PatternsRadar
Sign in

Language reference

Sift

A query language for candles. SQL’s structure and English’s readability, over 3,829 NSE instruments with daily history to 20 Nov 2026.

Try a scan144 worked examples210 fields, indicators and patterns

Overview#

The common case is one line with no ceremony. Everything else is optional.

close > ema(21)
Run

That is a complete query. It reads the latest bar of every stock in the universe and keeps the ones closing above their 21-day exponential moving average. There is no timeframe to declare, no latest to repeat, and no wrapper.

Compare the same three intents against Chartink, whose syntax makes you restate the timeframe on every single term:

IntentChartinkSift
Close above the 20 EMA( {cash} ( latest close > latest ema( latest close , 20 ) ) )close > ema(21)
Volume twice its 20-bar average( {cash} ( latest volume > 2 * latest sma( latest volume , 20 ) ) )volume > 2x avg(volume, 20)
RSI crossed 60 in the last 3 barsnot expressiblersi(14) crossed above 60 within 3 bars
Within 2% of the 52-week highmanual max plus arithmeticclose within 2% of high_52w

Sift is not Turing complete: no loops, no user-defined functions, no side effects. Every query is statically analysable, which is what lets the editor underline a mistake before you run anything and lets the engine bound the cost of a scan before it starts.

Query shape#

Five clauses, all optional, conventionally in this order.

on <timeframe>
from <universe>
where <condition>
sort by <expression> [asc | desc]
top <n>

A bare condition is a complete query — where is optional. Clauses may appear in any order. Newlines and indentation carry no meaning, and # or // starts a comment that runs to the end of the line.

on daily
from NSE
where close > ema(21) > ema(50)      # a trend in good order
  and volume > 2x avg(volume, 20)
  and delivery_pct > 55
sort by turnover desc
top 25
Run
ClausePurposeDefault
onCandle timeframe.daily — the only one this dataset carries
fromWhich universe to scan: `NSE` (all equities) or `fno` (stocks with listed futures & options).NSE
whereThe condition. Optional keyword.everything matches
sort byOrder the results by any expression.turnover, descending
topRow cap.300
sort by accepts any expression, not just a returned column — sort by (close - ema(50)) / ema(50) desc ranks by distance above the average.

Fields#

Written bare, with no parentheses. Each maps to one precomputed column.

Price

FieldMeaning
closeClosing price, adjusted for splits and bonuses.
openOpening price.
highSession high.
lowSession low.
hl2Midpoint of the session range.
hlc3Typical price.
ohlc4Average of all four session prices.
ha_openHeikin-Ashi open — the average of the previous HA open and close.
ha_highHeikin-Ashi high.
ha_lowHeikin-Ashi low.
ha_closeHeikin-Ashi close — the average of the bar's four prices.
high_52wHighest high of the last 252 sessions. The same as `highest_high(1y)`; write `highest_high(N)` for any other window.
low_52wLowest low of the last 252 sessions. The same as `lowest_low(1y)`; write `lowest_low(N)` for any other window.
pct_from_52w_highDistance below the 52-week high, as a negative percentage. The same as `pct_from_high(1y)`; write `pct_from_high(N)` for any other window.
pct_from_52w_lowDistance above the 52-week low, as a percentage. The same as `pct_from_low(1y)`; write `pct_from_low(N)` for any other window.
pivotClassic floor-trader pivot from the previous session.
pivot_r1First resistance above the pivot.
pivot_r2Second resistance above the pivot.
pivot_s1First support below the pivot.
pivot_s2Second support below the pivot.

Volume & delivery

FieldMeaning
volumeShares traded, restated across splits and bonuses like close, so close × volume is the rupees traded.
turnoverNSE's traded value where known, else close × volume, in rupees.
tradesNumber of trades executed.
delivery_pctShare of volume taken to demat rather than squared off intraday. High delivery on a rising day suggests genuine accumulation.
delivery_qtyShares taken to demat, restated across splits and bonuses like close.
rel_volumeVolume divided by its own 20-day average. 2 means twice normal.
obvRunning total of volume signed by the day's direction.
acc_distRunning total weighted by where the close sits in the range.
cmfAccumulation/distribution normalised over 20 bars. Positive means buying pressure.
force_indexPrice change times volume, smoothed over 13 bars.

Momentum

FieldMeaning
stoch_kSlow stochastic %K over 14 bars, smoothed by 3.
stoch_d3-bar average of stochastic %K.
stoch_rsiThe stochastic oscillator applied to RSI itself.
cciCommodity Channel Index over the typical price.
williams_rLike the stochastic, scaled −100 to 0.
mfiA volume-weighted RSI over the typical price.
rocPercentage change over 10 bars.

Trend

FieldMeaning
adxTrend strength, direction-agnostic. Above 25 is usually read as trending.
di_plusPositive directional indicator.
di_minusNegative directional indicator.
supertrendATR-banded trailing stop (10, 3).
supertrend_dir+1 while Supertrend is bullish, −1 while bearish.
psarParabolic stop-and-reverse (0.02, 0.2).
aroon_upHow recently the 25-bar high occurred.
aroon_downHow recently the 25-bar low occurred.
aroon_oscAroon up minus Aroon down.
ichimoku_conversionTenkan-sen, the 9-bar midpoint.
ichimoku_baseKijun-sen, the 26-bar midpoint.
ichimoku_span_aCloud edge A, unshifted.
ichimoku_span_bCloud edge B, unshifted.

Volatility

FieldMeaning
atrWilder's average true range over 14 bars.
true_rangeThis bar's true range.
bb_upper20-period SMA plus two standard deviations.
bb_mid20-period simple moving average.
bb_lower20-period SMA minus two standard deviations.
bb_pct_bWhere the close sits within the bands: 0 at the lower, 1 at the upper.
bb_widthBand separation as a fraction of the middle band. Low means a squeeze.
donchian_upperHighest high of the last 20 bars.
donchian_midMidpoint of the Donchian channel.
donchian_lowerLowest low of the last 20 bars.
keltner_upper20-EMA plus two ATRs.
keltner_lower20-EMA minus two ATRs.

Performance

FieldMeaning
changePercentage change against the previous close. The same as `return(1)`; write `return(N)` for any other span.
return_1wPercentage change over 5 sessions.
return_1mPercentage change over 21 sessions.
return_3mPercentage change over 63 sessions.
return_6mPercentage change over 126 sessions.
return_1yPercentage change over 252 sessions. The same as `return(1y)`; write `return(N)` for any other span, e.g. `return(3y)`.

