The Darvas Box Breakout Method
Nicolas Darvas · 1950s (his signature run came in the 1957–1958 U.S. stock bull market, long-only and unleveraged)
Published 2026.07.08
Nicolas Darvas was a dancer touring the world in the 1950s. Because he traded off nothing but stock prices delivered by telegram — not the trading floor — he pinned his rules to a single thing that was neither news nor rumor: "the box that price draws." When a stock that has just made a new high spends a while consolidating up and down, it forms a box; he climbed on only when price closed above its ceiling, and got out without exception when the floor broke. This piece breaks down, for educational purposes, the "structure" of the method he published in his book — not a claim that copying it makes money, but a lens for scanning where breakouts and invalidations occur.
- A box is complete only once both its ceiling (not exceeded for three consecutive trading days) and its floor (not broken for three consecutive trading days) are confirmed after a new high.
- Entry is not on an intraday spike but only when price closes above the ceiling, ideally accompanied by a volume surge.
- The stop is preset below the box floor at the moment of entry and moves only upward, step by step, with each new box (no averaging down).
- It's not about win rate but reward-to-risk — many small losses are offset by a few large trends. We look at its limits too: it lags and is long-only.
① Identification — How a Box Is Completed
Darvas's starting point was not the stock but the regime. He first checked whether the broad market and the sector were bullish and whether the stock was near its yearly, 52-week high. A falling market or a weak stock, however cheap it looked, was dropped from the shortlist — his principle was "never fight the trend."
- Confirm the new high: the stock prints a fresh new high.
- Fix the ceiling: if that high is not exceeded for three consecutive trading days, that high becomes the "box ceiling."
- Fix the floor: once the ceiling is set, if a low holds for three consecutive trading days it becomes the "box floor." The box is complete only when both the top and the bottom are set.
While price is consolidating up and down inside the box, the rule is to wait and watch. Chasing an "unfinished box" whose top and bottom are not yet fixed is the fast track to getting stopped out over and over by false breakouts (whipsaws).
② The Entry Frame — Observation, Not Prediction
Entry is not prediction but lagging observation. Only when price closes above the box ceiling — ideally with a volume surge — is it treated as a "valid breakout." A momentary breakout that pokes above the ceiling intraday but pulls back before the close does not count as a signal.
Even for the same "ceiling breakout," a momentary intraday breakout is an unconfirmed signal that can be reversed, whereas a close above the ceiling is a confirmed signal that the day's buyers won. The Darvas rule is always close-based — this is a confirmation procedure, not a buy instruction.
While price danced inside the box I waited, and I climbed on only when it broke up out of the box — the gist of the "box" idea Darvas left behind.
③ Risk, Sizing, and Pyramiding
At the same moment as buying, Darvas preset a stop just below the box floor (or the most recent valid low). It nails down "how much you'll lose" before you enter. The distance from the entry price to this stop is precisely 1R — the single unit of loss you accept on this trade.
- When price stacks a new box above, the previous ceiling becomes the new box's floor.
- Raise the stop one step, to below the new box floor — never lower it.
- Additional buys (pyramiding) happen only when price breaks above a new box ceiling again. Averaging down into a falling stock is forbidden.
Buying more at a losing position to lower your average is the exact opposite of the Darvas rule. You add only when you're in profit and a new breakout appears above. It's a reward-to-risk structure that cuts losses short and lets profits run, so if this rule breaks, the whole method collapses.
④ Invalidation and Limits
Invalidation is simple. If price closes below the current box floor, the structure is considered broken and you exit without exception. A snap back into the box right after breaking above (a false breakout, a whipsaw) is likewise treated as invalid.
① It lags — a box is a product of past prices, so the signal is always late, and you often climb on just before a top and get trapped. ② Weak in range-bound markets — you get stopped out repeatedly by false breakouts. In later backtests, daily-scale performance was poor and it worked only somewhat on the weekly (52-week) scale. ③ Hit rate is coin-flip-level — results depend entirely on a few large trends. ④ It's a long-only, unleveraged stock method, so evidence that it reproduces in other regimes is weak.
