alphax/app/core/signal_direction.py

"""Direction-aware signal hygiene for long/short opportunities."""

from __future__ import annotations

from copy import deepcopy
from typing import Any

from app.core.signal_taxonomy import signal_code
from app.core.trade_direction import normalize_trade_side


LONG_SUPPORT_CODES = {
    "box_breakout_pullback_1h",
    "box_breakout_pullback_4h",
    "breakout_pullback_d1",
    "breakout_pullback_w1",
    "ignition_1h_current",
    "ignition_4h_current",
    "ignition_d1_current",
    "dynamic_k_1h_bull",
    "dynamic_k_d1_bull",
    "breakout_15m_current",
    "pullback_15m_confirm",
    "top_trader_long",
    "sector_rotation",
    "rs_strong",
    "rs_independent_strength",
    "funding_negative_contrarian",
    "vcp_bull_breakout",
    "vcp_bull_forming",
}

SHORT_SUPPORT_CODES = {
    "breakdown_retest_1h_short",
    "retest_reject_15m_short",
    "market_risk_off_short",
    "vcp_bear_breakdown",
    "vcp_bear_forming",
}

RISK_OR_CONTEXT_CODES = {
    "volume_divergence_1h",
    "entry_quality_gate",
    "liquidity_remove_risk",
    "exchange_inflow_risk",
    "holder_concentration_risk",
    "funding_extreme",
    "trend_exhaustion",
    "false_breakout",
    "high_position_reject",
    "risk_reward_bad",
    "rs_weak",
    "oi_divergence_risk",
    "funding_positive_risk",
    "tf_alignment_single_penalty",
    "tf_alignment_conflict_penalty",
    "vp_path_blocked",
    "breakout_quality_low",
}

LONG_SUPPORT_TEXT = (
    "大户偏多",
    "BTC回调中独立走强",
    "板块联动",
    "龙头",
    "箱体突破回踩",
    "底部箱体突破",
    "日线需求区反弹",
    "动K(阳)",
    "阳动K",
    "当前多头起爆",
    "VCP多头",
    "资金费率负值反向看多",
    "空头拥挤",
)


def is_factor_supportive_for_side(code: str, side: object, score_delta: object = 0) -> bool:
    """Whether a positive factor can support the requested trade side."""
    trade_side = normalize_trade_side(side)
    code = str(code or "").strip()
    try:
        delta = float(score_delta or 0)
    except Exception:
        delta = 0.0
    if delta <= 0 or code in RISK_OR_CONTEXT_CODES:
        return True
    if trade_side == "short" and code in LONG_SUPPORT_CODES:
        return False
    if trade_side == "long" and code in SHORT_SUPPORT_CODES:
        return False
    return True


def is_signal_supportive_for_side(signal: object, side: object) -> bool:
    """Whether a display signal belongs in the main evidence list."""
    trade_side = normalize_trade_side(side)
    text = str(signal or "").strip()
    if not text:
        return False
    code = signal_code(text)
    if trade_side == "short":
        if code in LONG_SUPPORT_CODES:
            return False
        normalized = text.replace(" ", "")
        if any(token.replace(" ", "") in normalized for token in LONG_SUPPORT_TEXT):
            return False
    elif code in SHORT_SUPPORT_CODES:
        return False
    return True


def sanitize_signals_for_side(signals: list[Any], side: object) -> tuple[list[Any], list[str]]:
    clean: list[Any] = []
    conflicts: list[str] = []
    seen_clean: set[str] = set()
    seen_conflict: set[str] = set()
    for signal in signals or []:
        text = str(signal or "").strip()
        if not text:
            continue
        if is_signal_supportive_for_side(text, side):
            if text not in seen_clean:
                clean.append(signal)
                seen_clean.add(text)
        elif text not in seen_conflict:
            conflicts.append(text)
            seen_conflict.add(text)
    return clean, conflicts


def excluded_factor_delta(items: list[dict[str, Any]], side: object) -> float:
    total = 0.0
    for item in items or []:
        code = str(item.get("factor_code") or "")
        delta = item.get("score_delta") or 0
        if not is_factor_supportive_for_side(code, side, delta):
            try:
                total += float(delta or 0)
            except Exception:
                pass
    return round(total, 6)


def sanitize_factor_breakdown_for_side(summary: dict[str, Any], side: object) -> tuple[dict[str, Any], list[dict[str, Any]]]:
    """Remove side-inconsistent positive factors from an existing summary."""
    src = deepcopy(summary or {})
    kept: list[dict[str, Any]] = []
    removed: list[dict[str, Any]] = []
    for item in src.get("items") or []:
        code = str(item.get("factor_code") or "")
        delta = item.get("score_delta") or 0
        if is_factor_supportive_for_side(code, side, delta):
            kept.append(item)
        else:
            removed.append(item)

    groups: dict[str, dict[str, Any]] = {}
    for item in kept:
        group = item.get("factor_group") or "other"
        bucket = groups.setdefault(group, {"score_delta": 0.0, "items": 0})
        try:
            bucket["score_delta"] = round(float(bucket["score_delta"]) + float(item.get("score_delta") or 0), 3)
        except Exception:
            pass
        bucket["items"] += 1
    opportunity_groups = {"momentum", "participation", "structure", "positioning", "narrative", "onchain_flow"}
    src["items"] = kept
    src["groups"] = groups
    src["total_delta"] = round(sum(float(i.get("score_delta") or 0) for i in kept), 3)
    src["opportunity_score"] = round(sum(float(v.get("score_delta") or 0) for k, v in groups.items() if k in opportunity_groups), 3)
    src["entry_score"] = round(float(groups.get("entry_quality", {}).get("score_delta") or 0), 3)
    src["risk_score"] = round(abs(min(0.0, float(groups.get("risk", {}).get("score_delta") or 0))), 3)
    if removed:
        src["direction_filtered"] = {
            "side": normalize_trade_side(side),
            "removed_factor_codes": [str(x.get("factor_code") or "") for x in removed],
        }
    return src, removed