Method · Trending Value strategy

The rules, exactly

Trending Value comes from James O'Shaughnessy's What Works on Wall Street (4th edition, 2011). tickerbeat runs it on NSE stocks with a faster, more concentrated default: the cheapest 5% rather than the book's 10%, the 10 strongest rather than 25, and a rotation every 6 weeks on 4-week momentum, where the book rebalances once a year on 6-month momentum. The book's settings are one click away on the backtest.

1. Universe

Main-board NSE equities (series EQ and BE) that traded in the week and are worth at least ₹500 crore. SME platform stocks, ETFs and anything without quarterly results are left out. The size cut can be changed from ₹100 crore up.

2. Value Composite Two

Each stock is ranked against the rest of the universe on six ratios and given a percentile score from 1 (cheapest) to 100 (dearest):

  • P/E — market cap ÷ trailing-12-month net profit attributable to shareholders.
  • P/S — market cap ÷ trailing-12-month revenue from operations.
  • P/B — market cap ÷ shareholders' equity from the latest half-yearly balance sheet.
  • EV/EBITDA — (market cap + borrowings − cash) ÷ trailing-12-month EBITDA (profit before tax + finance costs + depreciation).
  • Price to cash flow — market cap ÷ (net profit + depreciation), the book's definition of cash flow.
  • Shareholder yield — dividends with an ex-date in the last 12 months ÷ price, plus the fall in share count over 12 months (buyback yield; a rising share count counts as zero).

Ratios are ranked as yields (earnings ÷ price and so on), so loss-makers and negative book values land at the expensive end automatically. A stock missing a ratio scores a neutral 50 on it. EV/EBITDA doesn't apply to lenders, so banks, NBFCs and insurers always score 50 there. The six scores are added (6 to 600) and the lowest 5% — the cheapest — go through (the book takes 10%).

3. The beat

From that cheapest 5%, the 10 stocks with the highest 4-week total return (price change with splits and bonuses undone and dividends reinvested) make the portfolio, in equal amounts. Ties go to the larger company.

4. Rotation

Every 6 weeks, on the last trading day of the week, the portfolio is rebuilt from scratch: new stocks are bought, dropped ones sold, and the survivors trimmed or topped up back to equal weight. Between rotations nothing changes. 6 weeks means about 9 rotations a year, so the backtest charges trading costs on every one, and can apply capital gains tax (almost every exit is short-term).

Data

  • Prices: NSE's daily bhavcopy, taking each week's last trading day.
  • Results: the XBRL files of every quarterly results filing on NSE (Financial Results to Dec 2024, Integrated Filing from Mar 2025). Consolidated figures are used when the last four quarters are all available consolidated; otherwise standalone.
  • Corporate actions: NSE's list of dividends, splits, bonuses and buybacks, used for total returns, share counts and dividend yield. A few splits and consolidations NSE never lists are added by hand.
  • Demergers: the shares of the new company aren't tracked, so the week of the ex-date counts as no change. If the demerger took more than a fifth of the price, the stock has no value ratios until all four quarters of its trailing results end after the ex-date, because earlier results still include the business that left.
  • Benchmarks: NSE's daily index closes (Nifty 500 is a price index; it excludes dividends, about 1–1.5% a year).

A result is only used from the day after it was filed, and only the first version published, so the backtest never sees numbers before the market did. Shares outstanding come from paid-up capital ÷ face value in the latest filing, adjusted for any later split or bonus.

Data as of 20 Nov 2026: 2,271 stocks with a market cap, 1,690 over ₹500 crore; 87% have all six ratios. Weekly history from 5 Jan 2018.

Known limitations

  • Short history. Free XBRL results start in mid-2018, so the backtest begins in 2019 — one value cycle, not forty-five years.
  • No balance sheets before mid-2022. NSE's XBRL results only carry the balance sheet from the September 2022 half-year onward. Before that, P/B and EV/EBITDA are missing for every stock, so all score a neutral 50 and the composite effectively ranks on the other four ratios.
  • Filing errors. Figures are read straight from company filings. A wrong unit or a mis-tagged line flows into the ratios; percentile ranking limits the damage but doesn't remove it.
  • Bank revenue is total income (interest plus other income), and bank cash flow is net profit alone, because the banking template reports depreciation only in the cash-flow statement.
  • Weekly prices. Everything is marked at the week's last close. A stock that stops trading keeps its last price.
  • Rights issues aren't adjusted for in returns; splits, bonuses and consolidations are.

Updates

Prices refresh every trading evening and results as they're filed; the published portfolio only changes on rotation days. The code that ranks stocks on this site is the same code that runs the backtest.