Even the master's own record is contested. In 1960, New York State Attorney General Louis Lefkowitz called much of Darvas's profit claim "unverifiable," saying the verifiable gain was only about $216,000 (Darvas countered that it was press suppression, and the court halted the inquiry). It means the "$2 million" headline figure in the book's title is itself in dispute. So it's safer to read this method only as the structure of its rules, not as a "profit record."
⑤ Porting It to Crypto (Perpetual Futures)
Darvas bought spot stocks long-only and unleveraged. To carry this skeleton over to crypto — especially high-leverage perpetual futures — you must not copy it as-is; four points have to be adjusted.
- Timeframe: in a 24-hour, high-volatility market, using the "3 trading days" rule as-is is very noisy. Define the box on a confirmed close of a higher timeframe such as the daily or weekly to filter low-timeframe whipsaws.
- Leverage: because of the multiplier, a box-floor stop can lead straight to the liquidation price. Lower the leverage and back out position size from risk relative to the stop width.
- Stop hunts: liquidity just below a box boundary can be deliberately swept, so a stop placed flush against the boundary is easy prey. Add some buffer or a close confirmation alongside it.
- Directionality: Darvas didn't short. Assume from the outset that a long-only skeleton is only half a method in a frequently-falling crypto market.
No adjustment guarantees future prices or profits, and none is an individual entry instruction. What's laid out here is only the "structure" of a method that one particular person made public.
Say an asset's box ceiling is 100 and its floor is 92. If price closes above the ceiling at 101 and you enter, placing the stop at 91 just below the box floor, then 1R = from 101 down to 91 = 10 (about 10 percent). The Darvas way doesn't nail down a target in advance; it drags the position along with a trailing stop "as long as new boxes keep stacking" — if the trend runs to 121 and you exit on a box-floor break, that's plus 20 = 2R; at 141 it's 4R. The point isn't whether you're right, but whether this single 3–4R covers the three or four false-breakout losses cut at 1R each beforehand. If you fix the account's 1R at 1 percent of capital, then the moment the stop width (about 10 percent) is set, the notional position size (about 10 percent of capital) is automatically backed out. The numbers are assumptions for explaining the concept and have nothing to do with any specific security or future price.
- Is this asset actually in an uptrend and near new highs right now, or has it just fallen a lot and looks cheap?
- Are the box's ceiling and floor each "confirmed" by the three-trading-day rule, or is it still consolidating (unfinished)?
- Is the signal I'm seeing a close-based breakout, or a momentary intraday spike that can be reversed?
- Before entering, have I written down the below-the-box-floor stop, the 1R, and the position size as numbers?
- (Crypto) Given the leverage, does this stop sit clear of the liquidation price, and is there no risk of liquidity just below the boundary being swept?
FAQ
Over how many days is a Darvas box drawn?
After a new high, if that high isn't exceeded for three consecutive trading days it becomes the box ceiling, and if a subsequent low then holds for three consecutive trading days it becomes the box floor. The box is complete only once both the top and bottom are confirmed, and until then the rule is to wait and watch. In a 24-hour, high-volatility market like crypto, you should redefine "three trading days" as a confirmed close on a higher timeframe like the daily or weekly to filter out low-timeframe noise.
If price breaks the ceiling intraday, can I buy right away?
The core of the Darvas rule is "close confirmation." Momentary intraday breakouts often turn into false breakouts that reverse before the close, so a valid breakout is recognized only when price closes above the ceiling and, ideally, volume surges. This is a lagging confirmation procedure, not a buy instruction.
Did Darvas really make $2 million?
His book is titled "How I Made $2,000,000 in the Stock Market," but in 1960 New York State Attorney General Louis Lefkowitz labeled much of it "unverifiable," saying the verifiable gain was only about $216,000 (Darvas countered that it was press suppression, and the court halted the inquiry). Because the headline figure itself is contested, it's safer to read this method only as the structure of its rules, not as a profit record.
Why is averaging down forbidden?
Darvas didn't buy more of a falling stock. Additions happen only when price stacks a new box above and breaks its ceiling again (pyramiding). That's because it's a reward-to-risk structure that cuts losses short and lets profits run, and especially at high leverage in crypto, averaging down easily turns into a forced liquidation and a "total loss."