, then make sure one of those definitions allows for the case where the
close are equal:
upCandle = close >= open downCandle = close < open
Most probably because you are trying to use a series integer instead of a simple integer as the second parameter (the length). Either use a simple integer or use the RicardoSantos replacements here. If you don’t know Ricardo, take the time to look at his indicators while you’re there. Ricardo is among the most prolific and ingenious Pine coders out there.
== is a comparison operator used to test for true/false conditions.
= is used to declare and initialize variables.
:= is used to assign values to variables after initialization, transforming them into mutable variables.
//@version=3 study("") a = 0 b = 1 plot(a == 0 ? 1 : 2) plot(b == 0 ? 3 : 4, color = orange) a := 2 plot(a == 0 ? 1 : 2, color = aqua)
:=operator to assign values to past values of a series?
No. Past values in Pine series are read-only, as is the past in real life. Only the current bar instance (
variableName) of a series variable can be assigned a value, and when you do, the
 history-referencing operator must not be used—only the variable name.
What you can do is create a series with the values you require in it as the script is executed, bar by bar. The following code creates a new series called
range with a value containing the difference between the bar’s
open, but only when it is positive. Otherwise, the series value is zero.
range = close > open ? close - open : 0.0
In the previous example, we could determine the value to assign to the
range series variable as we were going over each bar in the dataset because the condition used to assign values was known on that bar. Sometimes, you will only obtain enough information to identify the condition after a number of bars have elapsed. In such cases, a
for loop must be used to go back in time and analyse past bars. This will be the case in situations where you want to identify fractals or pivots. See the Pivots Points High/Low from the User Manual, for example.
Pine logical expressions have 3 possible values: true, false and na. Whenever a value used in the logical expression has na value, the result of the logical expression will be na. Furthermore, when a logical expression evaluates to na, the false branch of a conditional statement will be executed. This may lead to unexpected behavior and entails that special cases must be accounted for if you want your code to handle all possible logical expression results according to your expectations.
Let’s take a case where, while we are debugging code, we want to compare two variables that should always have the same value, but where one of the variables or both can have an na value. When that is the case, neither
a == b nor
a != b will return true or false, as they both return na.
When we undestand this, we can see why the first
bgcolor() line in the following code shows no background. While you could expect the
a != b logical expression to be true and thus the background to appear lime because the value of variable
a does not equal the value of
b, this is not the case. Because the logical expression returns na, the false branch of the ternary is executed and no color is plotted in the background.
bgcolor() line will produce the behavior we expect. You will see this if you comment out the first one and uncomment this second line. The other lines show different variations of the concept.
//@version=4 study("") int a = 1 int b = na bgcolor(a != b ? color.lime : na, transp = 20) // na, so goes to false branch. // bgcolor(a == b ? na : color.red, transp = 20) // na, so goes to false branch. // bgcolor(na((a != b)) ? color.orange : na, transp = 20) // true, so works. // bgcolor(a != b or na(a != b) ? color.fuchsia : na, transp = 20) // true, so works.
You’ll need to define your start and stop conditions and use logic to remember states and the level you want to plot.
plot() call using a combination of plotting
na and the
style = plot.style_linebr parameter to avoid plotting a continuous line, which would produce inelegant joins between different levels.
Also note how
plotchar() is used to plot debugging information revealing the states of the boolean building blocks we use in our logic. These plots are not necessary in the final product; they are used to ensure your code is doing what you expect and can save you a lot of time when you are writing your code.
//@version=4 study("Plot line from start to end condition", overlay=true) lineExpiryBars = input(300, "Maximum bars line will plot", minval = 0) // Stores "close" level when start condition occurs. var savedLevel = float(na) // True when the line needs to be plotted. var plotLine = false // This is where you enter your start and end conditions. startCondition = pivothigh(close, 5, 2) endCondition = cross(close, savedLevel) // Determine if a line start/stop condition has occurred. startEvent = not plotLine and startCondition // If you do not need a limit on the length of the line, use this line instead: endEvent = plotLine and endCondition endEvent = plotLine and (endCondition or barssince(startEvent) > lineExpiryBars) // Start plotting or keep plotting until stop condition. plotLine := startEvent or (plotLine and not endEvent) if plotLine and not plotLine // We are starting to plot; save close level. savedLevel := close // Plot line conditionally. plot(plotLine ? savedLevel : na, color = color.orange, style = plot.style_linebr) // State plots revealing states of conditions. plotchar(startCondition, "startCondition", "•", color = color.green, size=size.tiny, transp = 0) plotchar(endCondition, "endCondition", "•", color = color.red, size=size.tiny, location = location.belowbar, transp = 0) plotchar(startEvent, "startEvent", "►", color = color.green, size=size.tiny) plotchar(endEvent, "endEvent", "◄", color = color.red, size=size.tiny, location = location.belowbar)
To plot a continuous line in Pine, you need to either:
line.new()function available in v4. These are examples of three different techniques used to determine and draw support lines:
alertcondition()calls also count for one plot.
line.new()a garbage collector will preserve only the last ~50 objects of each type.
security()calls is 40.
See Working with colours by Kodify.
study() declaration statement, e.g.,:
study("My Script", overlay = true)
If your indicator was already in a Pane before applying this change, you will need to use Add to Chart again for the change to become active.
plot()calls in a
No, but you can use the v4
line.new() function in
You can use the
plot.style_columns style to plot them:
//@version=4 study("", "", true, scale = scale.none) cond = close > open plot(cond ? 10e20 : na, style = plot.style_columns, color = color.silver, transp=85)
There is a nice v4 function to plot a vertical line in this indicator: vline() Function for Pine Script v4.0+.
