I'm trying to achieve the following
- I want to count candles backwards for a specific timeframe (e.g. 1D) and only add the values to an array if they meet some criteria
- If I'm on a LTF chart (e.g. 1M, 5M, 15M) I want to make sure that my calculation results are unchanged.
- The idea is to then make a
request.security()call to get the criteria specific candle data (e.g. open[1], open[14], open[42])
I'm struggling a bit to implement this. My idea so far has been this, but it isn't working as expected.
step 1: Manually create the historical data / criteria (which is based on the economic calendar). Have to create manually since this isn't available to obtain within PineScript.
Step 2: Define the constants, variables and start the calculations. Basically check if today matches a historical event date and set these values to those variables. Then calculations.
Step 3: The first function calculates the timestamps for these event dates. The second one counts how many candles back this event happens to occur on (however this part isn't exactly correct)
Simplified code sample
//@version=6
indicator("Simplified Example", overlay = true, max_bars_back=1000)
// === User Inputs ===
tf_1day_enabled = input.bool(true, "D", "1 day", "row 1", "Timeframe")
// === Step 1: Historical Data Section ===
var int[] fomc_dates = array.new_int(0)
if bar_index == 0
array.push(fomc_dates, timestamp(2025, 3, 18, 0, 0))
array.push(fomc_dates, timestamp(2025, 2, 18, 0, 0)) // for testing purposes adjust to current date
array.push(fomc_dates, timestamp(2025, 1, 29, 0, 0))
array.push(fomc_dates, timestamp(2024, 12, 18, 0, 0))
array.push(fomc_dates, timestamp(2024, 11, 7, 0, 0))
// === Step 3: Functions Section ===
// Returns an array of past timestamp date values equal to the lookback period (e.g. lookback = 3, 3 values returned)
find_past_events(event_array, int lookback) =>
var int[] past_dates = array.new_int(0)
for i = 0 to array.size(event_array) - 1
if array.size(past_dates) < lookback and array.get(event_array, i) < timestamp(year, month, dayofmonth, hour, 0)
array.push(past_dates, array.get(event_array, i))
past_dates
// Returns an array of numbers showing how many candles counting back from current did the event news happen
find_matching_candles(event_array) =>
var int[] candles_back = array.new_int(0)
int i = 1
int j = 0
int max_allowable_candles = 1000
while i < max_allowable_candles and j < array.size(event_array) and array.size(candles_back) < array.size(event_array)
if dayofweek != dayofweek.saturday and dayofweek != dayofweek.sunday
historical_timestamp = timestamp(year, month, dayofmonth - i, 0, 0)
if historical_timestamp == array.get(event_array, j)
if array.size(candles_back) == 0 or array.get(candles_back, array.size(candles_back) - 1) < i
array.push(candles_back, i) // Add only if it is unique
j := j + 1
i := i + 1
candles_back
// === Step 2: Define Constants & variables & start calculations ===
calculation_lookback = 3
var string event_type = "normal"
var event_dates = array.new_int(0)
var past_candles_back = array.new_int(0)
if tf_1day_enabled
// Get past news event dates
if array.includes(fomc_dates, timestamp(year, month, dayofmonth, 0, 0))
event_type := "fomc"
event_dates := fomc_dates
past_event_dates = event_type != "normal" ? find_past_events(event_dates, calculation_lookback) : event_dates // Ignore normal days for now
past_candles_back := find_matching_candles(past_event_dates)
// log.info(str.tostring(past_candles_back)) // returns [1, 43, 107]
bgcolor(bar_index == 10 ? color.yellow : na) // plot something random for it to run
My issues are:
- This approach isn't calculating 1D candles backwards and it's estimating how many candles back it is based on time. Holidays and weekends have an affect when using time.
- The returned array I get is [1, 43, 107], but if I manually count the candles on the 1D chart... I should get [1, 14, 42]. I don't get why I have this large of a difference with counting backwards by 1 day (time).
- I feel that there is a better/easier way to do this and ideally actually counting the 1D candles. So I don't have to worry about weekends and holidays.
Can anyone help me with this?
I'm trying to achieve the following
- I want to count candles backwards for a specific timeframe (e.g. 1D) and only add the values to an array if they meet some criteria
- If I'm on a LTF chart (e.g. 1M, 5M, 15M) I want to make sure that my calculation results are unchanged.
- The idea is to then make a
request.security()call to get the criteria specific candle data (e.g. open[1], open[14], open[42])
I'm struggling a bit to implement this. My idea so far has been this, but it isn't working as expected.
step 1: Manually create the historical data / criteria (which is based on the economic calendar). Have to create manually since this isn't available to obtain within PineScript.
Step 2: Define the constants, variables and start the calculations. Basically check if today matches a historical event date and set these values to those variables. Then calculations.
