Types of trading strategies: 20 approaches and how quants test them

There are dozens of names for trading strategies, but most of them come down to a few ideas: some moves keep going, some snap back, related prices drift apart and come back together, and some risks pay a premium to whoever is willing to carry them. What separates a quant is how carefully they test an idea before trusting it with money.

This guide walks through 20 common strategy types, grouped into eight families. For each one you’ll find what it is, why researchers think it might work, how a quant would trade and improve it, and the mistake that most often makes a backtest of it look better than it is.

Twenty strategies, eight families

The 20 strategies at a glance

Trading styles vs. trading strategies

Day trading, swing trading and scalping often appear in lists of trading strategies, but they describe how long trades last, not why they should make money. Any of the 20 ideas below can be traded on different clocks.

• Scalping: seconds to minutes, with many small trades. Costs and fills decide almost everything.
• Day trading: in and out within one session, with no positions held overnight.
• Swing trading: days to a few weeks, aiming to catch one leg of a larger move.
• Position trading: weeks to months or longer, closest in spirit to trend following and carry.

A second split is how decisions get made. Discretionary traders use judgment. Systematic or algorithmic traders follow fixed rules a computer can run. Only rules can be backtested honestly, so a discretionary approach has to be written down as precise rules before it can be tested.

Trend and momentum

These strategies bet that a move already underway is more likely to continue than to reverse. They tend to lose often, in small amounts, and rely on a few large moves for their gains.

1. Momentum

Time frameWeeks to monthsNeedsDaily bars, several marketsUsed byFunds and individuals

Momentum trading means buying what has been rising and selling what has been falling. There are two versions, and they’re different bets. Cross-sectional momentum ranks markets against each other and buys the recent leaders. Narasimhan Jegadeesh and Sheridan Titman documented it in US stocks in 1993, using look-back periods of three to twelve months. Time-series momentum judges each market only against its own past: long if it has been rising, short if it has been falling. Tobias Moskowitz, Yao Hua Ooi and Lasse Pedersen found it in 58 futures and forward markets in 2012.

The usual explanation is that traders react slowly to new information, then overreact later. That’s why the effect fades and partly reverses over longer horizons.

How quants approach it: a rule such as “long if the past twelve months’ return is positive”, applied across many markets and rebalanced monthly, with positions sized by each market’s volatility so a quiet bond future and a wild energy future carry similar risk. Improvements come from diversification and risk control, not from searching for the perfect look-back.

The backtest trap: momentum has rare, sharp crashes, especially cross-sectional momentum, often when a falling market suddenly rebounds. A backtest that happens to miss those periods makes it look far smoother than it is.

2. Trend following

Time frameWeeks to monthsNeedsDaily bars, rolled futuresUsed byManaged-futures funds, individuals

Trend following, sometimes called trend trading, means going long markets that are rising and short markets that are falling, using rules such as a moving-average crossover or a new multi-week high. It’s time-series momentum put into practice, usually applied across many markets at once, and managed-futures funds are built around it. Research by Brian Hurst, Yao Hua Ooi and Lasse Pedersen traces trend-following rules back over a century of market data, and Ari Levine and Lasse Pedersen showed that moving-average rules and time-series momentum are closely related.

How quants approach it: simple entry rules, volatility-based position sizes, stops, and as many unrelated markets as the account allows. Trend followers accept being wrong most of the time. The edge, if there is one, comes from the size of the winners.

The backtest trap: futures expire, so a long test needs a continuous contract stitched together from many expiries. The way the contracts are joined can create profits that never existed, or hide real costs. Back-adjusted prices can even turn negative, which breaks any rule built on percentage moves.

3. Breakout trading

Time frameHours to monthsNeedsIntraday or daily barsUsed byIndividuals and funds

A breakout strategy enters when price moves beyond a recent range: the highest high of the last 20 days (often called a Donchian channel), the edge of a consolidation, or the high or low of the first part of the session, known as an opening-range breakout. The famous Turtle traders of the 1980s used 20-day and 55-day breakouts across futures markets. Their own rules spell out the catch: most breakouts don’t turn into trends, so most breakout trades lose.

How quants approach it: decide what counts as the range, how far beyond it is a breakout, where the stop goes and how to size the trade, often using average true range. Then test whether the result holds when each of those settings moves a little.

The backtest trap: fills. A backtest that buys exactly at the breakout level ignores gaps through it and slippage in fast markets. Because a few large winners carry the results, it’s also very easy to tune the look-back until the test catches them.

If your strategy is built from rules like these on futures charts, Kwaant can test it: you describe the rules in plain English, and it runs checks like realistic costs, walk-forward testing and a sealed final test.

Reversion

Reversion strategies bet the other way: that a stretched price is more likely to drift back than to keep going. They tend to win often, in small amounts, and lose rarely but heavily.