You need to use the
security() function. This script allows you to view normal candles on the chart, although depending on the non-standard chart type you use, this may or may not make much sense:
//@version=4 study("Plot underlying OHLC", "", true) // ————— Allow plotting of underlying candles on chart. plotCandles = input(true, "Plot Candles") method = input(1, "Using Method", minval = 1, maxval = 2) // ————— Method 1: Only works when chart is on default exchange for the symbol. o1 = security(syminfo.ticker, timeframe.period, open) h1 = security(syminfo.ticker, timeframe.period, high) l1 = security(syminfo.ticker, timeframe.period, low) c1 = security(syminfo.ticker, timeframe.period, close) // ————— Method 2: Works all the time because it use the chart's symbol and exchange information. ticker = tickerid(syminfo.prefix, syminfo.ticker) o2 = security(ticker, timeframe.period, open) h2 = security(ticker, timeframe.period, high) l2 = security(ticker, timeframe.period, low) c2 = security(ticker, timeframe.period, close) // ————— Get value corresponding to selected method. o = method == 1 ? o1 : o2 h = method == 1 ? h1 : h2 l = method == 1 ? l1 : l2 c = method == 1 ? c1 : c2 // ————— Plot underlying close. plot(c, "Underlying close", color = color.gray, linewidth = 3, trackprice = true) // ————— Plot candles if required. invisibleColor = color.new(color.white, 100) plotcandle(plotCandles ? o : na, plotCandles ? h : na, plotCandles ? l : na, plotCandles ? c : na, color = color.orange, wickcolor = color.orange) // ————— Plot label. f_print(_txt) => var _lbl = label(na), label.delete(_lbl), _lbl := label.new(time + (time-time)*3, c, _txt, xloc.bar_time, yloc.price, size = size.large) a = f_print("Underlying Close1 = " + tostring(c1) + "\nUnderlying Close2 = " + tostring(c2) + "\nChart's close = " + tostring(close) + "\n Delta = " + tostring(close - c))
The first thing required is to maintain a series containing the ids of the labels or lines as they are created. This is accomplished by assigning the returning value of the
line.new() function to a variable. This creates a series with value
na if no label or line was created from that bar, and with a value of type label or line when an element is created.
The next step will be to run a loop going back into the past from the current bar, jumping over a preset number of labels or lines and deleting all those following that, all the while doing nothing when an
na value is found since this means no label or line was created on that bar.
This first example illustrates the technique using labels:
//@version=4 //@author=LucF, for PineCoders maxBarsBack = 2000 study("Keep last x labels", "", true, max_bars_back = maxBarsBack) keepLastLabels = input(5, "Last labels to keep") // ————— Label-creating condition: when close is above ma. ma = sma(close,30) var aboveMa = false aboveMa := crossover(close, ma) or (aboveMa and not crossunder(close, ma)) // ————— Count number of bars since last crossover to show it on label. var barCount = 0 barCount := aboveMa ? not aboveMa ? 1 : barCount + 1 : 0 // ————— Create labels while keeping a trail of label ids in series "lbl". // This is how we will later identify the bars where a label exist. label lbl = na if aboveMa lbl := label.new(bar_index, high, tostring(barCount), xloc.bar_index, yloc.price, size = size.small) // ————— Delete all required labels. // Loop from previous bar into the past, looking for bars where a label was created. // Delete all labels found in last "maxBarsBack" bars after the required count has been left intact. lblCount = 0 for i = 1 to maxBarsBack if not na(lbl[i]) // We have identified a bar where a label was created. lblCount := lblCount + 1 if lblCount > keepLastLabels // We have saved the required count of labels; delete this one. label.delete(lbl[i]) plot(ma)
The second example illustrates the technique using lines:
//@version=4 //@author=LucF, for PineCoders maxBarsBack = 2000 study("Keep last x lines", "", true, max_bars_back = maxBarsBack) // On crossovers/crossunders of these MAs we will be recording the hi/lo reched until opposite cross. // We will then use these hi/los to draw lines in the past. ma1 = sma(close, 20) ma2 = sma(close, 100) // ————— Build lines. // Highest/lowest hi/lo during up/dn trend. var hi = 10e-10 var lo = 10e10 // Bar index of highest/lowest hi/lo. var hiBar = 0 var loBar = 0 // Crosses. crossUp = crossover(ma1, ma2) crossDn = crossunder(ma1, ma2) upTrend = ma1 > ma2 // Draw line in past when a cross occurs. line lin = na if crossUp or crossDn lin := line.new(bar_index[bar_index - hiBar], high[bar_index - hiBar], bar_index[bar_index - loBar], low[bar_index - loBar], xloc.bar_index, extend.none, color.black) // Reset hi/lo and bar index on crosses. if crossUp hi := high hiBar := bar_index else if crossDn lo := low loBar := bar_index // Update higher/lower hi/lo during trend. if upTrend and high > hi hi := high hiBar := bar_index else if not upTrend and low < lo lo := low loBar := bar_index plot(ma1, "MA1", color.aqua, 1) plot(ma2, "MA2", color.orange, 1) // ————— Delete all required lines. // Loop from previous bar into the past, looking for bars where a line was created. // Delete all lines found in last "maxBarsBack" bars after the required count has been left intact. keepLastLines = input(5) lineCount = 0 for i = 1 to maxBarsBack if not na(lin[i]) lineCount := lineCount + 1 if lineCount > keepLastLines line.delete(lin[i])
It’s possible, but not trivial. See this RicardoSantos script.
We will use a label to print our value. Labels however, require positioning relative to the symbol’s price scale, which is by definition fluid. The technique we use here is to create an indicator running in “No Scale” space, and then create an artificially large internal scale for it by using the
plotchar() call which doesn’t print anything. We then print the label at the top of that large scale, which does not affect the main chart display because the indicator is running in a separate scale.
Also note that we take care to only print the label on the last bar of the chart, which results in much more efficient code than if we deleted and re-created a label on every bar of the chart, as would be the case if the
if barstate.islast condition didn’t restrict calls to our
f_print() label-creating function.
//@version=4 //@author=LucF, for PineCoders // Indicator needs to be on "no scale". study("", "Daily ATR", true, scale = scale.none) atrLength = input(14) barsRight = input(5) // Adjust the conversion formatting string to the instrument: e.g., "#.########" for crypto. numberFormat = input("#.####") // Plot invisible value to give a large upper scale to indie space. plotchar(10e10, "", "") // Fetch daily ATR. We want the current daily value so we use a repainting security() call. dAtr = security(syminfo.tickerid, "D", atr(atrLength), lookahead = barmerge.lookahead_on) // Label-creating function puts label at the top of the large scale. f_print(_txt) => var _lbl = label(na), label.delete(_lbl), _lbl := label.new(time + (time-time) * barsRight, 10e10, _txt, xloc.bar_time, yloc.price, size = size.normal) // Print value on last bar only, so code runs faster. if barstate.islast f_print(tostring(dAtr, numberFormat))
hline()plots on and off?
showHline = input(true) hline(50, color = showHline ? color.blue : #00000000)
Use the following in your code:
ExternalIndicator = input(close, "External Indicator")
From the script’s Inputs you will then be able to select a plot from another indicator if it present on your chart. You can use only one such statement in your script. If you use more than one, the other indicator plots will not be visible from the Inputs dropdown. You cannot use this technique in strategies.
