What Trend Following Is: Directional Persistence and Three Market States
A trend is the phenomenon of price holding one direction over a period of weeks or months. It is not a single move or an intraday swing; it is a persistent posture that price maintains for an extended stretch of time. Trend following is built on a simple premise: identify that persistence as early as possible, participate while it lasts, and step aside before it fades. The basic strategy is to buy when an upward trend is just beginning and to exit before that trend ends. Markets are described as settling into one of three states: an uptrend, a downtrend, or a horizontal trend. The horizontal state is the reference point for everything else. In the Turtle program, participants were taught to buy when a market had just shifted from horizontal into upward, to sell short when a downtrend was about to begin, and to exit immediately once the trend ended. Trend ending was defined precisely: the market had turned from an upward or downward trend back into a horizontal one. This framing is what makes trend following mechanical rather than predictive. The trader is not forecasting where price will go; the trader is reacting to a change of state that has already been observed. The horizontal range is treated as a baseline of indecision, and a breakout out of that range is treated as evidence of a new direction. An important consequence follows. Because entries only occur after a range has been broken, and because many breakouts fail, trend following accepts a large number of small losses and long quiet periods in exchange for a few extended moves. The source material notes that most profits may come from just two or three successful trades. That distribution shapes every other rule in the method: risk control, exits, and the discipline required to keep taking signals after a run of losses.
The Breakout Method and Donchian Channels
The specific entry technique used in the Turtle program is called the breakout method, sometimes referred to as Donchian channels, because Richard Donchian popularized this channel breakout approach. The basic idea is straightforward: buy when the market exceeds the highest point of a specified past period, which means price has broken out beyond previously established levels. In practice, the channel is built from two reference lines. The upper line comes from the highest high of the lookback window, and the lower line comes from the lowest low. While price trades between those two lines, the market is behaving like a horizontal range. When price closes or trades beyond one of the lines, the breakout condition is met. After each trading day, the high and low breakout points are recalculated for the systems in use. The daily routine is modest: look back over the recent stretch of price data and identify one or two significant highs based on the visual character of the price movement. On most days the high does not change, so there is nothing to do. Action is only triggered when price actually moves beyond the level. The logic behind the breakout is that a prior high represents a level where sellers previously held the market back. When price exceeds that level, buyers have absorbed the supply that stopped the move before, and the market may be shifting from a horizontal state into a directional one. The same reasoning applies in reverse for the short side, where a break below the recent low suggests a shift into a downtrend. Breakout levels are not certainties. Many breakouts fail and price falls back into the range, which produces a small loss. That is precisely why the method is paired with a predefined stop rather than with discretion.
System 1 and System 2: Two Time Horizons Working Side by Side
The Turtle approach used two separate systems, each with its own lookback window. System 1, described as the intermediate-term system, determined its high and low breakout points from the past 20 days, or roughly four trading weeks. System 2 used a longer time span and determined its breakout points from the past 60 days, or about 12 trading weeks. Every trading day, after the close, the high and low breakout levels for both systems were recalculated. The two horizons produce different behavior. A shorter lookback reacts faster, so it produces more breakouts and engages with moves earlier. The trade-off is sensitivity to noise: in a choppy market, a 20-day high is easier to exceed temporarily, which can generate signals that do not develop into sustained trends. A longer lookback, such as 60 days, requires a larger move before it is triggered, so it produces fewer signals but each one reflects a more substantial change in price behavior. Running both systems together means the trader is monitoring two different definitions of what counts as a breakout at the same time. The shorter window can capture the early phase of a move, while the longer window only engages once the move has already proven itself against a broader history of prices. It is worth noting that the specific numbers — 20 and 60 days — are parameters rather than laws of nature. They define how sensitive the system is to price change and how frequently it will act. The source material emphasizes that the particular method used was not the decisive factor in the program's results, which implies that the choice of lookback window is a design decision rather than a secret ingredient. What matters is that the levels are calculated consistently according to a fixed rule.
Two Exit Criteria: The 2N Stop and the Trend-End Exit
Exits receive as much structure as entries in this framework. Each system is described as having two exit criteria. The first is a stop-loss exit measured in N, where the loss must not exceed 2N, meaning two average true ranges. Put differently, that decline is equivalent to about 2 percent of total account equity, because the method used to determine trade size in each market is also based on N. This coupling is the key point. Because position size is derived from N, and because the stop is set 2N away from the entry, the maximum loss on a trade is roughly the same fraction of the account regardless of which market is being traded or how volatile it happens to be. Risk per trade is normalized rather than left to chance. The second criterion is a trend-end exit. The position is closed when the trend ends, and trend ending has a specific definition: the market turns from an upward or downward trend back into a horizontal trend. For a long position, that means price has fallen back to the lower boundary of the recent range; the short side mirrors this logic. Both the stop level and the breakout boundaries are recalculated each day as new price data arrives, so the exit references move with the market. The stop is not an afterthought that is adjusted emotionally; it is a level determined before the trade is taken and updated according to a rule. This structure explains why exits are treated as mechanical events. The stop bounds the damage from a failed breakout, and the trend-end rule surrenders part of an open profit in exchange for confirmation that the directional phase is over. Together they define the trade in advance: where it begins, where it is wrong, and where it is finished.
