Profitable Trading Strategies: What Data Shows

The trading strategies that have most often shown an edge in historical tests are trend following, momentum, rotation with a market filter, breakout trading and disciplined rebalancing. None of them is reliably profitable: each has long losing stretches, and a strategy that made money in a backtest can lose money in real life once costs, overfitting, hindsight and human behaviour are counted. The honest answer is that data shows which ideas have worked before, not which will work next.

This guide walks through those strategy families, uses AlgolaneBots's own five backtests as worked examples, and explains why "profitable on paper" and "profitable in your account" are different claims.

What does "profitable" actually mean for a trading strategy?

A strategy is only interesting if it does something useful compared with a simple alternative. For most stock strategies that alternative is holding an index fund such as the S&P 500. For a gold strategy it is simply holding gold.

So three questions matter more than the headline return:

• Compared with what? A strategy that made 10% a year while the index made 14% was profitable but not useful.
• At what cost in falls? A higher return that came with a much deeper worst fall may not be worth it, because few people hold on through a 50% loss.
• Over what period, and how was it chosen? A test that starts just after a crash, or uses rules tuned on the same data, flatters itself.

Keep those questions in mind for everything below. Our guide on how to read a backtest covers them in more detail.

Which strategy families have shown an edge in tests?

Researchers and practitioners have tested thousands of trading ideas. A small number of families keep coming back because they are simple, they make sense, and they have shown some persistence across markets and decades. They also all have well-known weak spots.

Trend following and momentum

These two are close relatives. Trend following asks whether one asset is rising; momentum asks which of several assets has risen most. Both have a long record in academic research across many markets, and both go through painful periods when trends reverse. The trend following explainer covers the mechanics.

Rotation and regime filters

A regime filter is a rule that decides whether the market is healthy enough to take risk. When it says no, the strategy moves to bonds, cash-like funds or other defensive holdings. The appeal is avoiding the deepest part of a bear market. The cost is being out of the market when it rebounds fast.

Breakout

Breakout strategies wait for a period of low movement, then trade in the direction prices escape. They tend to have many small losses and fewer large wins, which is hard to sit through.

Rebalancing

Rebalancing is the least exciting and the most widely used. It keeps a portfolio from drifting into whatever happened to rise most. Its "edge" is mostly about controlling risk rather than adding return.

What do AlgolaneBots's five backtests show?

AlgolaneBots runs five rule-based strategies, and each one is built on one or more of the families above. Here are their results. These are historical backtests, meaning computer tests on past prices. Nobody earned these returns, and future results will differ.

Every one of these numbers comes with a caveat, and the caveats matter as much as the returns:

• Steady Growth: its seven names were picked because they already did well, which flatters the test.
• Big Tech Trend: the stock list was chosen with hindsight, and three stocks made 81% of the profit.
• Gold Breakout: doubling trading costs cuts its yearly result roughly in half.
• Market Switch: nearly half of its tested profit came from UVXY, and it failed outside the US. UVXY is a volatility fund that has lost most of its value over long periods, so the strategy depends on timing it well.
• Leveraged Growth: its rules may be fitted to the years after 2010, and leveraged funds can lose value quickly. Leveraged funds reset their leverage every day, so in choppy markets they can lose value even when the index ends up roughly flat. See leveraged ETF decay.

Notice the pattern. The higher the yearly figure, the deeper the worst fall, and the more the result depends on a specific period or a few lucky holdings. Leveraged Growth fell 53% in the 2020 Covid crash against 34% for the S&P 500. A simulated three-fund version of it fell 91% in the dot-com crash against 47% for the index, and took 115 months to get back against 78. That is the kind of history the headline number hides.

Why is a profitable backtest not the same as real profit?

A backtest is a controlled experiment on the past. Real trading adds four things it leaves out.

Costs and slippage

Every trade costs something: commissions, the gap between buying and selling prices, and fills at worse prices than expected. A strategy that trades often can see a large share of its edge eaten this way. Gold Breakout's caveat is a direct example: double the assumed costs and its yearly result roughly halves. AlgolaneBots's paper account takes 0.05% off every fill for this reason. Read more in paper trading slippage and costs.

Overfitting

With enough tries, you can find rules that fit any past data beautifully. Those rules often describe noise, not a real pattern, and fall apart on new data. A strategy with many settings, each tuned to its test period, is a warning sign. Our post on overfitting explains how to spot it.

Hindsight and survivorship

Choosing which stocks to test after you know which ones did well is a quiet form of cheating. It is why Steady Growth and Big Tech Trend carry the caveats they do. The companies that failed are not in the list, so the list looks better than any list chosen in advance could have. See hindsight and survivorship bias.

Behaviour

A backtest never panics. It holds through a 55% fall, waits 43 months for a recovery, and takes the next signal after five losses in a row. People often do not. Selling near the bottom and returning after the rebound can turn a strategy that worked on paper into a loss in practice. This gap is often the largest of the four.

How can you judge whether an edge is real?

No test can prove a strategy will make money. But some checks make an edge more believable:

1. A sensible reason. Can you explain in a sentence why the pattern might persist? Trend following has a story; a rule like "buy on the third Tuesday" does not.
2. Few settings. The fewer knobs, the less room for overfitting.
3. Different periods and markets. Does it hold up before and after the period it was designed on? Market Switch failing outside the US is the kind of result this check exposes.
4. Crash behaviour. How did it do in 2008, 2020 and 2022 compared with its benchmark? The stress test post shows this for all five.
5. Realistic costs. Does the result survive higher cost assumptions?
6. Forward testing. Does it behave as expected on live prices it has never seen? This is what paper trading is for.

Even a strategy that passes all six can have years of losses. The risk notice sets out what can go wrong in more detail.

This article is general education, not personal advice.

Key takeaways

• Trend following, momentum, rotation, breakout and rebalancing are the strategy families that have most often shown an edge in historical tests, and all of them have long losing stretches.
• Judge any result against a benchmark over the same period, alongside its worst fall and recovery time.
• In AlgolaneBots's backtests, higher yearly returns came with deeper falls and bigger caveats.
• Costs, overfitting, hindsight and behaviour are the main reasons backtest profits shrink or vanish in real life.
• No strategy is reliably profitable; forward testing on paper is the cheapest way to see how one behaves.

See the strategies on live prices

You can run all five strategies on a $10,000 paper account with real prices during a free 90-day trial, with no card needed and nothing charged automatically. Watch how each one trades, including its bad weeks, before deciding anything. Start your paper trial.