No. TradingView uses special code for these that is not available to standard Pine scripts.
No. The only way for now is through screen scraping.
The only thing that can be changed on the chart from within a pane is the color of the bars. See the
Sure, but start by looking at the scale each one is using. If you’re thinking of merging a moving average indicator designed to plot on top of candles and in relation to them, you are going to have problems if you also want to include and indicator showing volume bars in the same script because their values are not on the same scale.
Once you’ve made sure your scales will be compatible (or you have devised a way of normalizing/re-scaling them), it’s a matter of gathering the code from all indicators into one script and removing any variable name collisions so each indicator’s calculations retain their independence and integrity.
Note that if the indicators you’ve merged are CPU intensive, you may run into runtime limitations when executing the compound script.
TradingView backtesting evaluates conditions at the close of historical bars. When a condition triggers, the associated order is executed at the open of the next bar, unless
process_orders_on_close=true in the
strategy() declaration statement, in which case the broker emulator will try to execute orders at the bar’s close.
In the real-time bar, orders may be executed on the tick (price change) following detection of a condition. While this may seem appealing, it is important to realize that if you use
cal_on_every_tick=true in the
strategy() declaration statement to make your strategy work this way, you are going to be running a different strategy than the one you tested on historical bars. See the Strategies page of the User Manual for more information.
DateFilter = input(false, "═════════════ Date Range Filtering") FromYear = input(1900, "From Year", minval = 1900) FromMonth = input(1, "From Month", minval = 1, maxval = 12) FromDay = input(1, "From Day", minval = 1, maxval = 31) ToYear = input(2999, "To Year", minval = 1900) ToMonth = input(1, "To Month", minval = 1, maxval = 12) ToDay = input(1, "To Day", minval = 1, maxval = 31) FromDate = timestamp(FromYear, FromMonth, FromDay, 00, 00) ToDate = timestamp(ToYear, ToMonth, ToDay, 23, 59) TradeDateIsAllowed() => DateFilter ? (time >= FromDate and time <= ToDate) : true
You can then use the result of
TradeDateIsAllowed() to confirm your entries using something like this:
EnterLong = GoLong and TradeDateIsAllowed()
Note that with this code snippet, date filtering can be enabled/disabled using a checkbox. This way you don’t have to reset dates when filtering is no longer needed; just uncheck the box.
Because non-standard chart types use non-standard prices which produce non-standard results. See our Backtesting on Non-Standard Charts: Caution! - PineCoders FAQ indicator and its description for a more complete explanation.
The TradingView Help Center also has a good article on the subject.
Backtest Rookies has a blog post on the subject.
Here are two ways you can go about it:
//@version=4 // Mod of original code at https://www.tradingview.com/script/bHTnipgY-HOWTO-Plot-Entry-Price/ strategy("Plot Entry Price", "", true) longCondition = crossover(sma(close, 14), sma(close, 28)) if (longCondition) strategy.entry("My Long Entry Id", strategy.long) shortCondition = crossunder(sma(close, 14), sma(close, 28)) if (shortCondition) strategy.entry("My Short Entry Id", strategy.short) // ————— Method 1: wait until bar following order and use its open. var float entryPrice = na if longCondition or shortCondition entryPrice := open plot(entryPrice, "Method 1", color.orange, 3, plot.style_circles) // ————— Method 2: use built-in variable. plot(strategy.position_avg_price, "Method 2", color.gray, 1, plot.style_circles, transp = 0)
The best way to go about this is to write your strategies in such a way that their behavior depends the least possible on
strategy.* variables and
strategy.*() call parameters, because these cannot be converted into an indicator.
The PineCoders Backtesting-Trading Engine is a framework that allows you to easily convert betweeen strategy and indicator modes because it manages trades using custom Pine code that does not depend on an involved setup of
strategy.*() call parameters.
No. The brokers can only be used for manual trading. Currently, the only way to automate trading using TradingView is to:
alertcondition()calls in your indicator using your buy/sell conditions.
time[bar_index] will return that time in Unix format, i.e., the number of milliseconds that have elapsed since 00:00:00 UTC, 1 January 1970.
Use this function by RicardoSantos.
To do this, we will use the
security() function called at the 1D resolution and have it evaluate the
barstate.islast variable on that time frame, which returns true when the bar is the last one in the dataset, even if it is not a realtime bar because the market is closed. We also allow the
security() function to lookahead, otherwise it will only return true on the last bar of the chart’s resolution.
