Strategy Types: A Neutral Overview

The short answer

Most retail strategies are variations on four structures: trend following, mean reversion, breakout, and carry. Each is a bet that one specific market behaviour will persist, and each pays for its winners with a specific, recurring cost.

This lesson recommends none of them. No family is inherently more profitable than another. They differ in the shape of their returns and in what they demand of the person running them, and the second difference is usually the one that decides the outcome.

What a strategy family actually is

A family is defined by its assumption, not by its indicators.

Two traders using the same moving average can be running opposite methods: one buys when price is extended above it, betting the move continues; the other sells, betting it reverts. The indicator is decoration. The assumption is the strategy.

That matters when you evaluate your own results. A run of losses inside a family's normal failure mode is information about the market regime; a run of losses outside it is information about your implementation. You can only tell the two apart if you know which bet you are making.

Trend following

Assumes: moves that have persisted are more likely to continue than to reverse — that trend exhibits some autocorrelation over the timeframe being traded.

Outcome shape: a low proportion of winners, with a small number of large ones covering many small losses. Expectancy is concentrated in the tail.

What it costs: repeated small losses during ranges. Every false start is paid for in full, and ranges are common. It also costs a great deal of foregone entry — the method requires waiting for a move to establish itself, which means you are never early and often feel late.

How it fails: long stretches of sideways or choppy price action, where the entry condition triggers repeatedly and reverses. The drawdown is not usually one dramatic loss; it is forty unremarkable ones in a row.

Who can run it: someone who can tolerate a hit rate well under half without concluding the method is broken, sit through long flat periods, and keep position size constant while a run of losses accumulates. If you find a 30% hit rate psychologically intolerable, this family will not survive contact with you, regardless of its merits. See trend and moving averages for the mechanics.

Mean reversion

Assumes: price deviations from some reference — an average, a range boundary, a basis relationship — tend to revert.

Outcome shape: a high proportion of winners, each small. Expectancy is spread across many trades rather than concentrated.

What it costs: the left tail. The structure is short the extreme — it profits when nothing dramatic happens and loses when something does. Transaction costs also bite harder here, because the average win is small relative to a round trip.

How it fails: a deviation that does not revert. Because winners are small and frequent, a single continuation can remove the gains of many trades at once. This is the family where the temptation to average down is strongest and most expensive, for exactly the reasons set out in position management and scaling.

Who can run it: someone who will take a defined loss on an unreverted move without negotiating with it, who will not widen a stop on the argument that reversion is now even more likely, and who can accept that a run of small wins does not mean the method is safe.

Breakout

Assumes: a period of contracted range resolves into an expanded one, and the direction of the resolution carries information.

Outcome shape: somewhere between the two above, and highly dependent on the definition used. A tight breakout definition produces fewer trades with a lower hit rate; a loose one produces many trades, most of which fail.

What it costs: false breaks, and slippage at the moment of entry. Breakouts trigger precisely when volatility expands and liquidity at the touched level thins, so the realised entry is frequently worse than the intended one. That gap is a structural cost of the family, not an execution mistake.

How it fails: the break that reverses immediately and continues in the other direction, converting a small planned loss into a larger one as the move accelerates away. Repeated false breaks at the same level are the normal texture of ranging markets — see ranges, consolidation and breakouts.

Who can run it: someone available at specific times, because breakouts are events rather than states, and someone willing to accept a fill worse than the trigger price without abandoning the trade or chasing it further.

Carry

Assumes: an instrument pays — or costs — something for simply being held, and the accrual persists long enough to be worth the exposure.

The form varies by market: an interest rate differential, a dividend, a funding rate on perpetual futures, the roll between a futures price and spot. What is common is that the return arrives with the passage of time rather than from a directional move.

Outcome shape: frequent small accruals, punctuated by losses driven by the price of the underlying rather than by the carry itself.

What it costs: the accrual is usually small relative to the volatility of the thing you must hold to collect it. The position is exposed to the full move in the underlying while earning a fraction of it per period.

How it fails: a sharp adverse move that removes many periods of accrued carry at once. Carry can also invert — a funding rate that paid you can start charging you — which converts a carry position into a directional one you did not choose. In leveraged form, the failure is faster, because the same move that erases the carry can also reach a liquidation level. Funding and carrying costs covers the mechanics; the carry entry gives the short definition.

Who can run it: someone who monitors the accrual rather than the chart, sizes against the volatility of the underlying rather than the size of the yield, and does not read a long run of small credits as evidence that the position is safe.

Comparing them honestly

Read the table as a description of shape, not of quality. A high hit rate is not an advantage; it is a structural feature that comes with small average wins. A low hit rate is not a weakness; it is what concentrating returns in a tail looks like from the inside.

There is no ranking here because there is no ranking to give. Any of these can be implemented well or badly, in a market that suits it or one that does not, with costs that leave room for the edge or costs that consume it. Published results are not a guide either: they carry the biases described in backtesting and forward testing, and they never include whether the author actually followed the method.

What to do with this

Do not choose by appeal. Choose by constraint, and write the constraints down first.

How many hours are you genuinely at the screen, and when? What is the longest losing run you have sat through without changing something? Do small frequent wins settle you or make you complacent? Can you hold a position overnight without checking it?

Match a family to those answers, then specify it as rules using the tests in entry and exit rules that can be followed, and test it as described earlier in this course. Give it a sample large enough to mean something before you judge it, and expect its characteristic failure to show up — because when it does, that is the family behaving normally, not the method breaking.