Derivatives (F&O)

FieldMeaning
fut_oiTotal futures open interest across all expiries, in shares (index units for an index), restated onto today's split-adjusted share basis like close, so a split or bonus does not read as an OI jump.
fut_oi_valueFutures open interest times the cash price, in rupees — the size of the open position book, comparable across stocks. Write `fut_oi_value > 5000cr`.
fut_oi_change_pctDay-over-day change in total futures OI. Read it with price: price up with OI up is long buildup, price down with OI up is short buildup. The drop after an expiry day is genuine, not noise.
fut_volumeFutures contracts traded across all expiries.
fut_basis_pctFront-month futures premium (positive) or discount (negative) to the cash price, in percent. Compared on the raw price scale — corporate actions are bridged out. Reads near zero on expiry day by construction.
fut_rollover_pctShare of futures OI already sitting in later expiries. On expiry day this is the classic rollover number the derivatives desks quote.
pcr_oiPut OI divided by call OI across this stock's options, all expiries. Above 1 means more open puts than calls. NULL when no calls are open.
pcr_volPut contracts traded divided by call contracts traded, all expiries.
pcr_oi_frontPut OI divided by call OI on the nearest expiry only — weekly for the indices, monthly for stocks. On expiry day it reads the next expiry, not the one settling. NULL when no front-expiry calls are open.
max_painThe front-expiry strike at which option holders would collect the least if the stock settled there — the price option writers as a group would most like to see at expiry. On the same adjusted scale as `close`, so `close < max_pain` compares like with like.
max_pain_dist_pctHow far max pain sits from the close, in percent: positive when max pain is above the price. Write `abs(max_pain_dist_pct) > 5` for a price that has run well away from it.
call_oi_wallThe strike at or above the price with the most open call interest on the front expiry — where call writers have staked the most on the price staying below. Often read as resistance. Adjusted scale, like `close`.
put_oi_wallThe strike at or below the price with the most open put interest on the front expiry — where put writers have staked the most on the price staying above. Often read as support. Adjusted scale, like `close`.
nifty_pcr_oiThe Nifty 50 options' put-call ratio by OI, the same on every stock for a date: `pcr_oi` read off the index. Lets a stock scan ask about market positioning, e.g. `nifty_pcr_oi > 1.2`.
banknifty_pcr_oiThe Nifty Bank options' put-call ratio by OI, the same on every stock for a date.

Fundamentals

FieldMeaning
marketcapClose times shares outstanding, in rupees — write `marketcap > 5000cr`. The share count is restated onto the adjusted-price basis, so a split does not fake a jump.
ffmcMarket cap of the shares outside the promoter group — marketcap times (100 − promoter_pct) / 100, on the latest pattern filed by that date. What institutions can actually buy: write `ffmc > 1000cr`. NSE's own figure also excludes locked-in and strategic stakes, so a fresh IPO reads higher here than on NSE; it has no history, this one does.
pePrice to trailing-twelve-month earnings attributable to owners (excluding minority interest), from the last four filed quarters as known on that date. NULL when TTM earnings are negative or not yet filed — a loss-maker has no P/E rather than a misleading one.
eps_ttmTrailing-twelve-month earnings per share on the owners' share of profit (excluding minority interest), split-adjusted to match the adjusted price series (filed EPS is not — it is never used directly). Negative for loss-makers, NULL until four consecutive quarters are on file.
revenue_ttmRevenue from operations over the last four filed quarters, in rupees — write `revenue_ttm > 500cr`. NULL until four consecutive quarters are on file.
net_profit_ttmNet profit attributable to owners (excluding minority interest) over the last four filed quarters, in rupees — the denominator of P/E. The total profit stands in where a filing gives no owners' share. Negative for loss-makers.
net_margin_ttmTrailing-twelve-month net profit over trailing-twelve-month revenue. Four quarters smooth out the seasonality a single quarter's margin carries. Negative for loss-makers.
other_income_shareOther income (interest, dividends, investment gains) as a share of profit before tax, both over the trailing twelve months. High readings flag profit propped up by treasury income rather than the business; above 100 the operations lost money. NULL when PBT is not positive, and for banks, NBFCs and insurers, where that income is the business.
revenue_growth_yoyLatest filed quarter's revenue against the same quarter last year.
profit_growth_yoyLatest filed quarter's net profit against the same quarter last year. NULL when the base quarter was a loss — growth from negative earnings is not a number.
revenue_growth_qoqLatest filed quarter's revenue against the previous quarter.
profit_growth_qoqLatest filed quarter's net profit against the previous quarter. NULL when the base quarter was a loss.
profit_cagr_2yAnnualized growth in trailing-twelve-month net profit over two years — the shortest window where compounding says anything a single YoY does not.
profit_cagr_3yAnnualized growth in trailing-twelve-month net profit over three years. Wider coverage than the five-year window, and long enough to outlast one soft base year.
profit_cagr_4yAnnualized growth in trailing-twelve-month net profit over four years.
profit_cagr_5yAnnualized growth in trailing-twelve-month net profit over five years — the usual test of whether earnings compound or merely cycle. A company needs five unbroken years of filings on one basis to get a number; about half of covered names do.
profit_cagr_6yAnnualized growth in trailing-twelve-month net profit over six years. Sparse — few names have this much filed history yet.
profit_cagr_7yAnnualized growth in trailing-twelve-month net profit over seven years — the longest window the results store reaches, and the sparsest.
revenue_cagr_2yAnnualized growth in trailing-twelve-month revenue over two years.
revenue_cagr_3yAnnualized growth in trailing-twelve-month revenue over three years.
revenue_cagr_4yAnnualized growth in trailing-twelve-month revenue over four years.
revenue_cagr_5yAnnualized growth in trailing-twelve-month revenue over five years. Pair it with `profit_cagr_5y` to separate operating leverage from growth that only arrived through the top line.
revenue_cagr_6yAnnualized growth in trailing-twelve-month revenue over six years.
revenue_cagr_7yAnnualized growth in trailing-twelve-month revenue over seven years.
interest_cost_growth_yoyTrailing-twelve-month finance costs against a year earlier. This is a proxy for the direction of borrowing, not a measure of it — the debt itself is `debt_to_equity` and `net_debt`, which move only twice a year with the balance sheet, while this moves every quarter. Flat interest beside growing profit is the shape of growth funded from earnings, and a jump is the shape of fresh debt. Exclude lenders (for a bank interest is the cost of goods, not leverage), and expect NULL when a quarter's finance-cost line is missing.
promoter_pctPromoter shareholding from the latest pattern filed by that date.
public_pctPublic shareholding from the latest pattern filed by that date.
fii_pctForeign institutional holding from the latest pattern filed by that date. Sparse until the shareholding XBRL backfill completes.
dii_pctDomestic institutional holding from the latest pattern filed by that date. Sparse until the shareholding XBRL backfill completes.
promoter_pledged_pctShare of the promoter stake pledged as collateral. Zero is the healthy reading; a rising number is the classic distress tell.
promoter_change_qoqPercentage-point change in promoter holding against the previous quarter's pattern. Positive means promoters bought.
pbMarket cap over the equity attributable to shareholders on the latest filed balance sheet. NULL when book value is negative. The balance sheet is half-yearly, from Sep-2022.
roeReturn on equity: owners' net profit over the twelve months to the latest balance sheet, divided by owners' equity (equity_owners) averaged with the year before (the latest alone when there is no year-ago sheet). Negative for a loss year; NULL when equity is negative. Holds for banks and NBFCs too.
roceReturn on capital employed: profit before tax plus finance costs over the twelve months to the latest balance sheet, divided by equity plus borrowings (averaged with the year before). NULL for banks, NBFCs and insurers, whose borrowings are their raw material rather than capital.
debt_to_equityCurrent plus non-current borrowings over total equity on the latest balance sheet. Zero for a debt-free company. NULL for banks, NBFCs and insurers, where borrowing is the business, and when equity is negative.
net_debtBorrowings less cash, other bank balances and current investments, in rupees — write `net_debt < 0` for net cash or `net_debt > 1000cr`. NULL for banks, NBFCs and insurers.
current_ratioCurrent assets over current liabilities on the latest balance sheet. Below 1 means short-term obligations exceed short-term assets. NULL for lenders and insurers, whose statements have no current split.
fcf_ttmOperating cash flow less capex (purchases of fixed and intangible assets) over the trailing twelve months, in rupees. Cash flows are filed year-to-date twice a year, so at a half-year the year is this half plus the last full year less last year's half. NULL for banks, NBFCs and insurers, where operating cash flow tracks the loan book.
fcf_yieldTrailing-twelve-month free cash flow over market cap — the cash counterpart of an earnings yield. Negative when the company burns cash.
cash_conversionOperating cash flow over net profit for the same twelve months. Near or above 1 means profits arrive as cash; persistently low readings mean earnings are tied up in receivables or inventory. NULL when the profit is a loss, and for lenders and insurers.
mf_pctMutual funds' combined holding from the latest pattern filed by that date. Part of DII holding.
insurance_pctInsurance companies' holding (LIC, the private insurers) from the latest pattern filed by that date. Part of DII holding.
fpi_pctForeign portfolio investors (category I and II) from the latest pattern filed by that date. Part of FII holding, which also counts foreign direct investment and depositories.
retail_pctResident individuals holding up to ₹2 lakh of share capital each, from the latest pattern filed by that date.
hni_pctResident individuals holding more than ₹2 lakh of share capital each, from the latest pattern filed by that date.
mf_change_qoqPercentage-point change in mutual fund holding against the previous quarter's pattern. Positive means funds bought.
insurance_change_qoqPercentage-point change in insurance holding against the previous quarter's pattern.
fpi_change_qoqPercentage-point change in FPI holding against the previous quarter's pattern.
retail_change_qoqPercentage-point change in small-retail holding against the previous quarter's pattern. Rising while institutions sell is the usual shape of distribution.
hni_change_qoqPercentage-point change in HNI holding against the previous quarter's pattern.
shareholder_countNumber of shareholders on the latest pattern filed by that date.
shareholder_change_qoqPercent change in the number of shareholders against the previous quarter's pattern. A surge usually means retail arriving.
dividend_ttmRupees per share paid over the trailing twelve months, by ex-date, on the same split-adjusted basis as close. Zero for a stock that paid nothing. The same as `dividend(1y)`; write `dividend(N)` for any span.
dividend_yieldTrailing-twelve-month dividends per share as a percentage of close. The same as `dividend_yield(1y)`.
dividend_growth_yoyTrailing-year dividends per share against the trailing year before it. Null when the earlier year paid nothing. The same as `dividend_growth(1y)`.
dividend_streak_yearsYears spanned by an unbroken run of dividends, each paid within a year and 60 days of the last; drops to 0 when that passes with no payment. Capped at 10, a lower bound where the record starts in 2012.

