Spaces:
Sleeping
Sleeping
File size: 4,355 Bytes
558db1e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 | import math
import pandas as pd
import numpy as np
import numpy.linalg as la
from config import Color, logger, hr
def check_and_fix_bounds(tickers, asset_min, asset_max, sector_limit, sector_map, macro=None, gross_leverage_cap=1.0, silent=False):
"""
Validates and automatically resolves conflicting portfolio constraints.
Integrates with HMM regime detection to dynamically disable leverage in high-volatility crashes.
"""
n = len(tickers)
safe_min = asset_min
adj_gross_cap = gross_leverage_cap
warns = []
# ─────────────────────────────────────────────
# REGIME DETECTION: Dynamic Risk Management
# ─────────────────────────────────────────────
is_high_vol = False
if macro and "hmm_regime" in macro:
is_high_vol = macro["hmm_regime"].get("is_high_vol", False)
if is_high_vol and (safe_min < 0 or adj_gross_cap > 1.0):
safe_min = max(0.0, safe_min) # Disable shorts
adj_gross_cap = 1.0 # Disable leverage
warns.append(f" {Color.YELLOW}⚠ HMM High Volatility Regime: Short selling and leverage dynamically disabled to protect capital.{Color.RESET}")
# ─────────────────────────────────────────────
# STANDARD BOUNDS LOGIC
# ─────────────────────────────────────────────
if safe_min < 0:
if n * asset_max < 1.0:
asset_max = math.ceil(100.0 / n) / 100.0
warns.append(f" {Color.YELLOW}⚠ Maximum weight per asset raised to {asset_max:.0%} so the weights can add up to 100%.{Color.RESET}")
max_gl = n * max(asset_max, abs(safe_min))
if max_gl > 3.0:
warns.append(f" {Color.YELLOW}⚠ Your settings allow up to {max_gl:.1f}× gross leverage — make sure this is intentional.{Color.RESET}")
else:
if safe_min * n > 0.999:
safe_min = 0.999 / n
warns.append(f" {Color.YELLOW}⚠ Minimum weight reduced to {safe_min:.2%} so all weights can still sum to 100%.{Color.RESET}")
sector_groups = {}
for t in tickers:
sector = sector_map.get(t, "Other")
if sector not in sector_groups:
sector_groups[sector] = []
sector_groups[sector].append(t)
for sector, members in sector_groups.items():
if sector == "Other":
continue
if safe_min * len(members) > sector_limit:
reduced = (sector_limit / len(members)) * 0.95
safe_min = min(safe_min, reduced)
warns.append(f" {Color.YELLOW}⚠ '{sector}' sector: minimum weight × {len(members)} tickers would exceed the {sector_limit:.0%} sector cap. Minimum reduced to {reduced:.2%}.{Color.RESET}")
if safe_min >= asset_max:
safe_min = 0.0
warns.append(f" {Color.YELLOW}⚠ Minimum weight was ≥ maximum weight — reset minimum to 0 to avoid an impossible constraint.{Color.RESET}")
present_sectors = set(sector_map.get(t, "Other") for t in tickers)
num_sectors = max(1, len(present_sectors))
if num_sectors * sector_limit < 0.99:
old_lim = sector_limit
sector_limit = min(1.0, (1.0 / num_sectors) + 0.02)
warns.append(f" {Color.YELLOW}⚠ Sector limit mathematically forced from {old_lim:.0%} to {sector_limit:.0%} to ensure 100% portfolio coverage.{Color.RESET}")
if warns and not silent:
hr("─", 65, Color.YELLOW)
print(f"{Color.BOLD}{Color.YELLOW} Constraint Adjustments{Color.RESET}")
for w in warns:
print(w)
hr("─", 65, Color.YELLOW)
return safe_min, asset_max, adj_gross_cap, sector_limit
def make_nearest_psd(matrix):
"""Finds the nearest positive-definite matrix to input."""
sym_matrix = (matrix + matrix.T) / 2
eigval, eigvec = np.linalg.eigh(sym_matrix)
eigval[eigval < 0] = 1e-8
return eigvec @ np.diag(eigval) @ eigvec.T
|