web / pmct /ledger.py
LinJennyLin's picture
Deploy SharpFlow (nav + example wallets + Polymarket copy)
d4c51f2 verified
Raw History Blame Contribute Delete
4.07 kB
"""Normalize a wallet's raw activity feed into a tidy trade ledger.
The Data API returns loosely-typed dicts whose exact keys drift over time, so
every field is read with fallbacks. The output is one DataFrame per wallet with
a stable schema the rest of the pipeline relies on.
Cashflow convention (from the trader's perspective):
BUY -> cash out (negative)
SELL -> cash in (positive)
REDEEM / REWARD -> cash in (positive) # market resolved in their favor
Open positions are marked to market separately in metrics.py.
"""
from __future__ import annotations
import pandas as pd
LEDGER_COLS = [
"ts", "type", "side", "market", "asset", "outcome",
"size", "price", "usdc", "cashflow", "title",
]
# activity types that bring cash IN (winning redemptions + protocol incentives)
CASH_IN_TYPES = ("REDEEM", "CLAIM", "REWARD", "MAKER_REBATE", "TAKER_REBATE")
# the uncopyable subset: liquidity rewards + maker/taker rebates (market-making)
REBATE_TYPES = ("REWARD", "MAKER_REBATE", "TAKER_REBATE")
def _first(d: dict, *keys, default=None):
for k in keys:
if k in d and d[k] not in (None, ""):
return d[k]
return default
def _to_float(x, default=0.0) -> float:
try:
return float(x)
except (TypeError, ValueError):
return default
def normalize_activity(raw: list[dict]) -> pd.DataFrame:
rows = []
for a in raw:
typ = str(_first(a, "type", "activityType", default="TRADE")).upper()
side = str(_first(a, "side", default="")).upper()
size = _to_float(_first(a, "size", "shares", "amount"))
price = _to_float(_first(a, "price", "avgPrice"))
usdc = _first(a, "usdcSize", "usdc", "usdcValue", default=None)
usdc = _to_float(usdc) if usdc is not None else round(size * price, 6)
# ts may be unix seconds or ms
ts_raw = _to_float(_first(a, "timestamp", "ts", "time"))
if ts_raw > 1e12:
ts_raw /= 1000.0
ts = pd.to_datetime(ts_raw, unit="s", utc=True) if ts_raw else pd.NaT
# Protocol incentives (liquidity rewards + maker/taker rebates) are real
# income but NOT copyable directional skill — tag them so metrics can
# measure how rebate-driven a wallet is (market makers = uncopyable).
if typ in CASH_IN_TYPES or side == "SELL":
cashflow = +usdc
elif side == "BUY":
cashflow = -usdc
else: # unknown non-trade rows contribute no directional cashflow
cashflow = 0.0
rows.append({
"ts": ts,
"type": typ,
"side": side,
"market": _first(a, "conditionId", "condition_id", "market", default=""),
"asset": _first(a, "asset", "tokenId", "token_id", default=""),
"outcome": _first(a, "outcome", "outcomeIndex", default=""),
"size": size,
"price": price,
"usdc": usdc,
"cashflow": cashflow,
"title": _first(a, "title", "slug", "question", default=""),
})
df = pd.DataFrame(rows, columns=LEDGER_COLS)
if not df.empty:
df = df.dropna(subset=["ts"]).sort_values("ts").reset_index(drop=True)
return df
def attach_open_positions(positions: list[dict]) -> pd.DataFrame:
"""Current open positions -> per-market marked-to-market value & PnL."""
rows = []
for p in positions:
size = _to_float(_first(p, "size", "shares"))
avg = _to_float(_first(p, "avgPrice", "averagePrice"))
cur = _to_float(_first(p, "curPrice", "currentPrice", "price"))
rows.append({
"market": _first(p, "conditionId", "condition_id", "market", default=""),
"asset": _first(p, "asset", "tokenId", default=""),
"size": size,
"avgPrice": avg,
"curPrice": cur,
"cost_basis": size * avg,
"mark_value": size * cur,
"cashPnl": _to_float(_first(p, "cashPnl", default=size * (cur - avg))),
})
return pd.DataFrame(rows)