Insider trades

FieldMeaning
insider_net_buy_value_30dRupees promoters, directors and KMPs bought in the open market less what they sold, over disclosures broadcast in the last 30 days. Zero when none traded; write `insider_net_buy_value_30d > 1cr`.
insider_net_buy_value_90dRupees promoters, directors and KMPs bought in the open market less what they sold, over disclosures broadcast in the last 90 days.
promoter_net_buy_value_90dRupees the promoter group bought in the open market less what it sold, over disclosures broadcast in the last 90 days. Pledges, gifts and transfers inside the group do not count.
insider_buyers_90dDistinct promoters, directors and KMPs who bought in the open market, over disclosures broadcast in the last 90 days. Several insiders buying at once is the stronger signal.
days_since_promoter_buyCalendar days since the latest disclosed open-market promoter purchase became public. Empty for a stock with none on record.

Market context

FieldMeaning
mkt_pct_above_200dmaShare of the market's 1,000 most-traded equities closing above their own 200-day average. The same for every stock on a date — it gates a scan on the market's trend rather than the stock's.
mkt_pct_above_50dmaShare of the market's 1,000 most-traded equities closing above their own 50-day average — the faster breadth reading.
mkt_advancesHow many of the market's 1,000 most-traded equities closed up on the session.
mkt_declinesHow many of the market's 1,000 most-traded equities closed down on the session.
mkt_ad_ratioAdvances divided by declines. Above 1 more stocks rose than fell; `avg(mkt_ad_ratio, 10) > 1` asks for a fortnight of broad buying.
mkt_ad_lineRunning total of each session's net advance ratio, (advances − declines) ÷ (advances + declines), from 1995. Ratio-adjusted so a larger market does not make recent swings taller; read its direction, e.g. `mkt_ad_line > mkt_ad_line[-20]`, not its level.
mkt_mcclellanRatio-adjusted McClellan oscillator: the 19-day minus the 39-day EMA of the net advance ratio × 1000. Crossing above zero marks breadth momentum turning up; deep negative readings mark washed-out sessions.
mkt_new_highsHow many of the market's 1,000 most-traded equities made a new 52-week high on the session — a high above every other high of the past 252 sessions.
mkt_new_lowsHow many of the market's 1,000 most-traded equities made a new 52-week low on the session. `mkt_new_highs > mkt_new_lows` is the plainest healthy-market filter.
nifty_peThe NIFTY 50's price-to-earnings ratio as NSE published it that session, the same for every stock. Recorded from 2012-02; earlier sessions are empty. On 2021-03-31 NSE moved index P/E to consolidated earnings and the reading fell from 40 to 33 overnight, so compare levels on one side of that date.
index_peAn index's own P/E as NSE published it, on index symbols only — stocks read empty. Recorded from 2012-02 where the index existed; consolidated earnings from 2021-03-31, standalone before.