4. Mean reversion

Time frameHours to daysNeedsIntraday or daily barsUsed byIndividuals and funds

Mean reversion fades extremes. When price moves far from some idea of normal, such as a moving average, a band around it or an estimate of fair value, the strategy bets on a move back. Range trading, buying near support and selling near resistance, is a common form of it. Researchers offer a few reasons it might work: temporary mispricing, overreaction, and the reward for supplying liquidity to traders who need to buy or sell in a hurry.

How quants approach it: measure how stretched price is, for example in standard deviations from an average, enter at an extreme, and exit at the average, at a stop or after a set time. The improvements that matter most are realistic costs and a hard limit on how much can be lost when “stretched” keeps stretching.

The backtest trap: a market that’s trending has no fixed mean to return to. A reversion strategy can look excellent for years and then give it all back in one strong trend. Averaging down makes this worse and hides it until it happens.

5. Intraday reversal

Time frameMinutes to daysNeedsIntraday bars, ideally ticksUsed byIndividuals and market makers

Intraday reversal is mean reversion on a short clock: fading a sharp move, a spike or an opening gap, on the bet that part of it will undo itself. Academic work on individual US stocks found short-term reversals, measured against other stocks, at weekly (Bruce Lehmann, 1990) and monthly (Narasimhan Jegadeesh, 1990) horizons, and Stefan Nagel later linked these reversals to the reward for providing liquidity. Evidence for reliable reversals within a single futures market is much thinner. Short-term patterns can run the other way, too. Other research on the S&P 500 found intraday momentum, where the move from the previous close through the first half hour tends to continue into the last half hour.

How quants approach it: define the extreme precisely, use fixed targets or time exits, and pay close attention to session times and the cost of every trade.

The backtest trap: bid-ask bounce. Trades alternate between the bid and the ask, so a chart of last-traded prices shows small reversals that you could never have captured. A test built on those prices can find an edge that disappears once realistic fills and costs are included. Plenty of gaps never fill.

Relative value

Relative-value strategies don’t bet on whether a market goes up or down. They bet on how prices move against each other.

6. Pairs trading

Time frameDays to weeksNeedsSynchronized bars for both legsUsed byIndividuals and funds

Pairs trading takes two instruments that normally move together and trades the gap between them. When the gap, called the spread, widens unusually far, the strategy buys the cheap one and sells the expensive one, then exits when it closes. Evan Gatev, William Goetzmann and Geert Rouwenhorst studied the approach in a well-known 2006 paper. In futures, the same idea appears as calendar spreads (the same market, different expiry months) and intercommodity spreads between related markets.

How quants approach it: quants look for pairs that are cointegrated, not just correlated. Correlation means two prices tend to move together day to day. Cointegration means the gap between them, once one leg is scaled by a hedge ratio, keeps returning toward a stable level. Two highly correlated prices can still drift apart for good, which is exactly what a pairs trade can’t survive. Then come entry and exit levels for the spread, and a stop for when the relationship breaks.

The backtest trap: screen enough pairs and some will look cointegrated by chance. Choosing pairs with data from the test period makes it worse.

7. Statistical arbitrage

Time frameIntraday to daysNeedsA large group of instruments, low costsUsed byMostly quant funds

Statistical arbitrage, or stat arb, is pairs trading scaled up: a model looks for many small, temporary mispricings across a large group of instruments and holds a hedged set of long and short positions. Marco Avellaneda and Jeong-Hyun Lee described a classic version in 2010, trading the part of each stock’s move that its sector or the market doesn’t explain. The name is misleading: it carries real risk, and at best it profits on average across many trades.

How quants approach it: a model of what each instrument “should” do, a measure of how far it has strayed, and many small positions so no single one matters much. It mostly lives in stock markets and needs low costs and fast execution, which makes it mainly a fund strategy.

The backtest trap: crowding. In August 2007 many funds running similar strategies were forced to unwind at the same time, and losses spread through what had looked like a market-neutral book. A backtest can’t show how crowded a strategy will be when you trade it.

8. Cross-sectional alpha

Time frameDays to monthsNeedsMany instruments, ranked togetherUsed byMostly funds

Cross-sectional strategies rank many instruments on some measure at the same moment, buy the top of the list and sell the bottom. Often the long and short sides are roughly equal, so the result depends on relative performance rather than on the market’s direction. It’s the method behind most academic factors and cross-sectional momentum, and it contrasts with time-series strategies, which judge each market only against its own history.

How quants approach it: choose the ranking measure, split the list into groups, rebalance on a schedule and check that “market-neutral” really is neutral, with no hidden tilt toward one sector or toward the market itself.

The backtest trap: survivorship bias. The list of instruments has to include those that later disappeared or expired, exactly as it stood on each date. Testing today’s survivors makes the past look kinder than it was.

Factors and carry

These strategies try to earn a premium for owning certain traits or bearing certain risks, usually over months rather than days.

