# shared from s16_ec.py (definitions only)
# EXPIRY-CHEAP-COUNTER (PREREG 20:05 IST 27-Oct-2026). Signals, frequency facts, buyer exits a/b/c, controls. Real bars only; spot >=15:15 missing.
import pandas as pd,numpy as np,glob,os
L='/workspace/research/popular-strategies'; C=f'{L}/data/ec_cache'; CD='/workspace/research/common/data'
rng=np.random.default_rng(20261027)
DH={ix:pd.read_parquet(f'{CD}/{ix.lower()}_daily_hist.parquet').assign(date=lambda x:pd.to_datetime(x.date)).set_index('date') for ix in ('NIFTY','SENSEX')}
DH['NIFTY'].loc[pd.Timestamp('2026-10-27'),['open','close']]=[22708.80,22749.75]   # published official close (ET Now / Business Standard / Yahoo)
SP={}
for ix in ('NIFTY','SENSEX'):
    d=pd.read_parquet(f'{CD}/{ix.lower()}_1m_clean.parquet'); d['mi']=d.timestamp.dt.hour*60+d.timestamp.dt.minute-555
    for dt,g in d.groupby('date'):
        a=np.full(375,np.nan); ok=(g.mi>=0)&(g.mi<375)&(~g.missing); a[g.mi[ok].values]=g.close[ok].values; SP[(ix,dt)]=a
e=pd.read_parquet('/workspace/market-live-stack/bots/upstox/eod/2026-10-27/nifty_1m_2026-10-27.parquet'); ts=pd.to_datetime(e.timestamp); a=np.full(375,np.nan); a[(ts.dt.hour*60+ts.dt.minute-555).values]=e.close.values; SP[('NIFTY',pd.Timestamp('2026-10-27'))]=a
for k in SP: SP[k][360:]=np.nan          # spot 15:15-15:29 treated as missing (uniform rule)
LAB=pd.read_csv(f'{L}/results/regime_labels.csv',parse_dates=['date']).set_index('date')
W={'W1':(20,105),'W2':(105,225),'W3':(225,315),'W4':(315,345),'ALL':(20,345)}
def lot(ix,t):
    if ix=='NIFTY': return 25 if t<pd.Timestamp('2024-11-20') else (75 if t<pd.Timestamp('2026-01-01') else 65)
    return 10 if t<pd.Timestamp('2025-01-01') else 20
def chg(ix,t,buy,sells):
    Q=lot(ix,t); n=1+sum(1 for f,p in sells if p>0); brk=20.0*n; sv=sum(f*p for f,p in sells)*Q; bv=buy*Q
    stt=(0.0015 if t>=pd.Timestamp('2026-04-01') else 0.001)*sv; tx=(0.0003503 if ix=='NIFTY' else 0.000325)*(bv+sv); sebi=1e-6*(bv+sv); st=3e-5*bv; gst=0.18*(brk+tx+sebi)
    return (brk+stt+tx+sebi+st+gst)/Q
def load(ix,dt):
    d=pd.read_parquet(f'{C}/{ix}_{dt.date()}.parquet'); A={}
    for (t,K),g in d.groupby(['typ','K']):
        r=np.full((6,375),np.nan); m=g.mi.values; r[0,m]=g.o; r[1,m]=g.h; r[2,m]=g.l; r[3,m]=g.c; r[4,m]=g.v; r[5,m]=g.oi; A[(t,K)]=r
    return A,d.src.iloc[0]
def trend(s):
    tr=np.zeros(375,int); o=s[0] if not np.isnan(s[0]) else np.nan
    for t in range(20,346):
        if np.isnan(s[t]) or np.isnan(s[t-20]): continue
        up=s[t]>OPEN and s[t]-s[t-20]>0; dn=s[t]<OPEN and s[t]-s[t-20]<0; tr[t]=1 if up else -1 if dn else 0
    return tr
def pick(A,typ,t,spot,lo,hi,decay=False):
    best=None
    for (ty,K),r in A.items():
        if ty!=typ: continue
        if (typ=='PE' and K>=spot) or (typ=='CE' and K<=spot): continue
        c=r[3,t]
        if np.isnan(c) or c<lo or c>hi: continue
        if not np.isnan(r[4,t]) and r[4,t]<=0: continue
        if decay and not (np.nanmax(r[3,:t]) if t>0 and np.any(~np.isnan(r[3,:t])) else 0)>=2*hi: continue
        key=(abs(c-(lo+hi)/2),abs(K-spot))
        if best is None or key<best[0]: best=(key,K)
    return None if best is None else best[1]
def outcome(ix,dt,A,typ,K,t,slip,zero):
    r=A[(typ,K)]; ref=r[3,t]; hh=r[1,t+1:]; out=dict(ref=ref,sig_vol=r[4,t])
    mx=np.nanmax(hh) if np.any(~np.isnan(hh)) else np.nan; out['max_ratio']=mx/ref
    out['peak_min']=int(np.nanargmax(hh))+1 if not np.isnan(mx) else np.nan
    Sx=DH[ix].close.get(dt,np.nan); settle=max(0.0,(Sx-K) if typ=='CE' else (K-Sx)) if not np.isnan(Sx) else np.nan; out['settle']=settle; out['near_zero']=settle<=zero if not np.isnan(settle) else np.nan
    # fill: next 1m open (first real open within 5 min)
    f=None
    for y in range(t+1,min(t+6,375)):
        if not np.isnan(r[0,y]): f=y; break
    if f is None: out['status']='no_fill'; return out
    fill=r[0,f]+slip; out.update(status='ok',fill_min=f,fill=fill,fill_vol=r[4,f]); last=max([y for y in range(f,375) if not np.isnan(r[3,y])])
    out['last_close']=r[3,last]
    # (a) half at 2x, half at 3x or close
    s1=s2=None
    for y in range(f,last+1):
        if np.isnan(r[1,y]): continue
        if s1 is None and r[1,y]>=2*fill: s1=2*fill
        if r[1,y]>=3*fill: s2=3*fill; s1=s1 or 2*fill; break
    if s1 is None: s1=r[3,last]-slip
    if s2 is None: s2=r[3,last]-slip
    s1=max(s1,0.05); s2=max(s2,0.05); out['pnl_a']=0.5*(s1-fill)+0.5*(s2-fill)-chg(ix,dt,fill,[(0.5,s1),(0.5,s2)]); out['gross_a']=0.5*(s1+s2)-fill
    # (b) stop 50% (low, first), trail 50% of peak gain after gain>=2R (R=0.5 fill), on closes
    R=0.5*fill; peak=0; ex=None; why='close'
    for y in range(f,last+1):
        if np.isnan(r[2,y]): continue
        if r[2,y]<=fill-R and y>f or (y==f and r[2,y]<=fill-R): ex=fill-R-slip; why='stop'; break
        g=r[3,y]-fill; peak=max(peak,r[1,y]-fill if r[1,y]-fill>=2*R else g)
        if peak>=2*R and g<=0.5*peak: ex=r[3,y]-slip; why='trail'; break
    if ex is None: ex=r[3,last]-slip
    ex=max(ex,0.05); out['pnl_b']=ex-fill-chg(ix,dt,fill,[(1,ex)]); out['why_b']=why
    # (c) settlement intrinsic, no slippage
    out['pnl_c']=(settle-fill-chg(ix,dt,fill,[(1,settle)])) if not np.isnan(settle) else np.nan
    return out