Bulk & block deals

FieldMeaning
bulk_deal_net_qty_pct_30dShares bought minus shares sold in bulk deals over the last 30 days, as a percentage of shares outstanding. Positive means disclosed buyers outweighed disclosed sellers; `> 1` asks for over 1% of the company changing hands to net buyers. NULL where the share count is unknown.
block_deal_value_30dRupees traded in block deals over the last 30 days, each block counted once (its buy side). Write `block_deal_value_30d > 500cr`.
days_since_bulk_buyCalendar days since the last bulk deal on the buy side, 0 on the day of one. NULL if the stock has never had a bulk buy on record. `days_since_bulk_buy <= 5` finds stocks a large buyer has just stepped into.
price is accepted as an alias for close, so price between 50 and 5000 reads naturally.

Sectors & categories#

Closed sets of string values, tested with `is`, `is not` and `in`.

where sector is "Information Technology"
  and pe < 25 and pe > 0
Run
FormMeaning
sector is "Information Technology"Exactly this NSE sector.
sector is not "Financial Services"Everything but this sector.
industry in ("Banks", "Finance")Any of the listed industries.
macro_sector is "Consumer Discretionary"The broadest tier, above sector.

The values are NSE’s official classification — 22 sectors, 12 macro sectors, 58 industries — and the editor autocompletes them after is, so nobody has to remember that the exact spelling is “Oil Gas & Consumable Fuels”. Matching is case-insensitive; the compiler canonicalises onto the official name.

Classification is today’s snapshot — NSE publishes no history, so a historical scan or hit-rate replay applies the current sector retroactively. The compiler says so on every scan that filters on one.

Derivatives#

Per-stock daily aggregates from the NSE F&O bhavcopy, for the roughly 200 stocks with listed futures & options.

from fno where fut_oi_change_pct > 3 and change > 1
Run

Price and open interest rising together is a long buildup — fresh money agreeing with the move, rather than shorts giving up. Open interest is in shares, summed across all expiries and restated across splits and bonuses like close, so a split does not read as a jump in OI.

FieldMeaning
fut_oiTotal futures open interest, in shares.
fut_oi_valueFutures OI × price, in rupees — `fut_oi_value > 5000cr`.
fut_oi_change_pctDay-over-day change in futures OI. Rising OI with rising price is a long buildup.
fut_volumeFutures contracts traded.
fut_basis_pctFront-month futures premium (+) or discount (−) to cash.
fut_rollover_pctShare of OI already in later expiries.
pcr_oiPut-call ratio by open interest.
pcr_volPut-call ratio by contracts traded.
pcr_oi_frontPut-call ratio by OI on the nearest expiry only.
max_painFront-expiry strike where option holders collect least — `close < max_pain`.
max_pain_dist_pctMax pain against the close, in percent; positive when max pain is above.
call_oi_wallStrike at or above the price with the most front-expiry call OI, read as resistance.
put_oi_wallStrike at or below the price with the most front-expiry put OI, read as support.
nifty_pcr_oiThe Nifty 50's put-call ratio by OI, on every stock's row.
banknifty_pcr_oiThe Nifty Bank's put-call ratio by OI, on every stock's row.
pattern is long_buildupThe day's positioning: `long_buildup`, `short_buildup`, `short_covering`, `long_unwinding` — price direction against the change in live futures OI.
from fno where pcr_oi > 0.8 and close > sma(50)
Run
Stocks without listed derivatives carry NULL in every one of these fields and never match a condition on them. from fno makes that scope explicit — writing an F&O field without it earns a compiler warning rather than a silently thin result.

Fundamentals#

Valuation, growth and shareholding, as the market knew them on the scan date.

where marketcap > 20000cr and pe < 30 and pe > 0
Run

Fundamentals are point-in-time. A result filed after the 15:30 close becomes visible from the next session, and a restatement counts only from its own filing date — so a scan on any past date sees exactly what a trader could have known that day, never what the filings later became. That is what keeps the hit-rate replay honest for fundamental screens.

FieldMeaning
marketcapClose × shares outstanding, in rupees — `marketcap > 5000cr`.
ffmcFree-float market cap: marketcap less the promoter stake on the latest filed pattern — `ffmc > 1000cr`.
pePrice to trailing-twelve-month earnings. NULL for loss-makers rather than a misleading number.
eps_ttmTrailing EPS, split-adjusted to match the adjusted price series.
revenue_ttm / net_profit_ttmRevenue and net profit over the last four filed quarters, in rupees — `revenue_ttm > 500cr`.
net_margin_ttmTTM net profit over TTM revenue, as a percentage. Smooths a single quarter's seasonality.
other_income_shareTTM other income as a percentage of TTM profit before tax — high readings mean profit leans on treasury income. NULL for lenders, insurers and when PBT is not positive.
revenue_growth_yoy / profit_growth_yoyLatest filed quarter against the same quarter last year.
revenue_growth_qoq / profit_growth_qoqLatest filed quarter against the previous quarter.
profit_cagr_2y … profit_cagr_7yAnnualized profit growth over 2 to 7 years, trailing twelve months against the TTM N years earlier. NULL when the history is short or either end was a loss.
revenue_cagr_2y … revenue_cagr_7yThe same window on revenue — pair with the profit CAGR to see whether margins widened or only sales did.
interest_cost_growth_yoyTTM finance costs against a year ago. Moves every quarter; the debt itself is `debt_to_equity`, which moves twice a year.
pbMarket cap over the shareholders' equity on the latest balance sheet.
roe / roceReturn on equity, and on equity plus borrowings (pre-tax, before interest), over the twelve months to the latest balance sheet against capital averaged with the year before. ROCE is NULL for lenders and insurers.
debt_to_equity / net_debtBorrowings over equity, and borrowings less cash, bank balances and current investments in rupees — `net_debt < 0` is net cash. NULL for lenders and insurers.
current_ratioCurrent assets over current liabilities. NULL for lenders and insurers.
fcf_ttm / fcf_yieldOperating cash flow less capex over the trailing year, in rupees, and that as a percentage of market cap. NULL for lenders and insurers.
cash_conversionOperating cash flow over net profit for the same twelve months. NULL for a loss year.
promoter_pct / public_pct / fii_pct / dii_pctShareholding from the latest pattern filed by the scan date.
promoter_pledged_pctShare of the promoter stake pledged as collateral.
promoter_change_qoqPercentage-point change in promoter holding vs the previous quarter.
mf_pct / insurance_pct / fpi_pctMutual fund, insurance and foreign portfolio investor holding from the same filing. MF and insurance sit inside DII, FPI inside FII.
retail_pct / hni_pctResident individuals holding up to ₹2 lakh of share capital each (retail), and above it (HNI).
mf_change_qoq / insurance_change_qoq / fpi_change_qoq / retail_change_qoq / hni_change_qoqPercentage-point change against the adjacent quarter's filing. NULL across a gap in the filings.
shareholder_count / shareholder_change_qoqNumber of shareholders on the latest filing, and its percent change against the adjacent quarter.
dividend_ttm / dividend_yieldRupees per share paid over the trailing year by ex-date, split-adjusted to match close, and that as a percentage of close. Non-payers read 0.
dividend_growth_yoyTrailing-year dividends against the trailing year before. NULL when the base year paid nothing.
dividend_streak_yearsConsecutive trailing years that paid, counting back from today. Capped at 10; the record starts in 2012.
dividend(Ny) / dividend_yield(Ny)The same over any span of years — `dividend_yield(3y) > 9%`. One year reads the stored column; other spans are derived at scan time.
dividend_growth(Ny) / dividend_cagr(Ny)Trailing-year dividends against the trailing year that ended N years ago: total change, and annualised.
dividend_years(Ny)How many of the trailing N years paid — `dividend_years(5y) == 5` is a stock that never skipped.
where promoter_change_qoq > 0.5 and close > sma(200)
Run