//@version=4 study("") lastDay = security(syminfo.tickerid, "D", barstate.islast, lookahead = barmerge.lookahead_on) bgcolor(lastDay ? color.red : na)
//@version=4 //@author=mortdiggiddy, for PineCoders study("Detect today", "", true) currentYear = year(timenow) currentMonth = month(timenow) currentDay = dayofmonth(timenow) today = year == currentYear and month == currentMonth and dayofmonth == currentDay bgcolor(today ? color.gray : na)
Use the PineCoders
//@version=4 //@author=LucF, for PineCoders study("Multiple of current TF v4", precision = 8) // Get multiple. resMult = input(2, minval = 1) // Returns a multiple of current TF as a string usable with "security()". f_multipleOfRes(_mult) => // Convert target timeframe in minutes. _targetResInMin = timeframe.multiplier * _mult * ( timeframe.isseconds ? 1. / 60. : timeframe.isminutes ? 1. : timeframe.isdaily ? 1440. : timeframe.isweekly ? 10080. : timeframe.ismonthly ? 43800. : na) // Find best way to express the TF. _targetResInMin <= 0.0417 ? "1S" : _targetResInMin <= 0.167 ? "5S" : _targetResInMin <= 0.376 ? "15S" : _targetResInMin <= 0.751 ? "30S" : _targetResInMin <= 1440 ? tostring(round(_targetResInMin)) : _targetResInMin <= 43800 ? tostring(round(min(_targetResInMin / 1440, 365))) + "D" : tostring(round(min(_targetResInMin / 43800, 12))) + "M" // ————— Get string corresponding to current and target resolution. curResString = f_multipleOfRes(1) resString = f_multipleOfRes(resMult) // ————— Calculate current and target resolution RSI. // Current TF rsi. myRsi = rsi(close, 14) // No repainting target resolution TF rsi. myRsiHtf = security(syminfo.tickerid, resString, myRsi, lookahead = barmerge.lookahead_on) // Repainting target resolution TF rsi. myRsiHtf2 = security(syminfo.tickerid, resString, myRsi) // ————— Plots plot(myRsi, "Current TF RSI", color = color.silver) plot(myRsiHtf, "Target TF no repainting RSI", color = color.green) plot(myRsiHtf2, "Target TF repainting RSI", color = color.red) hline(0) hline(100) // Show resolution information label. var lbl = label(na) if barstate.islast label.delete(lbl) lbl := label.new(bar_index, max(max(myRsiHtf, myRsi), myRsiHtf2) * 1.1, "Current Res = " + curResString + "\nMultiple = " + tostring(resMult) + "\n Target Res = " + resString, xloc = xloc.bar_index, yloc = yloc.price)
For v3, use:
//@version=3 //@author=LucF, for PineCoders study("Multiple of current TF v3") // Get multiple. resMult = input(2, minval = 1) // Returns a multiple of current TF as a string usable with "security()". f_multipleOfRes(_mult) => // Convert target timeframe in minutes. _targetResInMin = interval * _mult * ( isseconds ? 1. / 60. : isminutes ? 1. : isdaily ? 1440. : isweekly ? 10080. : ismonthly ? 43800. : na) // Find best way to express the TF. _targetResInMin <= 0.0417 ? "1S" : _targetResInMin <= 0.167 ? "5S" : _targetResInMin <= 0.376 ? "15S" : _targetResInMin <= 0.751 ? "30S" : _targetResInMin <= 1440 ? tostring(round(_targetResInMin)) : _targetResInMin <= 43800 ? tostring(round(min(_targetResInMin / 1440, 365))) + "D" : tostring(round(min(_targetResInMin / 43800, 12))) + "M" // ————— Calculate current and target resolution RSI. // Current TF rsi. myRsi = rsi(close, 14) // No repainting target resolution TF rsi. myRsiHtf = security(tickerid, f_multipleOfRes(resMult), myRsi, lookahead = barmerge.lookahead_on) // Repainting target resolution TF rsi. myRsiHtf2 = security(tickerid, f_multipleOfRes(resMult), myRsi) // ————— Plots plot(myRsi, "Current TF RSI", color = silver) plot(myRsiHtf, "Target TF no repainting RSI", color = green) plot(myRsiHtf2, "Target TF repainting RSI", color = red) hline(0) hline(100)
Use the PineCoders
f_resInMinutes() function to get the current resolution expressed in minutes of type float.
You can then manipulate it and use the
f_resFromMinutes(_minutes) function to obtain a string usable in
//@version=4 study("Resolution in minutes", "", true) f_resInMinutes() => // Converts current timeframe into minutes of type float. _resInMinutes = timeframe.multiplier * ( timeframe.isseconds ? 1. / 60. : timeframe.isminutes ? 1. : timeframe.isdaily ? 1440. : timeframe.isweekly ? 10080. : timeframe.ismonthly ? 43800. : na) f_resFromMinutes(_minutes) => // Converts a resolution expressed in minutes into a string usable by "security()" _minutes <= 0.0417 ? "1S" : _minutes <= 0.167 ? "5S" : _minutes <= 0.376 ? "15S" : _minutes <= 0.751 ? "30S" : _minutes <= 1440 ? tostring(round(_minutes)) : _minutes <= 43800 ? tostring(round(min(_minutes / 1440, 365))) + "D" : tostring(round(min(_minutes / 43800, 12))) + "M" f_print(_txt) => var _lbl = label(na), label.delete(_lbl), _lbl := label.new(time + (time-time)*3, high, _txt, xloc.bar_time, yloc.price, size = size.large) resInMinutes = f_resInMinutes() resFromMinutes = f_resFromMinutes(resInMinutes) f_print("f_resInMinutes() = " + tostring(resInMinutes) +"\nf_resFromMinutes(_minutes) = " + resFromMinutes)
//@version=4 study("Resolution in minutes", "") higherRes = input("1D", "Interval used for security() calls", type = input.resolution) f_tfResInMinutes(_resolution) => // Returns resolution of _resolution period in minutes. // _resolution: resolution of other timeframe (in timeframe.period string format). _mult = security(syminfo.tickerid, _resolution, timeframe.multiplier) _res = security(syminfo.tickerid, _resolution, timeframe.isseconds ? 1 : timeframe.isminutes ? 2 : timeframe.isdaily ? 3 : timeframe.isweekly ? 4 : timeframe.ismonthly ? 5 : na) _return = _res == 1 ? _mult / 60 : _res == 2 ? _mult : _res == 3 ? _mult * 1440 : _res == 4 ? _mult * 10080 : _res == 5 ? _mult * 43800 : na higherResInMinutes = f_tfResInMinutes(higherRes) plot(higherResInMinutes, "higherResInMinutes", color.navy, linewidth = 10) f_print(_txt) => var _lbl = label(na), label.delete(_lbl), _lbl := label.new(time + (time-time)*3, higherResInMinutes, _txt, xloc.bar_time, yloc.price, size = size.large) f_print("Higher Resolution = " + tostring(higherResInMinutes))
security()on lower intervals than the chart’s current interval?
Yes but there are limits to using this technique:
security()at 1m from a 15m chart, but not 30sec.
If you call
security() at a lower resolution using a series argument such as
volume for its
security() returns the series’ value at the last intrabar, as in the
lastClose variable in the following script.
If you use a function as the
expression= argument, then that function will be executed on each intrabar, starting from the earliest one and ending at the most recent, even if the number of intrabars is sometimes irregular. The two functions used in the following code illustrate how you can use
_res is the chart’s current resolution) to detect the first intrabar the function is running on:
//@version=4 study("Intrabar inspection") insideBarNo = input(4, minval=1) // Current chart resolution. This needs to reflect the chart resolution you want the code working from. curRes = input("D", "Current resolution") // Lower TF we are inspecting. Cannot be in seconds and must be lower that chart's resolution. insideRes = input("60", "Inside resolution") f_qtyIntrabars(_res) => // Returns qty of intrabars in current chart bar. var int _initCnt = 0 _initCnt := change(time(_res)) ? 1 : _initCnt + 1 f_valueAtIntrabar(_src, _bar, _res) => // Returns series value at intrabar n. First intrabar is 1, starting from the earliest. var int _barNo = 0 var float _value = na _barNo := change(time(_res)) ? 1 : _barNo + 1 _value := _barNo == _bar ? _src : _value // Returns close of last intrabar in "curRes" chart bar. lastClose = security(syminfo.tickerid, insideRes, close) // Returns volume at "insideBarNo" intrabar. valueAtIntrabar = security(syminfo.tickerid, insideRes, f_valueAtIntrabar(volume, insideBarNo, curRes)) // Returns qty of "insideRes" intrabars in "curRes" chart bar. qtyIntrabars = security(syminfo.tickerid, insideRes, f_qtyIntrabars(curRes)) plotchar(lastClose,"lastClose", "", location = location.top) plotchar(valueAtIntrabar,"valueAtIntrabar", "", location = location.top) plot(qtyIntrabars,"qtyIntrabars")
This is an example of a script that uses the technique illustrated in the functions to calculate delta volume.
