scripts/market_metrics.py

import numpy as np
import pandas as pd
from typing import Tuple
import time
from tqdm import tqdm
from utils import convert_hex_to_int, ROOT_DIR, TMP_DIR
from get_mech_info import read_all_trades_profitability
from tools_metrics import prepare_tools
from predict_kpis import compute_markets_agent_roi


def determine_market_status(row):
    current_answer = row["currentAnswer"]
    """Determine the market status of a trade."""
    if (current_answer is np.nan or current_answer is None) and time.time() >= int(
        row["openingTimestamp"]
    ):
        return "pending"
    if current_answer is np.nan or current_answer is None:
        return "open"
    if row["fpmm.isPendingArbitration"]:
        return "arbitrating"
    if row["fpmm.answerFinalizedTimestamp"] and time.time() < int(
        row["fpmm.answerFinalizedTimestamp"]
    ):
        return "finalizing"
    return "closed"


def compute_market_metrics(all_trades: pd.DataFrame):
    print("Preparing dataset")
    all_trades.rename(
        columns={
            "fpmm.currentAnswer": "currentAnswer",
            "fpmm.openingTimestamp": "openingTimestamp",
            "fpmm.id": "market_id",
        },
        inplace=True,
    )
    all_trades["currentAnswer"] = all_trades["currentAnswer"].apply(
        lambda x: convert_hex_to_int(x)
    )
    all_trades["market_status"] = all_trades.apply(
        lambda x: determine_market_status(x), axis=1
    )
    closed_trades = all_trades.loc[all_trades["market_status"] == "closed"]
    print("Computing metrics")
    nr_trades = (
        closed_trades.groupby("market_id")["id"].count().reset_index(name="nr_trades")
    )
    total_traders = (
        closed_trades.groupby("market_id")["trader_address"]
        .nunique()
        .reset_index(name="total_traders")
    )
    final_dataset = nr_trades.merge(total_traders, on="market_id")
    markets = closed_trades[
        ["market_id", "title", "market_creator", "openingTimestamp"]
    ]
    markets.drop_duplicates("market_id", inplace=True)
    market_metrics = markets.merge(final_dataset, on="market_id")
    print("Saving dataset")
    market_metrics.to_parquet(ROOT_DIR / "closed_market_metrics.parquet", index=False)
    print(market_metrics.head())


def prepare_traders_data() -> Tuple[pd.DataFrame, pd.DataFrame]:
    """Prepare traders data for weekly metrics computation."""
    trades = read_all_trades_profitability()
    trades["creation_timestamp"] = pd.to_datetime(trades["creation_timestamp"])
    trades["creation_timestamp"] = trades["creation_timestamp"].dt.tz_convert("UTC")
    trades["creation_date"] = trades["creation_timestamp"].dt.date
    trades = trades.sort_values(by="creation_timestamp", ascending=True)
    unique_addresses = trades.trader_address.unique()
    closed_markets = trades.title.unique()
    # filter the mech requests done these traders on closed markets
    try:
        tools_df = pd.read_parquet(TMP_DIR / "tools.parquet")
        tools_df = prepare_tools(tools_df, total_included=False)
    except Exception as e:
        print(f"Error reading tools parquet file {e}")
        return None

    traders_activity = tools_df[
        tools_df["trader_address"].isin(unique_addresses)
    ].copy()
    traders_activity = traders_activity[traders_activity["title"].isin(closed_markets)]
    traders_activity["request_time"] = pd.to_datetime(
        traders_activity["request_time"], utc=True
    )
    traders_activity = traders_activity.sort_values(by="request_time", ascending=True)
    traders_activity["request_date"] = traders_activity["request_time"].dt.date
    return trades, traders_activity


def compute_weekly_trader_metrics():
    trades_data, mechs_data = prepare_traders_data()
    trades_data["week_start"] = (
        trades_data["creation_timestamp"].dt.to_period("W").dt.start_time
    )

    grouped_trades = trades_data.groupby("week_start")
    contents = []
    traders = trades_data.trader_address.unique()
    # Iterate through the groups (each group represents a week)
    for week, week_data in grouped_trades:
        print(f"Week: {week}")  # Print the week identifier

        # for all closed markets
        closed_markets = week_data.title.unique()
        for trader in tqdm(
            traders, total=len(traders), desc="Computing metrics for traders"
        ):
            # compute all trades done by the trader on those markets, no matter from which week
            trader_trades = trades_data[
                (trades_data["trader_address"] == trader)
                & (trades_data["title"].isin(closed_markets))
            ]
            if len(trader_trades) == 0:
                # no trading activity
                continue

            # filter mech requests done by the trader on those markets
            trader_mech_calls = mechs_data.loc[
                (mechs_data["trader_address"] == trader)
                & (mechs_data["title"].isin(closed_markets))
            ]
            # compute the ROI for that market, that trader and that week
            try:
                # Convert the dictionary to DataFrame before appending
                roi_dict = compute_markets_agent_roi(
                    trader_trades, trader_mech_calls, trader, "week", week
                )
                contents.append(pd.DataFrame([roi_dict]))
            except ValueError as e:
                print(f"Skipping ROI calculation: {e}")
                continue
    traders_weekly_metrics = pd.concat(contents, ignore_index=True)
    return traders_weekly_metrics


if __name__ == "__main__":
    all_trades = pd.read_parquet(TMP_DIR / "fpmmTrades.parquet")
    compute_market_metrics(all_trades)
    weekly_metrics_df = compute_weekly_trader_metrics()
    weekly_metrics_df.to_parquet(
        ROOT_DIR / "traders_weekly_metrics.parquet", index=False
    )