Dividends

where dividend_yield(1y) > 3% and dividend_years(5y) == 5 and dividend_cagr(5y) > 5
sort by dividend_yield(1y) desc
Run

Dividends are counted by ex-date, in rupees per share on the same split-adjusted basis as close, so a yield survives a split. The dividend functions take a number of years rather than bars — dividend(3y), dividend_yield(1y), dividend_cagr(5y) — and any span from one to ten years is derived at scan time from the ex-date record; the one-year forms read the stored columns. A stock that paid nothing reads 0, not NULL, so dividend_yield(1y) == 0 finds the non-payers. The record begins in 2012; a span reaching before it reads NULL rather than a partial total.

The filings backfill is in progress: a stock without parsed results has NULL P/E and growth, and NULL never matches. The compiler warns when a scan filters on a partial-coverage field, so a thin result is announced rather than mistaken for a market fact.

Market context#

Breadth and index valuation: one value per session, the same on every stock, so a scan can require a healthy market.

where mkt_pct_above_200dma > 50 and close crosses above max(high, 20 bars)
Run

A breakout scan matches in any market; this one only on sessions when most stocks are above their own 200-day average. Because the value sits on every row of its date, a hit-rate replay applies the gate session by session, and window functions work on it like any series — avg(mkt_ad_ratio, 10) > 1.

FieldMeaning
mkt_pct_above_200dma / mkt_pct_above_50dmaPercent of the market's 1,000 most-traded equities closing above their own 200- or 50-day average.
mkt_advances / mkt_declinesHow many of those stocks closed up, or down, on the session.
mkt_ad_ratioAdvances ÷ declines. `avg(mkt_ad_ratio, 10) > 1` is a fortnight of broad buying.
mkt_ad_lineRunning total of the net advance ratio since 1995; read its direction — `mkt_ad_line > mkt_ad_line[-20]`.
mkt_mcclellanRatio-adjusted McClellan oscillator: 19- minus 39-day EMA of the net advance ratio × 1000.
mkt_new_highs / mkt_new_lowsHow many of those stocks made a new 52-week high, or low, on the session.
nifty_peThe NIFTY 50's P/E as NSE published it that session. From 2012-02; consolidated earnings from 2021-03-31.
index_peAn index's own P/E, on index symbols only — stocks read NULL.
Breadth counts the 1,000 most-traded main-board equities on each session by 20-session traded value — no SME listings, ETFs or indices, and every equity before about 2011, when fewer traded. Percentages and ratios compare across the whole history from 1995; the raw counts from about 2011, once the universe is full.

Insider trades#

What promoters, directors and KMPs bought and sold in the open market, from their SEBI disclosures.

where promoter_net_buy_value_90d > 1cr and close > sma(200)
Run

Every disclosure is dated on the day NSE broadcast it — the next day when it landed after the 15:30 close — and never on the trade date, which routinely runs weeks earlier. Only open-market purchases and sales count: an ESOP allotment, a gift, a transfer inside the promoter group or a pledge moves shares without saying anything about the price. Debentures and warrants filed on the same forms are left out.

FieldMeaning
insider_net_buy_value_30d / insider_net_buy_value_90dRupees promoters, directors and KMPs bought in the open market less what they sold, over disclosures broadcast in the trailing 30 or 90 days — `insider_net_buy_value_90d > 1cr`. Zero when none traded.
promoter_net_buy_value_90dThe same for the promoter group alone, over 90 days. Pledges, gifts and inter-se transfers do not count.
insider_buyers_90dDistinct promoters, directors and KMPs who bought in the open market in the trailing 90 days.
days_since_promoter_buyCalendar days since the latest open-market promoter purchase became public. NULL for a stock with none on record.
The record begins in November 2015, and a window reaching before it reads NULL. A stock no insider traded reads 0, so insider_net_buy_value_90d > 0 is a real condition rather than a coverage filter.

Bulk & block deals#

Who has been buying or selling the stock in size, from NSE's daily deal disclosures.

where days_since_bulk_buy <= 5 and bulk_deal_net_qty_pct_30d > 0.5
Run

Stocks a large buyer stepped into this week, where disclosed buyers have taken over half a percent of the company more than disclosed sellers in the last month. Every value is a trailing 30-day window ending on the bar, so a replay reads exactly the deals published by each session.

FieldMeaning
bulk_deal_net_qty_pct_30dShares bought minus sold in bulk deals over 30 days, as a % of shares outstanding. Split-adjusted, so a split between two deals does not fake a buyer.
block_deal_value_30dRupees traded in block deals over 30 days, each block counted once — `block_deal_value_30d > 500cr`.
days_since_bulk_buyCalendar days since the last bulk deal on the buy side; 0 on the day. NULL if there has never been one.
A bulk deal is any client whose trades in a stock on one day add up to more than 0.5% of its shares; a block deal is a single large trade in the separate block window. Bulk deals are recorded from 2004 and block deals from November 2005 — earlier dates read NULL, and a stock with no deals in the window reads 0. The symbol page lists the deals themselves, client by client.