Two steps are required:
alertcondition()call in an indicator script.
See the User Manual page on
alertcondition(). Code to create an alert condition looks like:
triggerCondition = close > close alertcondition(triggerCondition, title = "Create Alert dialog box name", message = "Text sent with alert.")
When you need to create multiple alerts you can repeat the method above for every alert you want your indicator to generate, but you can also use the method shown in this indicator. Here, all the different alert conditions are bunched up in one
alertcondition() statement. In this case, you must provide the means for users to first select which conditions will trigger the alert in the Inputs dialog box. When all the required conditions are selected, the user creates an alert using the only alert this indicator makes available, but since TradingView remembers the state of the Inputs when creating an alert, only the selected conditions will trigger the alert once it’s created, even if Inputs selections are modified by the user after the alert is created.
When more than one condition can trigger a single alert, you will most probably need to have visual cues for each condition so that when users bring up a chart on which an alert triggered they can figure out which condition caused the alert to trigger. This is a method that allows users of your script to customize the alert to their needs.
When TradingView creates an alert, it saves a snapshot of the environment that will enable the alert to run on the servers. The elements saved with an alert are:
Note that while alert condition code will compile in strategy scripts, alerts are only functional in studies.
You need to create a separate alert for each symbol. There is currently no way to create an alert for all the symbols in a watchlist or for the Screener.
If one of the generic indicators supplied with the Screener suits your needs and your symbols are tagged with a color label, you can create an alert on those markets from within the Screener.
The string may vary conditionally, but it must be of type const string, which implies it must be known at compile time.
This requirement entails that neither the condition used to build the string nor values used to calculate the string itself can depend on:
close > open,
The first step when you are in doubt as to what can be used as an argument to a built-in function such as
alertcondition() is to look up the Reference Manual:
You now know that a const string is required as an argument.
The next step is to consult the automatic type casting rules diagram in the User Manual’s Type system page:
The diagram shows you where the const string type is situated in the casting rules, which allows you to determine:
This code shows examples that work and don’t work:
//@version=4 study("alertcondition arguments") // ————— These strings will not work. // The rsi() value can only be known at runtime time and it is a "series", // so "wrongMsgArg1" becomes a "series string". wrongMsgArg1 = "RSI value is:" + tostring( rsi(close, 14)) // This does not work because although the result can be calculated at compile time, // "tostring()" returns a "simple string" (a.k.a. "string"), // and automatic casting rules do not allow for that type to be cast to "const string". wrongMsgArg2 = "Enter at: " + tostring(100.3) // This fails because the condition can only be evaluated at compile time, // so the result of the ternary is a "series string". wrongMsgArg3 = close > open ? "Long Entry" : "Short Entry" // ————— These strings will work because: // ————— 1. They can be evaluated at compile time, // ————— 2. Their type is "literal string" or "const string". // Test condition "false" is known at compile time and result of ternary is a "const string". goodMsgArg1 = false ? "Long Entry" : "Short Entry" // Both values in the expression are literal strings known at compile time. Result is "const string". goodMsgArg2 = "AAA " + "BBB" alertcondition(true, title="Id appearing in Create Alert db", message = goodMsgArg1)
From the Pine Editor, go to the New menu and select the built-in you want to work with. Note that some built-ins like the three Volume Profile and the Volume indicators are not written in Pine and their behavior cannot be reproduced in Pine.
Use the CTRL-⌥ + ` (grave accent) keyboard shortcut or right click on the script’s name and choose Show Console.
With the script open in the editor, choose the Convert to v4 button at the upper right of the editor window, to the right of the Save button.
Since v4 the
var keyword provides a way to initialize variables on the first bar of the dataset only, rather than on every bar the script is run on, as was the case before. This has the very useful benefit of automatically taking care of the value’s propagation throughout bars:
//@version=4 study("Variable Initialization") // Initialization at first bar (bar_index=0) only. Value is propagated across bars. var initOnce = 0 initOnce := initOnce + 1 // Initialization at each bar. Value is not propagated across bars. initOnEachBar1 = 0 initOnEachBar1 := initOnEachBar1 + 1 // Initialization at each bar. Value is not propagated across bars, // so we must refer to the variable's previous value in the series, // while allowing for the special case on first bar where there is no previous value. initOnEachBar2 = 0 initOnEachBar2 := nz(initOnEachBar2) + 1 plot(initOnce, "initOnce", color.blue, 10) plot(initOnEachBar1, "initOnEachBar1", color.red) plot(initOnEachBar2, "initOnEachBar2", color.orange, 3, transp = 0)
(val1 + val2)/2. Note that
avg()accepts up to 6 values.
This script shows how to calculate a conditional average using three different methods.
See the discussion published with the PineCoders indicator How to avoid repainting when using security().
The easiest way is to use the following syntax for v4:
security(syminfo.tickerid, "D", close, lookahead = barmerge.lookahead_on)
And this for v3:
security(tickerid, "D", close, lookahead = barmerge.lookahead_on)
See the discussion published with the PineCoders indicator How to avoid repainting when NOT using security().
The general idea is to use the confirmed information from the last bar for calculations.
//@version=4 study("", overlay = true) len = input(3) cond = close > open and close > open trigger = cond and barssince(cond) > len - 1 plotchar(cond) plotchar(trigger, "", "O", color = color.red)
Use everget’s Chart Type Identifier.
