Breakout

Darvas Box

Nicolas Darvas, a dancer who made a fortune in the 1950s, traded stocks near their highs that paused in a tight 'box' of prices. He waited for the price to rise out of the box on volume, and stayed while the boxes kept stepping up.

Yearly return, after costs+16.2%Nifty 500 +11.5% over the same years

Worst fall−35.0%Win rate39%Matches now0

How it works

The signal every rule true at the close

The past 15 days traded in a box no taller than 12%
The top of the box is within 5% of the 52-week high
Close breaks above the top of the box
Volume at least 1.5 times the 20-day average

The exit whichever comes first

Closes below the bottom of the box
Or closes 10% below its highest close since the trade opened

Typical holdingWeeks to months

An example from the backtestWelspun Corp WELCORPThe rules were met at the close before 4 Jun 2026. The trade opened that morning and closed on 16 Sep 2026, 72 trading days later, +69.4% after costs. Exit: 10% below its highest close.

Stocks meeting the rules 0

At the close of 1 Oct 2026, among NSE's 500 most-traded stocks. A screen, not a recommendation: in the backtest, a trade opened at the next day's open.

No stock among the 500 most-traded met every rule at the last close. Strategies like this wait for their setup; the list fills again when one appears.

Backtest

1 Jan 2016 to 1 Oct 2026: a 10-stock portfolio following these rules on NSE's 500 most-traded stocks at each date, after 0.2% costs on each side, against the Nifty 500 index.

Yearly return+16.2%Index +11.5%

Compound growth a year (CAGR).

Total return+404%Index +224%

Gain or loss over the whole test.

Worst fall−35.0%Index −38.3%

Largest drop from a high (max drawdown).

Win rate38.6%

Share of trades that closed with a gain.

Profit factor2.09

Rupees gained for every rupee lost.

Average trade+5.20%

Average result per trade, after costs.

Average gain · loss+25.9% · −7.8%

A typical winning and losing trade.

Days held55.9

Average trading days per trade.

Trades415

About 39 a year.

Sharpe ratio1.11

Return for each unit of ups and downs.

Time invested89%

Average share of the money in trades.

Best · worst trade+883% · −19%

The extremes, after costs.

₹1 lakh following the rules

1 Jan 2016 to 1 Oct 2026: ₹5.04 L against ₹3.24 L in the Nifty 500.

Darvas BoxNifty 500

Falls from the high

Worst in this period −35.0%

Each calendar year

Return in the year, after costs.

Darvas BoxNifty 500

How the 415 trades ended

Each trade's return after costs, grouped.

GainLoss

How sturdy is it?

The same rules, tested again with one thing changed. A result worth trusting holds up across these; a big swing means the headline figure depends on luck or on the exact setup.

Beat the Nifty 500 in 5 of 5 versions and in 1 of 2 halves of the test.

Version	Yearly return	Worst fall	Win rate	Profit factor	Trades
As published10 stocks at a time, 0.2% costs on each side	+16.2%	−35.0%	38.6%	2.09	415
With the market filterNew trades only while the Nifty 500 is above its 200-day average	+13.1%	−39.8%	39.4%	2.02	376
Costs doubled0.4% on each side, as slippage in thinner stocks can be	+14.6%	−36.2%	37.8%	1.96	415
5 stocks at a timeA bigger share of the money in each trade	+21.4%	−40.5%	42.1%	2.46	216
20 stocks at a timeA smaller share in each trade	+14.5%	−33.0%	40.4%	1.97	809

First half · 1 Jan 2016 to 21 May 2021

Strategy+23.5%/yr

Nifty 500+12.8%/yr

Ahead of the index by 10.7 points a year.

Second half · 21 May 2021 to 1 Oct 2026

Strategy+9.4%/yr

Nifty 500+10.3%/yr

Behind the index by 0.9 points a year.

Recent trades in the backtest

The last 12 trades the rules closed, with 10 still open.

Stock	Opened	Closed	Days	Return	Why it closed
Nippon Life India Asset ManagementNAM-INDIA	21 Aug 2026	17 Sep 2026	18	−9.3%	Closed below the bottom of the box
Aarti IndustriesAARTIIND	13 Aug 2026	16 Sep 2026	23	−12.6%	Closed below the bottom of the box
IndusInd BankINDUSINDBK	7 Jul 2026	16 Sep 2026	50	−5.5%	10% below its highest close
Welspun CorpWELCORP	4 Jun 2026	16 Sep 2026	72	+69.4%	10% below its highest close
Manappuram FinanceMANAPPURAM	7 Jul 2026	8 Sep 2026	45	−5.2%	10% below its highest close
Himadri Speciality ChemicalHSCL	17 Jul 2026	21 Aug 2026	25	−3.5%	10% below its highest close
Aster DM Quality CareASTERDM	10 Jun 2026	21 Aug 2026	51	−3.7%	10% below its highest close
Astra Microwave ProductsASTRAMICRO	3 Aug 2026	13 Aug 2026	8	−11.7%	Closed below the bottom of the box
Vardhman TextilesVTL	18 Mar 2026	3 Aug 2026	91	+5.3%	10% below its highest close
JBCHEPHARMJBCHEPHARM	11 Mar 2026	17 Jul 2026	84	+12.5%	The stock stopped trading
Belrise IndustriesBELRISE	10 Jun 2026	7 Jul 2026	18	−5.8%	10% below its highest close
Coal IndiaCOALINDIA	6 Mar 2026	7 Jul 2026	80	−4.5%	10% below its highest close

How the tests work

Universe. The 500 most-traded NSE stocks at each date, by six-month average traded value, re-ranked every month: 1,353 different stocks over the years, 131 of which no longer trade. Testing all of them, not only today's survivors, keeps the results honest.
Period. 1 Jan 2016 to 1 Oct 2026, after a year of history for the long averages, on NSE's own end-of-day files: prices, delivery and futures open interest.
Portfolio. 10 slots, each a tenth of the portfolio's value when a trade opens. A signal at a day's close opens at the next day's open; an exit rule met at a close closes the trade at the next open. When there are more signals than free slots, the strategy's own ranking decides.
Costs. 0.2% on each side of every trade (STT, stamp duty, exchange charges and slippage). Tax and DP charges are not included.
Market filter. Every strategy is also tested opening new trades only while the Nifty 500 closes above its 200-day average. At the last close it was below it.
Robustness. Each strategy is run again with costs doubled, with 5 and with 20 stocks at a time, and judged separately in each half of the test. A result that only holds in one setup is luck, not an edge.
Corporate actions. Splits, bonus issues and demergers are applied from NSE's corporate-actions list, and symbol changes are joined up, so none of them shows up as a gain or a loss.
What is left out. Dividends (as in the Nifty 500 price index it is measured against), taxes, and the chance that a large order moves a thinly traded price.

Swing Lab is a screen and a history lesson, not advice. RupeeVerse is not registered with SEBI as an investment adviser or research analyst. A stock meeting a strategy's rules is not a recommendation to trade it, and past results do not predict future ones. Do your own research.