Results se pehle run-up? 7,940 events ka sach

6 min read · data: 437-stock NSE universe · descriptive study, not advice

Results season aate hi ek line har jagah sunai deti hai: “results se pehle stock mein run-up aata hai.” Iske peeche ek asli academic finding bhi hai — US research ke mutabik earnings se judi extra return ka bada hissa (ek study ke hisaab se ~71%) announcement se pehle ke ~10 trading dino mein aata hai.

Sawaal yeh tha: kya yeh India mein bhi dikhta hai? Humne measure kiya. Short answer: raw return positive hai, lekin woh results ki wajah se nahi — woh bas market ka normal chalna hai, results ka costume pehen ke. Yeh measurement hai, prediction nahi.

Pehle dates sahi kiye

Is tarah ki study ka sabse bada problem hota hai: results ki exact date. Galat date ho toh poora test bekaar. Humne do sources milaye:

• yfinance NSE earnings dates (~2001 se), aur
• IndianAPI board-meeting archive (~2021 se).

Dono sources 2 din ke andar match karein toh same event. Kul 10,987 real announcement dates mile; 3,499 dono sources se cross-verified. Spot-checks (RELIANCE ki saari 10 dates 2024–26, INFY Apr 2024, HDFCBANK Jan 2025, ADANIPORTS Jan 2025, RAMCOCEM Feb 2024) public records se match hue. Price data ke saath milane ke baad test ke liye 7,940 events, 187 F&O stocks, 2014–2026 bache.

Test kya tha (pehle se likha hua)

• Entry: results date se 10 trading din pehle ka close.
• Exit: results se 1 din pehle ka close (results ke through hold nahi).
• 0.08% friction (futures round-trip) minus.
• Do comparisons: (a) usi window mein poore F&O universe ka average return (drift), (b) usi stock ke random 9-din windows.
• Significance results-season clusters ke hisaab se (ek quarter ke saare events ek saath aate hain, unko independent maanna galat hota).

Result

Teen baatein saaf dikhti hain:

1. Raw return positive hai: +0.79%. Yahi number dekh ke log kehte hain “dekho, run-up hota hai.”
2. Usi 9 din mein poora F&O universe +0.69% chala. Matlab +0.79% ka lagbhag 87% sirf market ka drift tha — koi bhi F&O stock 9 din hold karte toh yeh milta.
3. Jo thoda extra bacha (+0.10%), woh bhi time ke saath zero ho gaya. 2023 ke baad ke dono periods mein excess +0.01–0.02% hai. Brokerage ke baad negative.

Aur random-window test mein toh 2023–24 mein results se pehle ka window, usi stock ke random 9-din windows se kharab tha (+0.48% vs +1.28%).

”Lekin mid-caps mein toh dikha?”

Humne kuch cuts bhi dekhe — large vs mid-cap, high vs low beta, quarter wise. Ek cut positive aaya: mid/small-cap, +0.19% excess (p=0.03). Sunne mein achha, lekin:

• ~15 cuts test kiye; ek ka p=0.03 aana chance se expected hai.
• Uska large-cap sibling bilkul flat tha (+0.005%).
• Period splits mein survive nahi kiya.
• Mid/small-cap mein asli friction 0.20%+ hota hai, jo isko negative kar deta hai.

Interesting baat: Q1 (April–June results, sabse heavy season) mein raw return hi negative tha (−0.17%). Q2/Q3/Q4 achhe dikhte hain, lekin woh drift-contaminated net numbers hain.

Iska matlab kya hai

Yeh hum “drift in costume” kehte hain: ek asli phenomenon (market ka upar jaana) kisi kahani (results run-up) ka costume pehen leta hai. Jab tak aap drift subtract nahi karte, dono alag nahi dikhte.

Results season mein jo sach mein badalta hai woh hai volatility — results ke din aur aas-paas moves bade hote hain. Humare research mein kuch stocks ke liye results “binary” events the (bade jhatke), kuch ke liye non-events. Yeh risk ka sawaal hai, direction ka nahi.

Aur results ke baad? (PEAD)

Results ke baad ka famous idea hai PEAD — post-earnings announcement drift: achhe results pe jo stock upar gaya, woh agle kuch hafte aur chalta hai. Humne yeh bhi test kiya: 3,984 results events (2024–2026), reaction ke baad agle din ke open pe entry, 5/10/21 din hold, aur wahi drift benchmark.

Pehli, kachhi report mein “+0.87% net, t=2.74” aaya tha — achha lagta tha. Lekin us window mein equal-weight market khud +0.94% chala tha. Drift subtract karne ke baad: training period mein +0.11%, test period mein −0.14%. 18 cells mein se ek bhi multiple-testing correction ke baad nahi bacha. Yaani results ke pehle aur baad — dono taraf ki kahani drift ka costume nikli.

Kya dekhein (kya kharidein nahi)

• Upcoming results dates humare /stocks/ pages aur event calendar pe dikhte hain — unhe volatility ke din ki tarah dekhiye.
• Koi “pre-results run-up” ka chart dikhaye toh poochiye: usi window mein market ne kya kiya?
• Results ke baad wale drift ko bhi humne alag se test kiya (neeche dekhiye).

The limitation

Yeh study sirf 187 F&O stocks pe hai, chhote non-F&O stocks pe nahi. Results ke content (beat/miss) ko yeh test nahi dekhta — sirf calendar window. Universe current listed stocks se bana hai. Base rates past describe karte hain; future predict nahi karte.

Methodology note. Events: first trading bar on/after the announcement date; entry close[D−10], exit close[D−1]; same-symbol events < 7 days apart de-duplicated; 0.08% friction. Nulls: (a) date-matched equal-weight F&O universe drift over each event’s window, (b) 2,000-draw same-stock random non-event 9-day windows. Inference: one-sided cluster-robust t, clustered by results-season year-quarter (49 clusters). Dates: yfinance + IndianAPI reconciled within 2 calendar days. Source: tietiy internal research (pre_earnings_premium_2026-07, tested 2026-07-24; pead_honest_validation_2026-07-18, 3,984 events, train 2024-01 → 2025-06, test 2025-07 → 2026-05).