N: The Volatility Unit Behind Position Sizing and Stops
N is the average true range, a measure of how much a market typically moves in a day. The true range accounts for the full extent of daily movement, including gaps between one session and the next, rather than only the distance between the day's high and low. Averaging that figure over a period produces N, a volatility yardstick for the market in question. N serves two distinct roles. First, it measures volatility, telling the trader how much daily movement is normal for that market at that time. Second, it becomes the unit in which both stop distance and position size are expressed. In the Turtle method, trade size in each market was determined on the basis of N, so that a move of 1N corresponded to a fixed fraction of the account. This produces a useful form of comparability. A market with a large N is volatile and therefore receives a smaller position, while a quiet market with a small N receives a larger one. Because the 2N stop is scaled to the same unit, a losing trade in either market costs approximately the same share of the account. Risk becomes a constant across markets rather than a variable. N is also dynamic. As volatility expands, N grows, and position sizes shrink accordingly; as volatility contracts, the opposite happens. The system therefore adapts automatically to changing conditions rather than holding a fixed exposure. One general limitation is worth noting. Averages look backward, so a volatility estimate can lag behind a sudden change in market conditions. When price behavior shifts abruptly, the most recent N may understate current movement, which is one reason volatility-based sizing is a framework for control rather than a guarantee.
Why the Rules Matter More Than the So-Called Secret
Over the years, many people have studied the rules associated with the Turtle program, and some have earned substantial sums by teaching those rules to others. The source material notes something important about this: the particular set of rules used was only an incidental factor. Many other well-known trend following methods were equally successful, or even more successful. The method in question was already widely known at the time the program took place. This observation is counterintuitive. If the rules were public, and if comparable approaches worked just as well, then the rules themselves cannot be the source of the advantage. The material concludes that success can be achieved through long-established ideas and concepts, but only if those rules are applied consistently. Consistency of execution is described as the secret of trading and the secret of the program's results. The practical implication is that searching endlessly for a superior indicator may be a distraction. Many breakout systems share the same essential logic: define a range, act when price leaves it, limit losses using a volatility-based stop, and remain in the trade while the direction persists. Variation between systems often comes down to parameters and details rather than to fundamentally different ideas. A public rule set also has a behavioral advantage: there is nothing to hide behind. Since the method is not proprietary, the trader cannot blame a lack of information for poor results. What remains is execution — taking every signal, accepting the small losses, and holding positions long enough for the occasional large move to arrive. That reframing moves the problem from the chart to the trader, which is exactly where the source material places the real difficulty.
The Four Pillars: Edge, Risk Control, Consistency and Simplicity
The training content is summarized in four points, and each one addresses a different failure mode. First, master an edge: find a trading strategy with positive expectancy, because over the long run it creates positive returns. Positive expectancy does not mean every trade wins. It means the average outcome across many trades is favorable, which allows for a large number of losing trades inside a profitable process. Second, manage risk: control risk and hold your ground, otherwise even a positive-expectancy system will be abandoned before it can produce results. This pillar connects directly to the 2N stop and volatility-based sizing. Capital preservation is not a side activity; it is what keeps the trader present long enough for the edge to express itself. Third, be unwavering: only through consistent execution does a trader actually capture a system's positive expectancy. A mathematically sound system that is followed irregularly is no longer that system. Skipping signals introduces a new and unmeasured strategy, and the expectancy calculation no longer applies to it. Fourth, keep it simple: the essence of the method is to catch every trend. Most profit may come from two or three successful trades, so missing a trend can undo an entire year of work. The material adds a candid observation: this is simple and easy to understand, but not easy to do. Read together, the four pillars describe a process rather than a prediction. The edge defines the statistical basis, risk control keeps the trader in the game, consistency is what converts the edge into realized results, and simplicity is what makes consistency possible in the first place.
Where Trend Following Struggles: Whipsaws, Drawdowns and Discipline
Horizontal markets are the primary hazard for a breakout approach. When price is ranging, breakouts tend to fail: price pushes beyond the recent high or low, then returns inside the range, and the stop is triggered. The same sequence can repeat several times. A cluster of small losses in a range is the cost of waiting for a trend, but it is also the experience most likely to make a trader abandon the rules. The trend-end exit rule has its own cost. Because an exit only occurs once the market has visibly turned back into a horizontal state, part of an open profit is given back before the position is closed. The rule is designed to keep a trader in a move long enough to capture a large portion of it, not to capture the exact turning point. There is also parameter risk. Lookback windows such as 20 days and 60 days are choices, not fixed truths. Adjusting them changes how often signals appear and how early the system engages. Fitting these numbers closely to past price data can produce impressive historical results that do not carry forward, because the fit reflects specific past conditions rather than a durable mechanism. The underlying distribution of results adds another layer of difficulty. A method that earns most of its profit from two or three trades implies a low win rate overall, with many small losses funded by a few large winners. That structure is psychologically demanding, and it creates a specific temptation: overriding the rules during a drawdown, precisely when the conditions that produce the large moves are most likely to appear. Discretion applied at that moment breaks the expectancy the system was built on. Based on study notes, for educational purposes only and not investment advice.
