F&O ban list aur MWPL: crowding kya batata hai
7 min read · data: 437-stock NSE universe · descriptive study, not advice
Har kuch din mein NSE ki “securities in ban period” list mein naye naam aate hain, aur trading groups mein do kahaniyan chalti hain: “ban mein aaya, ab shorts cover honge, girega” aur “ban se nikla, ab pop aayega.” Dono ke peeche ek logical mechanism hai — forced unwinding. Humne 8.5 saal ka data leke dekha ki in kahaniyon mein kitna dum hai. Yeh measurement hai, prediction nahi.
Pehle basics: MWPL aur ban kya hai
- MWPL (Market-Wide Position Limit): har F&O stock ke liye ek limit ki poore market mein total kitna open interest (OI — khule futures/options positions) ho sakta hai.
- Ban: jab kisi stock ka OI uske MWPL ke 95% ko cross kar jaata hai, toh woh ban period mein chala jaata hai — naye positions nahi ban sakte, sirf purane kam ho sakte hain. Jab OI 80% se neeche aata hai, ban hat jaata hai.
- Crowding ratio (OI/MWPL): limit ka kitna hissa bhar chuka hai. 0.9 matlab 90% bhara hua.
Data: NSE ki official combineoi archive — 2,108 trading din (Jan 2018 – Jul 2026),
3.9 lakh symbol-day rows, 353 F&O stocks. Isme har din har stock ka MWPL, OI aur “next
day limit” hai — yahi file ban decide karti hai.
Part 1: ban list — direction ka koi signal nahi
126 stocks kabhi na kabhi ban mein gaye → 1,277 ban entries. Price data ke saath 984 usable entries aur 983 exits bache. Har event ke aas-paas market-adjusted return (stock minus NIFTY) mapa.
Ban mein jaate waqt:
| Window | Return (NIFTY minus) |
|---|---|
| Ban se 5 din pehle se 2 din pehle tak | +1.40% |
| Ban ka din | −0.52% |
| Ban ke baad 10 din | +0.65% |
Matlab bada crowded move ban se pehle hi ho chuka hota hai. Ban entries mein se 25% stocks pehle hi ±20% move kar chuke the. Ban ke baad “forced unwinding se girna” data mein nahi dikha — ulta halki si positive drift.
Ban se nikalte waqt:
| Window | Return (NIFTY minus) |
|---|---|
| Release se 2 din pehle se release tak | +1.67% |
| Release ke baad 10 din | +1.30% |
+1.30% achha lagta hai. Lekin yahan asli test aata hai: unhi 92 stocks ke random non-ban dino pe 10 din hold karte toh kya hota? 3,000 random draws: average +0.21%, lekin spread (standard deviation) 8.4%. Asli result ka z-score: +0.13 — yaani noise ke andar. Yeh volatile stocks hain; inme kisi bhi din bade moves aate hain.
Saal dar saal bhi sign badalta raha: 2018 −1.6%, 2019 −6.9% (mid-cap bear market), phir 2020–2025 positive. Yeh market regime follow karta hai, koi fixed mechanism nahi.
Verdict: ban list ek direction signal ke roop mein dead hai.
Part 2: crowding — direction nahi, risk batata hai
Yahan kahani interesting hoti hai. Humne har stock-day ko crowding ratio ke hisaab se 10 hisson (deciles) mein baanta aur agle 5 din dekhe (349,430 symbol-days, 286 stocks):
| Crowding decile | Avg 5-day return | Volatility | Bura 5% din (p05) | Sharpe* |
|---|---|---|---|---|
| 1 (sabse kam crowded) | +0.28% | 4.45% | −6.31% | 0.45 |
| 5 | +0.28% | 4.52% | −6.68% | 0.44 |
| 9 | +0.31% | 5.60% | −7.96% | 0.39 |
| 10 (sabse zyada crowded) | +0.25% | 6.04% | −8.27% | 0.29 |
*Sharpe = return per unit risk; zyada = behtar risk-adjusted.
Dhyan se dekhiye: average return lagbhag same hai (0.25% vs 0.28%). Matlab crowding direction nahi batati. Lekin volatility, bura-din loss aur drawdown crowding ke saath seedha badhte hain. Crowded stocks wilder hote hain — dono taraf. (Unka upside bhi bada hota hai: top decile ki skew +1.92 hai, yaani kabhi kabhi bade rally bhi.)
Aur ek honest twist: ban hone wale stocks agle 5 din mein zyada chale (0.74% vs 0.27%). Toh crowded stocks ko “avoid list” ki tarah use karna return bhi kaat sakta hai.
Ek portfolio test
Phir humne dekha: ek equal-weight F&O basket mein agar crowded stocks (OI/MWPL > 0.75) ko temporarily cash mein rakh dein, toh basket ka risk-adjusted performance kaisa badalta hai? Basket ka Sharpe 0.48 → 0.57 hua, aur yeh har strict comparison (random veto, volume-matched, volatility-matched) ko beat kar gaya, dono periods (2018–21 aur 2022–26) mein.
Lekin teen honest caveats, jo humne research mein khud likhe:
- Effect chhota hai (Sharpe +0.07 se +0.09).
- Yeh in-sample hai — poora 2018–2026 data use hua, koi forward holdout nahi.
- Mechanism woh nahi nikla jo socha tha. Yeh individual stock ke crash ka nahi, balki crowded stocks ke ek saath girne (correlated stress) ka effect lagta hai.
Isliye yeh abhi sirf forward paper observation mein hai, kisi live use mein nahi.
Kya dekhein (kya kharidein nahi)
- Ban list mein naam aane ka matlab hai: move pehle ho chuka, aur stock abhi bahut volatile hai. Direction ka hint nahi.
- High crowding = high risk state. Position size aur risk ke sandarbh mein yeh dekhne layak hai.
- Humare /stocks/ pages pe F&O stocks ki OI context daily refresh hoti hai.
The limitation
Delist ya rename hue stocks (jaise DHFL, RCOM) price data mein missing hain — 293 ban events isi wajah se chhoote. Crowding result in-sample hai aur forward test baaki hai. Market-adjusted return NIFTY ke against hai, beta-adjusted version ne same conclusion diya. Base rates past describe karte hain; future predict nahi karte.
Methodology note. Ban study: source NSE NSCCL combineoi_DDMMYYYY archive,
2,108 days, 2018-01 → 2026-07-23; ban = “No Fresh Positions” next day; spell = run of
consecutive ban days; abnormal return = stock − NIFTY; nulls = 3,000 matched-stock random
non-ban dates (±15-day buffer) + event-date cluster bootstrap. Crowding study: 349,430
symbol-days, 286 F&O stocks; ratio known at close T, forward close-to-close; overlay =
equal-weight daily-rebalanced basket, crowded names to cash at 6%/yr, 8 bps turnover
cost; nulls include random, dollar-volume-matched, realized-vol-matched and
turnover-fair spell-preserving vetoes. Source: tietiy internal research
(banlist_unwind_2026-07, mwpl_veto_2026-07).