Notice how we take advantage of the fact that script execution begins at the first bar of the dataset and executes once for each successive bar. By working this way we don’t need a
for loop to go inspect past bars, as our script is already running in a sort of giant loop taking it on each of the dataset’s bars, from the oldest to the realtime bar. Scripts with calculations structured in the following way will execute much faster than ones using
//@version=4 study("Plot history's high and low", "", true) var hi = 0. var lo = 10e20 hi := max(hi, high) lo := min(lo, low) plot(hi, trackprice = true) plot(lo, trackprice = true)
Also note that we are using the
var keyword to initialize variables only once on the first bar of the dataset. This results in the variable’s value being automatically propagated throughout bars so we don’t have to use the equivalent of what was necessary in v3 to fetch the value of the variable from the previous bar:
//@version=3 study("Plot history's high and low", "", true) hi = 0. lo = 10e20 hi := max(nz(hi), high) lo := min(nz(lo), low) plot(hi, trackprice = true) plot(lo, trackprice = true)
barssince() built-in function is the simplest way of doing it, as is done in Method 1 in the following script. Method 2 shows an alternate way to achieve the same result as
barssince(). In Method 2 we watch for the condition as the script is executing on each successive bar, initialize our distance to 0 when we encounter the condition, and until we encounter the condition again, add 1 to the distance at each bar.
In either case the resulting value can be used as an index with the
 history-referecing operator.
//@version=4 study("Track distance from condition", "", true) // Plot the high/low from bar where condition occurred the last time. // Conditions. upBar = close > open dnBar = close < open up3Bars = dnBar and upBar and upBar and upBar dn3Bars = upBar and dnBar and dnBar and dnBar // Method 1, using "barssince()". plot(high[barssince(up3Bars)], linewidth = 10, transp = 80) plot(low[barssince(dn3Bars)], color = color.red, linewidth = 10, transp=80) // Method 2, doing manually the equivalent of "barssince()". var barsFromUp = 0 var barsFromDn = 0 barsFromUp := up3Bars ? 0 : barsFromUp + 1 barsFromDn := dn3Bars ? 0 : barsFromDn + 1 plot(high[barsFromUp]) plot(low[barsFromDn], color = color.red) plotchar(barsFromUp, "barsFromUp", "", location.top) plotchar(barsFromDn, "barsFromDn", "", location.top)
This script shows how to keep track of the number of bars since the last cross using the same two methods:
//@version=4 study("Bars between crosses", "", true) maS = sma(close,30) maF = sma(close,5) masCross = cross(maF, maS) // ————— Count number of bars since last crossover: manually or using built-in function. var barCount1 = 0 barCount1 := masCross ? 0 : barCount1 + 1 barCount2 = barssince(masCross) // ————— Plots label.new(bar_index, high + tr, "barCount1: " + tostring(barCount1) + "\nbarCOunt2: " + tostring(barCount2), xloc.bar_index, yloc.price, size = size.small) plot(maF) plot(maS, color = color.fuchsia)
This code shows how to do that without using
security() calls, which slow down your script. The source used to calculate the highs/lows can be selected in the script’s Inputs, as well as the period after which the high/low must be reset.
//@version=4 //@author=LucF, for PineCoders study("Periodic hi/lo", "", true) showHi = input(true, "Show highs") showLo = input(true, "Show lows") srcHi = input(high, "Source for Highs") srcLo = input(low, "Source for Lows") period = input("D", "Period after which hi/lo is reset", input.resolution) var hi = 10e-10 var lo = 10e10 // When a new period begins, reset hi/lo. hi := change(time(period)) ? srcHi : max(srcHi, hi) lo := change(time(period)) ? srcLo : min(srcLo, lo) plot(showHi ? hi : na, "Highs", color.blue, 3, plot.style_circles) plot(showLo ? lo : na, "Lows", color.fuchsia, 3, plot.style_circles)
We use the intrabar inspection technique explained here to inspect intrabars and save the high or low if the intrabar is whithin user-defined begin and end times.
//@version=4 //@author=LucF, for PineCoders study("Pre-market high/low", "", true) begHour = input(7, "Beginning time (hour)") begMinute = input(0, "Beginning time (minute)") endHour = input(9, "End time (hour)") endMinute = input(25, "End time (minute)") // Lower TF we are inspecting. Cannot be in seconds and must be lower that chart's resolution. insideRes = input("5", "Intrabar resolution used") startMinute = (begHour * 60) + begMinute finishMinute = (endHour * 60) + endMinute f_highBetweenTime(_start, _finish) => // Returns low between specific times. var float _return = 0. var _reset = true _minuteNow = (hour * 60) + minute if _minuteNow >= _start and _minuteNow <= _finish // We are inside period. if _reset // We are at first bar inside period. _return := high _reset := false else _return := max(_return, high) else // We are past period; enable reset for when we next enter period. _reset := true _return f_lowBetweenTime(_start, _finish) => // Returns low between specific times. var float _return = 10e10 var _reset = true _minuteNow = (hour * 60) + minute if _minuteNow >= _start and _minuteNow <= _finish // We are inside period. if _reset // We are at first bar inside period. _return := low _reset := false else _return := min(_return, low) else // We are past period; enable reset for when we next enter period. _reset := true _return highAtTime = security(syminfo.tickerid, insideRes, f_highBetweenTime(startMinute, finishMinute)) lowAtTime = security(syminfo.tickerid, insideRes, f_lowBetweenTime(startMinute, finishMinute)) plot(highAtTime, "High", color.green) plot(lowAtTime, "Low", color.red)
sum() function is the most efficient way to do it, but its length (the number of last bars in your sample) cannot be a series float or int. This script shows three different ways of achieving the count:
forloop to go back on past bars at every bar. Examining all length bars at every bar is unnecessary since all of them except the last bar have already been examined previously when the script first executed on them. This makes for slower code and will be detrimental to chart loading time.
Method 2 is a very good example of the Pine way of doing calculations by taking advantage of series and a good understanding of the Pine runtime environment to code our scripts. While it is useful to count occurrences of a condition in the last x bars, it is also worth studying because the technique it uses will allow you to write much more efficient Pine code than one using a
for loop when applied to other situations. There are situations when using a
for loop is the only way to realize what we want, but in most cases they can be avoided.
for loops are the only way to achieve some types of backward analysis when the criteria used are only known after the bars to analyze have elapsed.
