1. Momentum and trend: betting that the move continues

Theory 6: A survey of real strategies

'What goes up a lot must come down soon.' That intuition, applied without judgment, has cost entire generations of investors money.

You now have all the conceptual tools from the previous levels: signal, entry, exit, position sizing, backtesting and evaluation metrics. This final level walks through, without code and at a conceptual level, the major families of strategies that dominate real algorithmic trading.

We start with the most intuitive of them all: momentum and trend following.

The premise: the move persists

The premise of momentum is easy to state and, at the same time, counterintuitive for anyone coming from the world of 'buy low, sell high'.

An asset that has risen strongly in the recent past tends, on average and in a statistically detectable way, to keep rising for a while longer. And an asset that has fallen strongly tends to keep falling.

This persistence of recent movement has been documented in dozens of academic studies, across multiple markets and multiple decades.

It flatly contradicts the intuition we opened with. And that intuition, badly applied, has led people to sell too early stocks that kept rising, or to buy too early stocks that kept falling.

Why momentum exists

There is no single academic consensus, but the most cited explanations combine two kinds of factor.

On the behavioral side: investors tend to react incompletely to new information (underreaction). They adjust their expectations gradually, instead of incorporating all the relevant information at once. That produces a prolonged price drift after the initial event, rather than an instantaneous adjustment.

On the structural side: many institutional investors enter established trends in stages rather than instantaneously. We mean index funds rebalancing, or funds following mandates to 'buy what is rising'. By entering little by little, they feed the move itself for a while.

Trend following: riding, not guessing

Trend following is the oldest and most widespread practical application of momentum.

It does not try to predict from scratch which way an asset will move. It simply detects that a move is already under way and rides it. It typically detects that through:

• Moving averages.
• Price channels.
• Breakouts of recent highs and lows.

The trade-off is explicit: it accepts a late entry in exchange for a reasonably high probability that the established move continues.

The result profile, and what it demands of you

A distinctive feature of well-designed trend strategies is their result profile:

• A relatively low win rate. Many signals trigger and then the move reverses almost immediately, producing repeated small losses.
• A high payoff ratio. On the few occasions when the trend really does develop with force, the strategy captures a substantial part of that large move.

It is exactly the expectancy pattern we saw in the previous module: many small losses, few large gains, and a positive expectancy overall if the design is sound.

Psychologically, that profile demands considerable tolerance for long runs of consecutive small losses without abandoning the system. It is precisely the ground where systematic discipline is most needed over discretionary judgment, as we saw at the start of the previous level.

Where it works and where it does not

Momentum and trend following work particularly well in markets with sustained, clear trends.

And they work particularly badly in sideways markets, or in markets with many abrupt changes of direction. There the signals trigger and reverse repeatedly, without ever developing a large move to offset the accumulated small losses. It is a period known informally among systematic traders as a 'whipsaw'.

No momentum strategy performs equally well across every market regime. Recognizing that limitation up front, instead of expecting a single strategy to work equally well forever, is an essential part of managing realistic expectations about any system of this kind.

Two momentums that get confused

It is also worth distinguishing two variants that are frequently mixed up.

Time-series momentum looks only at an asset's own past behavior relative to itself, without comparing it to anything else. If a stock has risen over the past few months, the strategy bets that this same asset keeps rising.

Cross-sectional momentum compares the relative behavior of many assets against one another over the same period. It builds the portfolio by buying those that have risen most relative to the rest of the group and, in more sophisticated versions, shorting those that have risen least.

The nature of the bet changes: you are not betting that a particular asset rises in absolute terms, but that the relatively strong keep outperforming the relatively weak within that same comparison universe.

Both variants rest on the same evidence of persistence in price moves, but they imply quite different design and infrastructure decisions. The most sophisticated institutional strategies commonly combine both perspectives within the same system.