Trend Following

Hurst, Ooi and Pedersen describe a basic form of Trend Following as Time-Series Momentum: positive past returns lead to long exposure and negative past returns to short exposure in their model. This is a research rule across multiple futures markets, not a general instruction to trade Bitcoin in both directions. Every implementation must specify the past-return horizon and when the signal becomes known. [1]

A moving average of recent completed closes is one possible tool: a three-point SMA adds three consecutive Close values and divides by three. Other window lengths or a comparison of two averages produce different signals. For a test, always state whether a completed Close above or below the SMA makes the decision, how equality is treated, and whether the signal merely changes exposure or generates an order. [2] [5]

An average uses past prices, so it responds with a lag and cannot know whether a new move will last. StockCharts and CME warn of repeated false switches when a market lacks a strong trend; faster settings can respond sooner but change the number and character of signals. A filter chosen in advance may remove some switches at the cost of further delay. [2] [3]

Saying that price is rising is not enough to define a signal. Specify whether an upward SMA cross moves from no exposure to a long position and whether a downward cross means exiting, switching short, or only reducing position size. Equality between Close and SMA, repeated signals, and the interval between checks need the same predefined interpretation for successful and unsuccessful cases. [2] [5]

If the SMA includes the just-completed bar, the full signal value is known only after its Close. TradingView's default model therefore fills a market order on the next available tick, typically at the following bar's open. Backtesting must not credit an already completed move as an executed trade; for Bitcoin, specify the exchange, price feed, Slippage, fees and liquidity conditions separately. [4]

The same signal direction can imply very different risk depending on position size, Leverage, exit method and the number of concurrent positions. A strategy must describe how it limits loss, financing, possible gaps and simultaneous moves across markets; even diversification does not remove the possibility of loss. Short futures exposure and merely leaving a spot position are different decisions. [1] [4]

AQR studies a diversified Time-Series Momentum strategy across futures and currencies over a long history. That cannot establish the return of a particular Bitcoin SMA, one exchange, or guaranteed behavior in the next regime. Separate data used to develop a rule from data used to evaluate it, and include costs, losing stretches and sensitivity to parameter changes. [1] [5]

Before evaluation, fix the data, market, Timeframe, SMA length, crossover rule, exposure, execution and costs. CFA Institute warns about look-ahead and survivorship bias; TradingView also describes distortion from improper order timing. Assess the whole period, including sideways moves and losing switches, and validate the rule on data that did not determine its parameters. [5] [4]