How is Performance calculated in the Screener?

Screener Performance data is calculated using the formula:

//@version=5
indicator(title="Screener Performance")

rateOfreturn(ref) =>
	if ref < 0 and close > 0
		na
	else
		(close - ref) * 100 / math.abs(ref)

rr(bb, maxbarsback) =>
	open[maxbarsback] * 0 + bb == 0 ? na : rateOfreturn(open[bb])

perfYTD() =>
	var lastYearOpen = open
	if year > year[1]
		lastYearOpen := open
	rateOfreturn(lastYearOpen)

fastSearchTimeIndex(x, maxbarsback) =>
	mid = 0 * time[maxbarsback]
	right = math.min(bar_index, maxbarsback)
	left = 0
	if time < x
		0
	else
		for i = 0 to 9 by 1
			mid := math.ceil((left + right) / 2)
			if left == right
				break
			else if time[mid] < x
				right := mid
				continue
			else if time[mid] > x
				left := mid
				continue
			else
				break
		mid

week1 = 7
week_ago = timenow - 1000 * 60 * 60 * 24 * week1
week_ago_this_bar = time - 1000 * 60 * 60 * 24 * week1
countOfBarsWeekAgo = fastSearchTimeIndex(week_ago, week1)

month1 = 30
month_ago = timenow - 1000 * 60 * 60 * 24 * month1
countOfBars1MonthAgo = fastSearchTimeIndex(month_ago, month1)

month3 = 90
months3_ago = timenow - 1000 * 60 * 60 * 24 * month3
countOfBars3MonthAgo = fastSearchTimeIndex(months3_ago, month3)

month6 = 180
months6_ago = timenow - 1000 * 60 * 60 * 24 * month6
countOfBars6MonthAgo = fastSearchTimeIndex(months6_ago, month6)

years1 = 365
oneYearAgo = timenow - 1000 * 60 * 60 * 24 * years1
barsCountOneYear = fastSearchTimeIndex(oneYearAgo, years1)

years3 = 365 * 3
years3_ago = timenow - 1000 * 60 * 60 * 24 * years3
countOfBars3YearAgo = fastSearchTimeIndex(years3_ago, years3)

years5 = 365 * 4 + 366
years5_ago = timenow - 1000 * 60 * 60 * 24 * years5
countOfBars5YearAgo = fastSearchTimeIndex(years5_ago, years5)

years10 = (365 * 4 + 366) * 2
years10_ago = timenow - 1000 * 60 * 60 * 24 * years10
countOfBars10YearAgo = fastSearchTimeIndex(years10_ago, years10)

perfYTD = perfYTD()
plot((close - open[4]) / open[4] * 100, title='Perf.5D')
plot(rr(countOfBarsWeekAgo, week1), title='Perf.W')
plot(rr(countOfBars1MonthAgo, month1), title='Perf.1M')
plot(rr(countOfBars3MonthAgo, month3), title='Perf.3M')
plot(rr(countOfBars6MonthAgo, month6), title='Perf.6M')
plot(rr(barsCountOneYear, years1), title='Perf.Y')
plot(rr(countOfBars3YearAgo, years3), title='Perf.3Y')
plot(rr(countOfBars5YearAgo, years5), title='Perf.5Y')
plot(rr(countOfBars10YearAgo, years10), title='Perf.10Y')
plot(perfYTD, title='Perf.YTD')

Note: this script values are different on history and realtime because of timenow, see https://www.tradingview.com/pine-script-docs/en/v4/essential/Indicator_repainting.html

For visual displays, you can add this script to your chart through the Pine Editor using the chart's daily timeframe. An indicator will appear on the chart, the plots of which will show the values for each type of performance.

Change % vs Performance %:

Let's say today is Tuesday. Weekly Change calculates the difference between the current close (Tuesday) and the close from the last week (the closing price of the previous Friday). Weekly Performance calculates the difference between the current close (Tuesday) and the open from a week ago exactly (the previous Tuesday).