Indicators#

Called with a period, over close.

where rsi(14) < 40 and close > sma(200)
Run
IndicatorPeriodsMeaning
sma(n)5, 10, 20, 50, 100, 200, anyUnweighted mean close over the period.
ema(n)9, 21, 50, 200, anyExponentially weighted mean close, seeded from the SMA.
wma(n)20, anyLinearly weighted mean — the newest bar counts most.
hma(n)21, anyHull's low-lag moving average.
tema(n)20, anyTriple-smoothed EMA, with much of the lag removed.
rma(n)14, anyWilder's smoothing, as used inside RSI and ATR.
vwma(n)20, anyMean close weighted by each bar's volume.
rsi(n)14, anyWilder's relative strength index.
atr(n)14, anyWilder's average true range.
adx(n)14, anyTrend strength, direction-agnostic.
cci(n)20, anyCommodity Channel Index.
mfi(n)14, anyVolume-weighted RSI.
cmf(n)20, anyAccumulation/distribution normalised over a window.
roc(n)10, anyPercentage change over the period.
return(n)1, 5, 10, 21, 63, 126, 252, anyPercentage change in close over the period. The listed periods read the stored return columns; any other, `return(3y)` say, is derived at scan time.
highest_high(n)252, anyHighest high over the period. 252 reads the stored 52-week column; any other period is derived at scan time.
lowest_low(n)252, anyLowest low over the period. 252 reads the stored 52-week column; any other period is derived at scan time.
pct_from_high(n)252, anyDistance below the highest high of the period, as a negative percentage. 252 reads the stored 52-week column.
pct_from_low(n)252, anyDistance above the lowest low of the period, as a percentage. 252 reads the stored 52-week column.
williams_r(n)14, anyLike the stochastic, scaled −100 to 0.
dividend(years)1y, anyRupees per share paid over the trailing N years by ex-date, split-adjusted to match close. Zero for a stock that paid nothing.
dividend_yield(years)1y, anyDividends per share over the trailing N years as a percentage of close. Over one year this is the ordinary dividend yield.
dividend_growth(years)1y, anyTotal change in trailing-year dividends per share against the trailing year that ended N years ago. Null when that base year paid nothing.
dividend_cagr(years)1y, anyAnnualised growth in trailing-year dividends per share over N years. Over one year it equals the total change.
dividend_years(years)anyHow many of the trailing N years paid a dividend. `dividend_years(5y) == 5` is a stock that paid every year.

Every indicator accepts any period: the listed ones read a precomputed column and anything else is computed at scan time, so ema(37), adx(7) and bb(50, 2.5) all work. Multi-output indicators take their full parameter list — macd(8, 21, 5), supertrend(14, 2) — and written bare, macd() keeps meaning the stored 12/26/9.

The fixed-window price fields work the same way. Every one of them is the stored period of an indicator that takes any window: high_52w is highest_high(1y), return_1m is return(1mo), and highest_high(3y), return(2y) or pct_from_low(6mo) are derived at scan time from the same bars. A period can be written as a duration — lowest_low(5y) — or as a bar count. The dividend functions count years instead; see Fundamentals.

An indicator reads close and only close. There is no ema(high, 20): a moving average of another series is a window function, avg(high, 20), which is computed at scan time over any expression.

Multi-output indicators

Indicators producing more than one line take empty parentheses and a sub-field. Omitting the sub-field picks the one marked (default) — bb() is bb().mid.

where macd().line crosses above macd().signal
  and close > bb().upper
Run
IndicatorSub-fieldsMeaning
macd().line (default) .signal .hist12/26 EMA difference, with a 9-period signal line.
bb().upper .mid (default) .lower .pctb .width20-period SMA with two-standard-deviation bands.
stoch().k (default) .dSlow stochastic oscillator.
supertrend().value (default) .dirATR-banded trailing stop.
donchian().upper .mid (default) .lowerThe rolling 20-bar high/low envelope.
keltner().upper (default) .lowerA 20-EMA with ATR-scaled bands.
ichimoku().conversion .base (default) .span_a .span_bConversion, base and cloud edges.
aroon().up .down .osc (default)How recently the window's extremes occurred.

Window functions#

Rolling aggregates over any period, computed at scan time rather than read from a column.

where volume > 2x avg(volume, 20)
  and close > max(high, 20 bars)[-1]
Run
FunctionReturns
avg(x, n)Mean of x over the last n bars.
max(x, n)Highest value of x over the last n bars.
min(x, n)Lowest value of x over the last n bars.
sum(x, n)Total of x over the last n bars.
stdev(x, n)Population standard deviation of x.
median(x, n)Median of x over the last n bars.

The window takes a bar count or a duration — avg(volume, 20), max(high, 20 bars) and min(low, 52w) are all valid. The first argument is any expression, so avg(high - low, 10) gives the mean daily range.

A window returns nothing until it is full. A 50-day average over a stock with 12 days of history is not a 50-day average, so avg(close, 50) and sma(50) agree exactly rather than differing on new listings.

Math functions#

One bar at a time, so they compose with everything.

FunctionReturns
abs(x)Distance of x from zero.
ceil(x)x rounded up to a whole number.
floor(x)x rounded down to a whole number.
round(x)x rounded to the nearest whole number.
square(x)x multiplied by itself.
sqrt(x)Square root of x. Nothing where x is negative.
log(x)Natural log of x. Nothing where x is zero or less.
log10(x)Base-10 log of x. Nothing where x is zero or less.
greatest(a, b, …)The largest of its values.
least(a, b, …)The smallest of its values.
where abs(close - open) > 2% above avg(abs(close - open), 20)
Run

Unlike a window these read a single bar, so one can sit inside a window — avg(abs(close - open), 20) is the mean candle body over twenty sessions. greatest and least take two or more values; the rest take one.

Where a function is undefined — the square root of a negative, the log of zero — it returns nothing rather than failing. A scan reads every bar of history, so one bad row would otherwise take the whole query down with it.

Counting bars#

How many of the last N — not all of them.

FunctionReturns
count(cond, n)How many of the last n bars matched. `count(close > open, 10) >= 7` is seven up days in ten.
countstreak(cond, n)How many matched in a row, counting back from the scan bar. Stops at the first that did not.
where count(volume > 2x avg(volume, 20), 10) >= 3
  and countstreak(close > sma(20), 30) >= 10
Run

These take a condition where everything else takes a value. has been … for N bars demands every bar; this is how you ask for most of them.

A bar whose condition is unknown counts as not matched, so a count is always a number from 0 to the window and never nothing. A count cannot sit inside a window function.

Time travel#

Past bars are negative. There is no future.

WrittenMeans
closeThis bar — the scan date.
close[-1]The previous bar.
prev closeThe previous bar, spelled out. `prev` takes a field, not an expression.
close[-5]Five bars ago.
avg(volume, 20)[-1]The 20-bar average as of yesterday.
max(high, 20 bars)[-1]The 20-bar high excluding today — a breakout level.
where open > high[-1] and close > open
Run
A positive index is rejected rather than guessed at: close[1] would name a bar that has not happened, and silently treating it as the past is how look-ahead bias gets into a scan.

Operators#

The usual comparisons and arithmetic, plus chaining.

OperatorsNotes
> >= < <= = !=`=` is accepted as `==`.
+ - * /Division by zero yields no value rather than an error.
and or not`and` binds tighter than `or`. Parenthesise when mixing.
( )Grouping, for both conditions and arithmetic.

Chained comparisons

A chain means what it looks like — each neighbouring pair must hold.

where close > ema(21) > ema(50) > ema(200)
Run

That is exactly equivalent to writing the three comparisons out and joining them with and.

Sugar#

Shorthand for the arithmetic other screeners force you to write by hand.

