BASE RATE
Method: har figure kaise banta hai, aur kahan ruk jaata hai.
Har study pehle ek public log mein likhi jaati hai, uske baad hi data chhua jaata hai. Har event ke saamne usi date ke milte-julte securities rakhe jaate hain, taaki bazaar ki apni chaal alag ho sake. Phir farak ke saath uncertainty ka interval chhapta hai. Har study ke figures uske English page par hain.
Pattern pehle register hota hai, phir compute. Ulta kabhi nahi.
Har event ke saath usi date ka matched control chalta hai.
Universe point-in-time hai — delist ho chuke naam bhi andar rehte hain.
Jo studies kuch nahi dikha payin, woh bhi usi list mein chhapti hain.
TAREEKA
Pehle likho, phir gino
Har pattern ki poori definition, uska poora parameter grid aur uske horizons ek dated public log mein likhe jaate hain — data chhune se pehle. Headline us grid ke pehle se tay kiye gaye beech wale setting par chhapta hai, us cell par nahi jo baad mein sabse achha dikha.
Yeh kram hi poora argument hai. Agar setting natija dekhne ke baad chuni jaaye, toh number kisi cheez ka saboot nahi rehta — woh sirf yeh batata hai ki kitni koshishon mein se ek achhi nikli.
CONTROL
Har number ke saath ek placebo
Har event ke liye usi date ke aise securities chune jaate hain jo shart poori nahi karte the, lekin price band, volatility band aur traded-value band mein us jaise hi the. Unhe bilkul usi tareeke se naapa jaata hai. Wahi placebo hai.
Iski zaroorat isliye hai ki is daur mein bazaar 3 mein se lagbhag 2 rolling 20-din wali windows mein upar gaya. Aisi haalat mein koi bhi upar ki taraf wali shart apne aap ek badi dikhne wali frequency de degi. Isliye chhapne laayak number sirf farak hai — pattern aur uske control ke beech ka.
GINTI
Sample kitna dikhta hai, aur kitna hai
Ek hi din par bane events aapas mein azaad nahi hote — un sabne ek hi bazaar ki chaal share ki. Isliye raw ginti sample ko bada dikhati hai. Effective sample size wahi imaandaar ginti hai jo iss clustering ko hata kar bachti hai, aur har interval aur har gate usi par chalta hai.
Interval khud bootstrap se aata hai: data ko baar baar dobara khincha jaata hai, poori dates ko blocks mein le kar, taaki ek hi din ka clustering aur aage-peeche overlap dono uske andar aa jaayein.
SEEMA
Yeh method kya nahi kar sakta
- Yeh nahi batata ki agli baar kya hoga. Aage aane wala din us sample ka hissa hi nahi tha.
- Yeh kisi security ko pehchanta nahi. Har figure sainkdon naamon par saalon ka jod hai.
- Yeh aapki laagat, aapka size aur aapke tax nahi ginta — na hi haar ke daur mein tike rehne ki taakat.
- Control se farak apne aap koi edge nahi hai. Woh ek sample ka observation hai, uncertainty ke saath.
FIGURE
Figures kahan milenge
Har study ke figures, uska interval aur effective sample uske English page par hain.
In Hinglish pages par har study ki shart, grid, control aur limits samjhayi gayi hain.
Woh log jisme sab pehle likha gaya tha
Method ka sabse zaroori hissa kram hai. Registration log wahi kram dikhata hai — kaunsa pattern kab likha gaya, kis grid ke saath, aur uska status abhi kya hai.