Darvas Box Traps Elusive Returns

Darvas called himself a techno-fundamental trader and began identifying trading candidates with a quick fundamental screen. He wanted to buy stocks in the industries of tomorrow, those with the strongest growth potential. In his day, this included electronics and missile technologies. It wasn't possible for Darvas to follow too many stocks because this required obtaining quotes through telegrams. The fundamental screen, therefore, was needed to narrow the number of stocks he was following.

From there, he turned to finding stocks with good technical characteristics. What he looked for, in the words of modern technical analysts, was a short-term rectangle forming on weekly charts. In each issue of Barron's, there is a page of stock charts featuring stocks in the news, and he looked for stocks breaking out of narrow trading ranges on high volume.

Figure 1: An example of trading the Darvas box

In Figure 1 above, the Darvas box is shown. The box is created when the stock trades within a rectangle formation. This is characterized by prices trading within a narrow range for at least three weeks. There is some subjective decision-making in this process, but if the trader must question whether a stock is in a narrow range, he or she should pass on that stock and look for a clearer signal. Buys are taken when prices break through the top of the box. Well-behaved stocks will form new boxes at higher levels. Stocks should be sold when the price breaks below the bottom of the highest box.

The rules can be explained so that modern tools like scanning software can identify trading candidates. To quantify the box, traders should look for stocks in which the difference between the high and the low price over the past four weeks is less than 10% of the stock's high during that time. As a formula, it can be written as:

(100 * ((High – Low) / High)) <>

Traders can use a larger percentage to get more stocks on their potential buy lists. The buy should be taken at the market's open the morning after the stock closes outside the box by at least half a point on a volume that is greater than the average 30-day volume. The initial stop should be set a quarter point below the lowest price of the box. It should be raised as new boxes form, always a quarter point below the low.

Darvas in Action
Figure 2 shows an example of a Darvas box pattern. Deere (NYSE:DE), best known as a tractor and farm equipment manufacturer, would not have passed Darvas' fundamental screen, but eliminating this screen offers an advantage to modern-day traders who can follow an unlimited number of stocks. In addition, scanning software, available at a low cost to traders today, was not available to Darvas.


Figure 2: Deere & Co demonstrates that the Darvas box method of trading stocks still applies in today's markets.

The first box in the Figure 2 chart began forming in late 2005. DE traded in a very narrow range of less than 5% for eight weeks. When a stock is range-bound for so long, the trader can use a stop order to set the entry price. In this case, the box was drawn with a high near 35.70, and traders could place buy stop orders near 36.

The trade was entered on January 26, 2006, when the stock broke out to the upside. Immediately after the buy order was filled, the trader entered a stop-loss order at the bottom of the box, 33.75 in this case. The risk of about 6.25% is then known by the trader.

Two new boxes are drawn as DE moves higher. Boxes are drawn on the chart when at least three weeks of a narrow range are identified. This can be monitored visually on the chart, or adept programmers can easily code these criteria into their trading systems. Each time a new box is drawn, the stop-loss order is changed to reflect the higher bottom.

In June 2006, the bottom of the third box is broken, and the trade is closed after 18 weeks with a profit of more than 10%. DE declines a little bit and forms a new box near the same level as the original box. In short order, it gives another buy signal by breaking out of the box.

The sell signal again comes 18 weeks later, after a gain of more than 18%. But, the sell signal turns out to be wrong and the stock quickly reverses.

What Would Darvas Do?
When Darvas saw this happen, he would quickly get back into the stock, using his original stop. However, automated testing of trading systems leaves many traders unwilling to jump right back into a trade like this. The problem we face today is one of too much information. We see the cost of whipsaw trades and we know they occur frequently with moving average or oscillator-based systems. Darvas didn't have this knowledge, and he operated under the assumption that "the trend is your friend", and what has worked so well in the past will work well in the future. Without fear, he was able to buy because the chart told him to.

With the Deere trade, we exercised caution and waited for the price to confirm it was resuming the uptrend - but we didn't have to wait long. Another box quickly formed, and within two months, we were back in DE. This trade was the big winner, up more than 50% over eight months.

Handling such a big winner presents challenges to the trader. Using only boxes, the stop-sell order to close the trade is more than 20% below the closing price. That's a lot of profit to give back. In a case like this, it might be better to use the top of the box, which sets the stop a little closer to the current market. Alternatively, traders can use an oscillator such as the relative strength index (RSI), which would yield an overbought reading given this price action, and sell when the stock breaks below the overbought level, although this could sacrifice future gains.

(by Michael Carr)