WrittenEquivalent to
volume > 2x avg(volume, 20)volume > 2 * avg(volume, 20)
close > 5% above ema(50)close > ema(50) * 1.05
close < 3% below sma(200)close < sma(200) * 0.97
close within 2% of high_52wclose between high_52w * 0.98 and high_52w * 1.02
price between 50 and 5000price >= 50 and price <= 5000
price is not between 50 and 5000not (price >= 50 and price <= 5000)
close up 3% over 5 bars(close - close[-5]) / close[-5] >= 0.03
close down 2% over 5 bars(close[-5] - close) / close[-5] >= 0.02
close up 3% from prev close(close - close[-1]) / close[-1] >= 0.03

These are not approximations. Each form is checked against its longhand equivalent in the test suite and must select exactly the same stocks.

Where two spellings mean the same thing, the block editor and the API’s normalized field print one of them: has stayed above comes back as has been above, crosses … within n bars as crossed … within n bars, and up 3% from prev close as up 3% over 1 bars. The query you wrote and the query it prints select the same stocks.

Event operators#

Crossings, persistence, runs and extremes — the reason the language exists.

Every one of these compiles to a bounded window expression. In a screener without them you would hand-roll the same thing out of offset arithmetic, and get it subtly wrong.

Crossings

where sma(50) crosses above sma(200)
Run

A crossing is defined on two bars: strictly across now, and not across on the bar before. Add a recency window to catch one that happened a few sessions ago.

where rsi(14) crossed above 30 within 3 bars
  and close > sma(200)
Run

Persistence

Whether something has held for a stretch, rather than being true on one lucky day.

where close has been above ema(21) for 10 bars and adx > 25
Run

Monotonic runs

where volume rising for 3 bars and close > close[-3]
Run

Window extremes

is highest in compares against a window that includes the current bar, so it is true exactly when this bar sets the extreme.

where close is highest in 52w and volume > 1.5x avg(volume, 20)
Run
FormExampleTrue when
x crosses above ysma(50) crosses above sma(200)x is above y now and was at or below on the previous bar.
x crosses below yclose crosses below ema(21)x is below y now and was at or above on the previous bar.
x crossed above y within n barsrsi(14) crossed above 30 within 3 barsThat crossing happened on any of the last n bars.
x crosses above y within n barssma(50) crosses above sma(200) within 5 barsThe same thing — `crosses` and `crossed` both take a recency window.
x has been above y for n barsclose has been above ema(21) for 10 barsx was above y on every one of the last n bars.
x has stayed above y for n barsrsi(14) has stayed below 30 for 3 bars`stayed` reads better after a level and means exactly `has been`.
x rising for n barsvolume rising for 3 barsx increased on each of the last n bars.
x falling for n barsclose falling for 5 barsx decreased on each of the last n bars.
x is highest in nclose is highest in 52wNo bar in the window has a higher x.
x is lowest in nlow is lowest in 20 barsNo bar in the window has a lower x.
x is not highest in nclose is not highest in 20 barsThe negation of the above — this bar does not set the extreme.
pattern is ppattern is bullish_engulfingThe pattern was detected on the scan bar.
pattern is p within n barspattern is hammer within 2 barsIt was detected on any of the last n bars.
pattern is not ppattern is not dojiExcludes a pattern rather than requiring one.

Candlestick patterns#

Detected at build time and queried as a first-class value.

where pattern is bullish_engulfing
  and close within 3% of sma(50)
Run

Add within n bars for recency, or pattern is not … to exclude one. Definitions use proportional tolerances, so they behave the same on a ₹30 stock and a ₹30,000 one.

PatternShape
dojiOpen and close nearly equal — indecision.
hammerLong lower wick, small body at the top.
shooting_starLong upper wick, small body at the bottom.
marubozuAlmost no wicks — one side controlled the session.
bullish_engulfingAn up bar whose body swallows the previous down bar.
bearish_engulfingA down bar whose body swallows the previous up bar.
bullish_haramiA small up bar contained inside the previous down bar.
bearish_haramiA small down bar contained inside the previous up bar.
morning_starDown bar, pause, then a strong up bar through the midpoint.
evening_starUp bar, pause, then a strong down bar through the midpoint.
three_white_soldiersThree consecutive strong up bars.
three_black_crowsThree consecutive strong down bars.
inside_barRange contained entirely within the previous bar's.
outside_barRange containing the whole previous bar's.
long_buildupPrice up, futures OI up — fresh longs.
short_buildupPrice down, futures OI up — fresh shorts.
short_coveringPrice up, futures OI down — shorts closing.
long_unwindingPrice down, futures OI down — longs closing.

Numbers & literals#

Indian and Western magnitudes, both native.

WrittenValue
1k1,000
5L5,00,000 — five lakh
2.5m25,00,000
10cr10,00,00,000 — ten crore
1b100,00,00,000
₹500 / $50500 / 50 — the symbol is read and discarded
20 bars / 52w / 3moA duration, converted to trading sessions
d / w / mo / yA day, week, month and year: 1, 5, 21 and 252 sessions.
where turnover > 10cr and volume > 5L
Run

Durations convert at 252 sessions a year — 21 a month, 5 a week — so 52w, 12mo and 1y are all 252 bars, and max(high, 52w) is exactly the stored high_52w. No term may reach back more than 1260 bars: 5y is the widest window there is, and 6y is refused. As an argument to a dividend function, a duration is read in whole years instead.

Universes#

Which stocks a scan considers, chosen in the toolbar rather than in the query.

TierContains
Top 100Most traded 100 stocks by 20-day turnover
Top 250Most traded 250 stocks by 20-day turnover
Top 500Most traded 500 stocks by 20-day turnover
Top 1000Most traded 1000 stocks by 20-day turnover
All equitiesEvery actively traded NSE equity
These tiers are ranked by 20-day average turnover, not by index membership — “Top 100” means the hundred most traded stocks, not the NIFTY 100. Market cap exists as the marketcap field for conditions, but the tiers stay turnover-ranked because turnover is dense from day one while fundamentals coverage is still filling in.

from fno narrows a scan to the roughly 200 stocks with listed futures & options — the set the derivatives fields cover.

from fno where fut_oi_change_pct > 10 and close up 2% over 1 bars
Run

Errors#

Every mistake is reported with the reason and the fix, before the scan runs.

The parser and analyser run in the browser as you type — they are pure and need no database — so the editor underlines a problem immediately, and the identical code validates again on the server.

