Zen v1
Systematic · inception 5 Oct 2026
Zen v1 ranks NSE companies on a blend of fundamental and market factors and holds a concentrated portfolio of its highest ranked names. It reviews the portfolio on a fixed schedule and only ever trades on information that was public before the open. I wrote the rules down before I ran the backtest, kept the last three and a half years back for an out-of-sample test, and have not changed a rule since. The rules themselves stay private.
It started from cash on 5 October 2026 with the same rebalance the backtest ran, and its scheduled reviews continue from there.
Zen v3
Systematic, purely fundamental · inception not set
Zen v3 is built on company fundamentals alone, looking for improving businesses at a sensible price with firm limits on balance sheet risk. The rules were fixed before any test was run, and the portfolio will start from cash once the backtest has been completed and checked independently.
Keeping the record honest
Every decision is written to a file and timestamped before the market opens on the day it trades. The live pages show only the paper record from each portfolio's own start date, and nothing on them is ever restated. The backtest has its own pages and never shares a chart or a table with the live record, so simulated and real results cannot be mixed up.
Orders fill at the opening price of the next NSE session, with 0.1% slippage on each side. Every fill carries the full set of Indian charges, from STT and stamp duty to exchange fees, GST and depository charges, itemised and stored with the trade.
I only buy whole shares, sell before I buy and never let cash go negative. There is no leverage and no shorting, and idle cash earns nothing. If a stock does not trade at the open, the order waits up to five sessions and is then cancelled.
Dividends and corporate actions
Splits and bonuses adjust the holding before the ex-date's open, and any fractional share is paid out in cash. Dividends are credited on the ex-date to the shares held at the previous close, before tax, which is how funds report them. Anything the code cannot handle on its own, such as a demerger, is flagged on the dashboard rather than guessed.
Each portfolio starts with fifty lakh rupees in cash and is valued at every close from the same NSE prices the research archive uses. A stock that did not trade is marked at its last close and flagged.
I measure performance against the Nifty 500 on a total return basis, and show the Nifty 50, the midcap and smallcap indices and NSE's factor indices alongside it. Risk statistics need time, so anything based on fewer than sixty trading days is greyed out, and annualised return waits for a full year of history.
Every decision is saved with its own SHA-256 hash and committed before the open it trades at. Fills and corporate actions go into append-only ledgers in which each line is chained to the hash of the line before it, so nothing can be edited or removed quietly. Everything on this site is rebuilt from that ledger every evening, and a separate check confirms that rebuilding from the ledger gives the same result.
An automated job runs at 8 in the morning IST on weekdays and makes any decision that is due before the market opens. A second job runs at 9 in the evening, once the day's NSE data is in. It fills the day's orders, applies corporate actions, values each portfolio and rebuilds this site.
This is a read-only public copy of my private dashboard, rebuilt by the same evening job. It leaves out my discretionary portfolio and has no trade controls, and the full version sits behind Cloudflare Access. The pages load only their own files and one pinned charting library, with no trackers.