//@version=4 //@author=LucF, for PineCoders // TimesInLast - PineCoders FAQ // v1.0, 2019.07.15 19:37 — Luc // This script illustrates 3 different ways of counting the number of occurrences when a condition occured in the last len bars. // By using the script's Settings/Inputs you can choose between 4 types of length to use with the functions. // If you look at results in the Data Window, you will see the impact of sending different types of length to each of the functions. // Conclusions: // - Unless your length is of series type, use Method 1. // - Use Method 2 if you need to be able to use a series int or series float length. // - Never use Method 3. study("TimesInLast - PineCoders FAQ") // Change this value when you want to use different lengths. // Inputs cannot be change through Settings/Inputs; only the form-type. deflen = 100 // ————— Allow different types to be specified as length value. // This part is only there to show the impact of using different form-types of length with the 3 functions. // In normal situation, we would just use the following: len = input(100, "Length") LT1 = "1. input int", LT2 = "2. input float", LT3="3. series int", LT4="4. series float" lt = input(LT1, "Type of 'length' argument to functions", options=[LT1, LT2, LT3, LT4]) len1 = input(deflen, LT1, type=input.integer, minval=deflen, maxval=deflen) len2 = input(deflen, LT2, type=input.float, minval=deflen, maxval=deflen) var len3 = 0 len3 := len3 == deflen ? len3 : len3 + 1 var len4 = 0. len4 := len4 == deflen ? len4 : len4 + 1 // Choose proper form-type of length. len = lt == LT1 ? len1 : lt == LT2 ? len2 : lt == LT3 ? len3 : lt == LT4 ? len4 : na // Condition on which all counts are done. condition = close > open // ————— Method 1. This function uses Pine's built-in function but only accepts a simple int for the length. f_ideal_TimesInLast(_cond, _len) => sum(_cond ? 1 : 0, _len) // ————— Method 2. This function is equivalent to using sum() but works with a float and series value for _len. f_verboseButEfficient_TimesInLast(_cond, _len) => // For first _len bar we just add to cumulative count of occurrences. // After that we add count for current bar and make adjustment to count for the tail bar in our mini-series of length=_len. var _qtyBarsInCnt = 0 var _cnt = 0 if _cond // Add to count as per current bar's condition state. _cnt := _cnt + 1 if _qtyBarsInCnt < _len // We have not counted the first _len bars yet; keep adding to checked bars count. _qtyBarsInCnt := _qtyBarsInCnt + 1 else // We already have a _len bar total, so need to subtract last count at the tail of our _len length count. if _cond[_len] _cnt := _cnt - 1 _qtyBarsInCnt == _len ? _cnt : na // Use this to return na until first _len bars have elapsed, as built-in "sum()" does. // _cnt // Use this when you want the running count even if full _len bars haven't been examined yet. // ————— Method 3. Very inefficient way to go about the problem. Not recommended. f_verboseAndINEFFICIENT_TimesInLast(_cond, _len) => // At each bar we loop back _len-1 bars to re-count conditions that were already counted in previous calls, except for the current bar's condition. _cnt = 0 for _i = 0 to _len - 1 if na(_cond[_i]) _cnt := na else if _cond[_i] _cnt := _cnt + 1 _cnt // ————— Plots v1 = f_ideal_TimesInLast(condition, int(len)) v2 = f_verboseButEfficient_TimesInLast(condition, int(len)) v3 = f_verboseAndINEFFICIENT_TimesInLast(condition, int(len)) plot(v1, "1. f_ideal_TimesInLast", color.fuchsia) plot(v2, "2. f_verboseButEfficient_TimesInLast", color.orange) plot(v3, "3. f_verboseAndINEFFICIENT_TimesInLast") // Plot red background on discrepancies between results. bgcolor(v1 != v2 or v2 != v3 ? color.red : na, transp = 80)
//@version=4 study("On/Off condition", "", true) upBar = close > open // On/off conditions. triggerOn = upBar and upBar and upBar triggerOff = not upBar and not upBar // Switch state is implicitly saved across bars thanks to initialize-only-once keyword "var". var onOffSwitch = false // Turn the switch on when triggerOn is true. If it is already on, // keep it on unless triggerOff occurs. onOffSwitch := triggerOn or (onOffSwitch and not triggerOff) bgcolor(onOffSwitch ? color.green : na) plotchar(triggerOn, "triggerOn", "▲", location.belowbar, color.lime, 0, size = size.tiny, text = "On") plotchar(triggerOff, "triggerOff", "▼", location.abovebar, color.red, 0, size = size.tiny, text = "Off")
One way to do it is by using
barssince(). This way is more flexible and faster:
//@version=4 //@author=LucF, for PineCoders study("AB or BA", "", true) // ————— Trigger conditions. upBar = close > open condATrigger = upBar and upBar condBTrigger = not upBar and not upBar // ————— Conditions. These variable will only be true/false on the bar where they occur. condA = false condB = false // ————— State variable set to true when last triggered condition was A, and false when it was condition B. // This variable's state is propagated troughout bars (because we use the "var" keyword to declare it). var LastCondWasA = false // ————— State transitions so that we allow A►B or B►A, but not A►A nor B►B. if condATrigger and not LastCondWasA // The trigger for condA occurs and the last condition set was condB. condA := true LastCondWasA := true else if condBTrigger and LastCondWasA // The trigger for condB occurs and the last condition set was condA. condB := true LastCondWasA := false bgcolor(LastCondWasA ? color.green : na) plotchar(condA, "condA", "▲", location.belowbar, color.lime, 30, size = size.tiny, text = "A") plotchar(condB, "condB", "▼", location.abovebar, color.red, 30, size = size.tiny, text = "B") // Note that we do not plot the marker for triggers when they are allowed to change states, since we then have our condA/B marker on the chart. plotchar(condATrigger and not condA, "condATrigger", "•", location.belowbar, color.green, 0, size = size.tiny, text = "a") plotchar(condBTrigger and not condB, "condBTrigger", "•", location.abovebar, color.maroon, 0, size = size.tiny, text = "b")
The answer depends on whether you know the minimum/maximum possible values of the signal to be rescaled. If you don’t know them, as is the case for volume where the maximum is unknown, then you will need to use a function that uses past history to determine the minimum/maximum values, as in the
normalize() function here. While this is an imperfect solution since the minimum/maximum need to be discovered as your script progresses left to right through historical bars, it is better than techniques using
highest() over a fixed length, because it uses the minimum/maximum values for the complete set of elapsed bars rather than a subset of fixed length. The ideal solution would be to know in advance the minimum/maximum values for the whole series prior to beginning the normalization process, but this is currently not possible in Pine.