WrittenReported asHint
clos > 100Unknown field `clos`Did you mean `close`?
sma > 100`sma` needs a periodTry `sma(5)`, `sma(10)`, `sma(20)`.
rsi(5000) > 50`rsi` period must be between 2 and 1260e.g. `rsi(14)`.
dividend(3mo) > 0`dividend` takes whole years, not 3moWrite whole years, e.g. `dividend(1y)`.
close[1] > 100`[1]` looks like a future barPast bars are negative — write `[-1]` for 1 bar ago.
ema(high, 20) > 0`ema` is only precomputed over closeWrite `ema(20)`, or use a window function such as `avg(high, 20)`.
min(low, 6y) > 0A 1512-bar window is too longWindows reach back at most 1260 bars.
rsi(14) crosses 30`crosses` must be followed by `above` or `below`
on 15m where close > 100Timeframe `15m` is not availableThis dataset is end-of-day only, so `on daily` is the only timeframe.
from BSE where close > 100Universe `BSE` is not availableAvailable universes: `NSE` (all equities) and `fno` (stocks with listed futures & options).
top 10% by marketcapPercentile limits are not supportedUse a plain count, e.g. `top 25`.
pattern is wibbleUnknown pattern `wibble`
macd().wibble > 0`macd()` has no sub-field `wibble`Available: line, signal, hist.

Recipes#

Complete scans worth stealing. Each one runs.

52-week high breakout

A close at the highest level in a year, on volume half again its own 20-day average. The Darvas entry, written out.

where close is highest in 52w and rel_volume > 1.5
Run

Within 3% of the 52-week high

Within 3% of the yearly high and above the 50-day EMA. Plenty of traders prefer this to the breakout candle itself.

where close within 3% of high_52w and close > ema(50)
Run

Volume breakout

Twice the 20-day average volume with a gain over 3%. Something changed today.

where volume > 2x avg(volume, 20)
  and change > 3
  and close > sma(20)
Run

Volume shockers

Twice the ten-day average volume and a move of more than 5%.

where volume > 2x avg(volume, 10) and change > 5
Run

20-day channel breakout

Price above yesterday's 20-day channel top on 1.5 times normal volume. The Turtle entry.

where close > donchian_upper[-1]
  and volume > 1.5x avg(volume, 20)
Run

Bollinger band breakout

A close above the upper Bollinger band with volume 1.5 times its average.

where close > bb().upper and rel_volume > 1.5
Run

Keltner channel breakout

Clearing the upper Keltner channel, which is built on ATR and shifts more slowly than a Bollinger band.

where close > keltner_upper and rel_volume > 1.5
Run

Crossing above pivot R1

The close crossing above the first pivot resistance on 1.5 times average volume. Floor traders have watched R1 for decades.

where close crosses above pivot_r1 and rel_volume > 1.5
Run

The scan library has 144 of these, each testable against a year of history.

Data caveats#

What the newer data can and cannot honestly answer.

CaveatThe honest version
F&O coverageRoughly 200 NSE stocks have listed derivatives. Everywhere else the F&O fields are NULL and never match — `from fno` makes the scope explicit.
Fundamentals coverageThe filings backfill is in progress; a stock without parsed results has NULL P/E and growth, and never matches those conditions. Coverage rises weekly.
Institutional holdingsPromoter and public shares of equity are filed every quarter and are the two the shareholding view is built on. FII, DII, employee-trust and pledge percentages come from a separate filing stage whose backfill has reached about 1% of instruments — they are NULL almost everywhere, including for the largest companies on the exchange, and never match.
Quarterly resultsThe per-quarter profit and loss starts in 2017 and the shareholding pattern in 2015. Only the half-year and annual filings carry a balance sheet (from September 2022) and a cash flow statement (from September 2020), so `pb`, `roe`, `roce`, `debt_to_equity`, `net_debt`, `current_ratio`, `fcf_ttm`, `fcf_yield` and `cash_conversion` change twice a year and are NULL before those dates.
Sector historyClassification is today's snapshot — NSE publishes no history. A historical scan or hit-rate replay applies the current sector retroactively.
OI unitsFutures open interest is in shares (index units for an index), summed across expiries and restated onto today's split-adjusted basis like close, so a split or bonus does not read as an OI jump. Futures volume is in contracts.

Fundamentals are point-in-time: a scan sees the numbers as the market knew them on the scan date. A result filed after the 15:30 close becomes visible from the next session, and a restatement counts only from its own filing date — which is what keeps the hit-rate replay honest for fundamental screens.

Limits#

What a query is compiled against. Crossing one is an error with a span, never a silent truncation.

LimitWhat it bounds
1260 bars of lookbackThe furthest back any one term may reach — five years of sessions: a window length, a bar offset, an event window, or a window plus its offset. `5y` fits; `6y` does not. Dividend spans are counted in years against a separate table and are capped at 10.
8,000 characters of sourceThe longest query the compiler accepts, in the browser and over the API alike.
5,000 rowsThe hard cap on `top`. A larger number is clamped rather than refused.
`top` defaults to 300And `sort by <expr>` with no direction sorts descending.
Indicator periods 2–1260Per-parameter bounds come from the catalog — `bb`'s standard deviation is 0.5–5, `supertrend`'s multiplier 0.5–10, and the recursive kernels (`stoch`, `keltner`, `aroon`) stop at 400.

The lookback ceiling is the one real queries reach, and it is what makes a scan’s cost knowable before it runs: the analyser computes the deepest bar any term needs, so the engine reads that many and no more. min(low, 3y) is 756 bars and compiles; min(low, 6y) is 1512 and does not. A scan that reaches past a year of history reads proportionally more rows, and only that scan pays for it.

Not supported#

Parts of the language spec this dataset cannot honour, and why.

FeatureWhy not
on 15m / 1h / weeklyThe dataset is end-of-day only. `on daily` is the only timeframe.
and on daily: …Inline multi-timeframe scopes need a second timeframe to scope to, and there is only one.
from BSE / US / NIFTY500NSE only, and there is no index-constituent list to filter by. `from fno` is the one list-like universe.
from NSE except … / watchlist "…"Set subtraction and user lists are not implemented; `except` is reserved but unused.
vwapNeeds intraday data, which end-of-day bars cannot provide.
ema(high, 20)Indicators read close only. For another series use a window function — `avg(high, 20)`.
close 3 bars agoThe English offset was dropped for one spelling. Write `close[-3]`.
… as golden / … since goldenEvent sequencing — one condition holding after another — is not implemented. `as` and `since` are reserved but unused.
scan "name" { … }A query is the scan. Naming and saving one is an account action, not syntax.
per-strike option screeningF&O fields are per-stock daily aggregates. Strike-level OI is a chain view, not a screener column.
pattern is … formingGeometric pattern detection with a confidence score is not built. Only completed candlestick patterns are available.
backtest { } / alert { }Not implemented. The hit-rate panel on each scan is the nearest thing.
top 10% by …Percentile limits are rejected; use a plain count.
A query using any of these fails with a specific message naming the limitation, rather than silently returning something plausible.

API & MCP#

Everything on this page works without the browser.

A scan is one POST with the Sift source in the body, and the MCP endpoint gives a coding agent this whole reference as a tool — so Claude Code or Codex can write and run scans against your account unaided. Keys are free on every plan.

Read the API & MCP docs