Step 3: The first function calculates the timestamps for these event dates. The second one counts how many candles back this event happens to occur on (however this part isn't exactly correct)
Simplified code sample
//@version=6
indicator("Simplified Example", overlay = true, max_bars_back=1000)
// === User Inputs ===
tf_1day_enabled = input.bool(true, "D", "1 day", "row 1", "Timeframe")
// === Step 1: Historical Data Section ===
var int[] fomc_dates = array.new_int(0)
if bar_index == 0
array.push(fomc_dates, timestamp(2025, 3, 18, 0, 0))
array.push(fomc_dates, timestamp(2025, 2, 18, 0, 0)) // for testing purposes adjust to current date
array.push(fomc_dates, timestamp(2025, 1, 29, 0, 0))
array.push(fomc_dates, timestamp(2024, 12, 18, 0, 0))
array.push(fomc_dates, timestamp(2024, 11, 7, 0, 0))
// === Step 3: Functions Section ===
// Returns an array of past timestamp date values equal to the lookback period (e.g. lookback = 3, 3 values returned)
find_past_events(event_array, int lookback) =>
var int[] past_dates = array.new_int(0)
for i = 0 to array.size(event_array) - 1
if array.size(past_dates) < lookback and array.get(event_array, i) < timestamp(year, month, dayofmonth, hour, 0)
array.push(past_dates, array.get(event_array, i))
past_dates
// Returns an array of numbers showing how many candles counting back from current did the event news happen
find_matching_candles(event_array) =>
var int[] candles_back = array.new_int(0)
int i = 1
int j = 0
int max_allowable_candles = 1000
while i < max_allowable_candles and j < array.size(event_array) and array.size(candles_back) < array.size(event_array)
if dayofweek != dayofweek.saturday and dayofweek != dayofweek.sunday
historical_timestamp = timestamp(year, month, dayofmonth - i, 0, 0)
if historical_timestamp == array.get(event_array, j)
if array.size(candles_back) == 0 or array.get(candles_back, array.size(candles_back) - 1) < i
array.push(candles_back, i) // Add only if it is unique
j := j + 1
i := i + 1
candles_back
// === Step 2: Define Constants & variables & start calculations ===
calculation_lookback = 3
var string event_type = "normal"
var event_dates = array.new_int(0)
var past_candles_back = array.new_int(0)
if tf_1day_enabled
// Get past news event dates
if array.includes(fomc_dates, timestamp(year, month, dayofmonth, 0, 0))
event_type := "fomc"
event_dates := fomc_dates
past_event_dates = event_type != "normal" ? find_past_events(event_dates, calculation_lookback) : event_dates // Ignore normal days for now
past_candles_back := find_matching_candles(past_event_dates)
// log.info(str.tostring(past_candles_back)) // returns [1, 43, 107]
bgcolor(bar_index == 10 ? color.yellow : na) // plot something random for it to run
My issues are:
- This approach isn't calculating 1D candles backwards and it's estimating how many candles back it is based on time. Holidays and weekends have an affect when using time.
- The returned array I get is [1, 43, 107], but if I manually count the candles on the 1D chart... I should get [1, 14, 42]. I don't get why I have this large of a difference with counting backwards by 1 day (time).
- I feel that there is a better/easier way to do this and ideally actually counting the 1D candles. So I don't have to worry about weekends and holidays.
Can anyone help me with this?
Share Improve this question asked 2 days ago AlwaysBeLearningAlwaysBeLearning 451 silver badge5 bronze badges 3 |1 Answer
Reset to default 0In case someone gets stuck with wanting to count candles backwards on a HTF (which is flawed logic in PineScript), I figured out how to solve my problem correctly.
Original problem re-stated
How do I get specific candle data from a HTF chart for select, past dates (in my case based on economic calendar news dates), so I can perform some calculations on them?
Solution
It is actually really easy when you think that PineScript runs a continuous loop from the first candles available until the latest ones. The trick is to pass a function into the request.security() call to perform the (naturally looping) calculations on the HTF chart.
This is a simplified version of how I ended up solving my problem.
//@version=6
indicator("Simplified Example", overlay = true)
fomc_date = timestamp(2025, 1, 29, 0, 0)
find_candles_on_htf() =>
var open_prices = array.new_float(0)
if timestamp(year, month, dayofmonth) == fomc_date
// do calculations or get price info & store in local variable
array.push(open_prices, open)
open_prices
open_price_htf = request.security(syminfo.tickerid, "D", find_candles_on_htf())
// then you can use `open_price_htf` on LTF charts for whatever you need
security()and store it somewhere. Then apply your conditions to that array and not try to call thesecurity()function conditionally. – vitruvius Commented yesterday[]is not an index value, but essentially past bars. And PS is looping in nature. The thing that trips me up is looping 1 bar on the LTF does not match 1 bar on the HTF. So I get all available 1D data fromsecurity(). Then how does one process that historical data on the LTF? – AlwaysBeLearning Commented yesterdaysecurity()when you know exactly how many candles back. If you request everything...then how do you loop through the HTF open values on a LTF (because the LTF chart doesn't go that far back candle wise)? – AlwaysBeLearning Commented yesterday