If you know the minimum/maximum values of the series, then you should use the
//@version=4 //@author=LucF, for PineCoders study("Normalizer") // ————— When scale of signal to rescale is unknown. // Min/Max of signal to rescale is determined by its historical low/high. normalize(_src, _min, _max) => // Normalizes series with unknown min/max using historical min/max. // _src: series to rescale. // _min: minimum value of rescaled series. // _max: maximum value of rescaled series. var _historicMin = 10e10 var _historicMax = -10e10 _historicMin := min(nz(_src, _historicMin), _historicMin) _historicMax := max(nz(_src, _historicMax), _historicMax) _min + (_max - _min) * (_src - _historicMin) / max(_historicMax - _historicMin, 10e-10) plot(normalize(volume, -100, 100)) // ————— When scale of signal to rescale is known. rescale(_src, _oldMin, _oldMax, _newMin, _newMax) => // Rescales series with known min/max. // _src: series to rescale. // _oldMin: minimum value of series to rescale. // _oldMax: maximum value of series to rescale. // _newMin: minimum value of rescaled series. // _newMax: maximum value of rescaled series. _newMin + (_newMax - _newMin) * (_src - _oldMin) / max(_oldMax - _oldMin, 10e-10) plot(rescale(rsi(close, 14), 0, 100, -100, 100), color = color.fuchsia)
Use the code from the PineCoders Script Stopwatch. You will be able to time script execution so you can explore different scenarios when developing code and see for yourself which version performs the best.
This code will show a label containing the current values of the variables you wish to see. Non-string variables need to be converted to strings using
tostring(). The label will show when price changes in the realtime bar, so the code needs to run on a live chart.
//@version=4 study("f_print()", "", true) f_print(_txt) => var _lbl = label(na), label.delete(_lbl), _lbl := label.new(time + (time-time)*3, high, _txt, xloc.bar_time, yloc.price, size = size.large) f_print("Multiplier = " + tostring(timeframe.multiplier) + "\nPeriod = " + timeframe.period + "\nHigh = " + tostring(high))
The solution is to use the
plotchar() function, but without actually printing a character, and using the fact that values plotted with
plotchar() will appear both:
The reason for using the
location = location.top parameter is that
location.abovebar as the default when the
location= parameter is not specified, and this puts price into play in your indicator’s scale, even if no character is actually plotted by
Note that you may use
plotchar() to test variables of string type, but only by comparing them to a single string, as is done in the second
plotchar() call in the following code:
//@version=4 study("Printing values with plotchar()") plotchar(bar_index, "Bar Index", "", location = location.top) // This will be true (1) when chart is at 1min. Otherwise it will show false (0). plotchar(timeframe.period == "1", "timeframe.period='1'", "", location = location.top)
1printed by the second call to
plotchar(), indicating that we are on a 1 min chart.
This code displays green or red squares corresponding to the two different states of four different conditions, and colors the background when they are either all true or all false:
//@version=4 study("Debugging states with plotshape() and bgcolor()") cond1 = close > open cond2 = close > close cond3 = volume > volume cond4 = high - close < open - low cond5 = cond1 and cond2 and cond3 and cond4 cond6 = not (cond1 or cond2 or cond3 or cond4) plotshape(9, "cond1", shape.square, location.absolute, cond1 ? color.green : color.red, size = size.tiny) plotshape(8, "cond2", shape.square, location.absolute, cond2 ? color.green : color.red, size = size.tiny) plotshape(7, "cond3", shape.square, location.absolute, cond3 ? color.green : color.red, size = size.tiny) plotshape(6, "cond4", shape.square, location.absolute, cond4 ? color.green : color.red, size = size.tiny) bgcolor(cond5 ? color.green : cond6 ? color.red : na, title = "cond5/6")
You could also use
plot() to achieve a somewhat similar result. Here we are plotting the condition number only when the condition is true:
//@version=4 study("Debugging states with plot() and bgcolor()") // ————— States cond1 = close > open cond2 = close > close cond3 = volume > volume cond4 = high - close < open - low cond5 = cond1 and cond2 and cond3 and cond4 cond6 = not (cond1 or cond2 or cond3 or cond4) plot(cond1 ? 1 : na, "cond1", linewidth = 4, style = plot.style_circles) plot(cond2 ? 2 : na, "cond2", linewidth = 4, style = plot.style_circles) plot(cond3 ? 3 : na, "cond3", linewidth = 4, style = plot.style_circles) plot(cond4 ? 4 : na, "cond4", linewidth = 4, style = plot.style_circles) bgcolor(cond5 ? color.green : cond6 ? color.red : na, title = "cond5/6")
When building compound conditions that rely on the accuracy of multiple underlying conditions used as building blocks, you will usually want to confirm your code is correctly identifying the underlying conditions. Here, markers identifying them are plotted at the top and bottom of the chart using
plotshape(), while the compound conditions 5 an 6 are marked above and below bars using
plotshape(), and one bar later using
plotchar() and a Unicode character:
//@version=4 study("Plotting markers with plotshape()", "", true) cond1 = close > open cond2 = close > close cond3 = volume > volume cond4 = high - close < open - low cond5 = cond1 and cond2 and cond3 and cond4 cond6 = not (cond1 or cond2 or cond3 or cond4) plotshape(cond1, "cond1", shape.circle, location.top, color.silver, text = "1", size = size.small) plotshape(cond2, "cond2", shape.diamond, location.top, color.orange, text = "2", size = size.tiny) plotshape(cond3, "cond3", shape.circle, location.bottom, color.fuchsia, text = "3", size = size.small) plotshape(cond4, "cond4", shape.diamond, location.bottom, color.aqua, text = "4", size = size.tiny) plotshape(cond5, "cond5", shape.triangleup, location.belowbar, color.green, 0, text = "cond5", size = size.tiny) plotshape(cond6, "cond6", shape.triangledown, location.abovebar, color.maroon, 0, text = "cond6", size = size.tiny) // Place these markers one bar late so they don't overprint the "plotshape()" triangles. plotchar(cond5, "cond5", "⮝", location.belowbar, color.lime, 0, size = size.tiny) plotchar(cond6, "cond6", "⮟", location.abovebar, color.red, 0, size = size.tiny)
You will find lists of Unicode arrows here and here. Because they are not all mapped in the MS Trebuchet font TV uses, not all characters will work with
plotchar(). Some work as arguments to the
text= parameter, but not as arguments to