holdings-import

holdings-import

Prerequisites

User has dropped CSVs into imports/YYYY-MM-DD/:

• kite-tradebook.csv (required for lot entry dates)
• kite-holdings.csv (optional, for reconciliation)
• coin-holdings.csv (optional — current Coin MF balance)
• indmoney-trades.csv (required)
• indmoney-holdings.csv (optional)

If any REQUIRED file is missing, abort and tell the user what's needed.

Steps

1. Find the latest import folder:

ls -1d imports/*/ | sort -r | head -1

2. Parse each file and build lots per account. FIFO runs per (account, symbol) — do not pool trades across accounts. Invoke:

python - <<'PY'
import yaml
from datetime import date
from utils.csv_parsers import parse_kite_tradebook, parse_indmoney_trades
from utils.fifo import build_lots

# Each account gets its own trade stream + enrichment metadata.
per_account = [
    {"id": "kite-eq",     "market": "IN", "instrument": "equity", "currency": "INR",
     "trades": parse_kite_tradebook("imports/LATEST/kite-tradebook.csv")},
    {"id": "indmoney-us", "market": "US", "instrument": "equity", "currency": "USD",
     "trades": parse_indmoney_trades("imports/LATEST/indmoney-trades.csv")},
]

all_lots = []
for acc in per_account:
    lots = build_lots(acc["trades"])
    for l in lots:
        lot = {
            "account": acc["id"],
            "symbol": l.symbol,
            "market": acc["market"],
            "instrument": acc["instrument"],
            "qty": l.qty,
            "entry_date": l.entry_date.isoformat(),
        }
        if acc["currency"] == "INR":
            lot["avg_cost_inr"] = l.cost_basis / l.qty
            lot["cost_basis_inr"] = l.cost_basis
        else:
            lot["avg_cost_usd"] = l.cost_basis / l.qty
            lot["cost_basis_usd"] = l.cost_basis
        all_lots.append(lot)

doc = {
    "as_of": str(date.today()),
    "fx": {"USDINR": 0.0},  # filled by portfolio-state at read time
    "lots": all_lots,
}
print(yaml.safe_dump(doc, sort_keys=False))
PY

Replace LATEST with the actual folder.

3. USD→INR cost basis enrichment. For each US-currency lot in the output, fetch the USDINR rate on the lot's entry_date via mcp__market-data__history (or a fallback spot if historical is unavailable) and populate avg_cost_inr and cost_basis_inr alongside the USD fields. Indian tax computations require INR basis.

4. Reconcile the computed open-lot totals against kite-holdings.csv (current qty per symbol). Emit a diff report. If a symbol in kite-holdings.csv has no derived lot OR quantities disagree by >0.5%, flag loudly.

5. Write the assembled doc to portfolio/holdings.yaml.

6. Emit a diff vs the prior holdings.yaml (if any): new lots, closed lots, quantity changes.

Upgrade hook: Kite MCP

If portfolio/accounts.yaml sets source: mcp for any Kite-based account AND the Kite MCP is reachable (mcp__kite__* tools present), skip CSV parsing for that account and call the MCP's holdings endpoint instead. Resolve trades/lots from the MCP's position + transaction endpoints. Output contract is identical.

Rules

• Do not write holdings.yaml if any required input is missing.
• Do not write holdings.yaml if reconciliation fails — write to holdings.yaml.proposed and ask the user to review.