9. Factor investing

Time frameMonthsNeedsMany instruments, often fundamentalsUsed byMostly funds

Factor investing tilts a portfolio toward traits that research has linked to different average returns. Eugene Fama and Kenneth French added size and value to the market in 1993, and later profitability and investment. Momentum, quality and low volatility are other well-known factors. Factors also show up outside stocks: Clifford Asness, Tobias Moskowitz and Lasse Pedersen found value and momentum in country stock-index futures, government bonds, currencies and commodity futures as well. Why factors pay is still debated. Some researchers see a reward for risk, others behavioral mistakes or limits on what investors can do.

How quants approach it: sort a list by the trait, hold the top and short the bottom, rebalance monthly or quarterly, and combine factors that don’t move together.

The backtest trap: the “factor zoo”. Researchers have reported hundreds of factors, and Campbell Harvey, Yan Liu and Heqing Zhu argued that a new one should clear a much higher statistical bar because so many have been tried. Research by David McLean and Jeffrey Pontiff also found that many effects weaken after they’re published.

10. Carry trade

Time frameMonthsNeedsPrices across the futures curveUsed byFunds and individuals

Carry is what an asset earns if its price doesn’t change. The classic carry trade borrows a low-interest-rate currency and invests in a high-interest-rate one. Ralph Koijen, Tobias Moskowitz, Lasse Pedersen and Evert Vrugt generalized the idea in their paper “Carry”, showing it can be measured in advance across many asset classes. In futures, carry shows up in the shape of the curve. If the curve doesn’t change, a long position earns a positive roll when later contracts are cheaper than nearer ones (backwardation), and the roll costs it money when they’re more expensive (contango).

How quants approach it: rank markets by carry, hold the highest and short the lowest, and rebalance monthly, watching how the curve changes.

The backtest trap: carry tends to earn small, steady gains and then lose sharply in a crisis, when many carry trades unwind together. A short backtest can miss that entirely. Curve shapes can flip quickly, too, so “free roll yield” is never free.

Events and regimes

11. Event-driven

Time frameSeconds to monthsNeedsExact event times and dataUsed byFunds and individuals

Event-driven strategies trade around something identifiable: an earnings release, a merger, a scheduled economic report or a central-bank decision. In stocks, Victor Bernard and Jacob Thomas studied post-earnings drift, where prices keep moving in the direction of an earnings surprise, and Mark Mitchell and Todd Pulvino showed that merger arbitrage behaves a lot like selling insurance against a market crash. In futures, the events are mostly economic releases, which research has shown cause jumps in stock-index, bond and currency futures.

How quants approach it: a precise calendar with exact release times, a clear rule for before, at and after the event, and realistic assumptions about how wide spreads get when the news hits.

The backtest trap: look-ahead bias. Economic numbers are often revised later, and testing with the revised figure, or the wrong timestamp, uses information no trader had at the time. Slippage around releases is also far worse than on a normal bar.

12. Regime switching

Time frameWeeks to monthsNeedsLong history, many conditionsUsed byFunds and individuals

Regime switching means changing a strategy’s exposure or approach when the market’s state changes, for example from calm to turbulent or from trending to choppy. A strategy that works in one state may fail in another, so the aim is to tell which state the market is in and adjust to match. The statistical foundation comes from James Hamilton’s 1989 work on models where the state can’t be seen directly and has to be estimated from the data. In practice it’s usually an overlay on another strategy, such as a filter that only trades a breakout system when volatility is low.

How quants approach it: define regimes with a few simple measures, such as volatility or trend strength, decided before testing, and then check whether the strategy’s results really differ between them.

The backtest trap: hindsight. It’s easy to label regimes once you know how things turned out. Some models even estimate a past state using data from after it, which a trader at the time couldn’t have had. Even the US recession dates economists rely on are only announced months after the fact.

Options and volatility

13. Options strategies

Time frameDays to monthsNeedsHistorical option pricesUsed byIndividuals and funds

Options strategies combine calls, puts and the underlying market to shape a payoff. A covered call collects a premium but caps the upside. A protective put sets a floor at the cost of a premium. A vertical spread defines both the risk and the reward. A straddle or strangle bets on a big move either way (or on calm, if sold), and an iron condor profits if price stays in a range. On their own, these mostly reshape risk rather than create an edge. The edge, if there is one, comes from a view on direction or volatility. Most options on futures exercise into a futures position. American-style options can be exercised, and so assigned, on any trading day before expiry; European-style ones only at expiry.

How quants approach it: start from a view, choose the structure that expresses it with known worst cases, and test with real historical option prices, not prices estimated from the underlying chart.

The backtest trap: “defined risk” isn’t the same as “safe”. Early assignment can break a spread apart, closing quotes are often stale or wide, and a short option without protection can lose far more than it collects.

14. Volatility arbitrage

Time frameDays to weeksNeedsOption prices, implied volatilityUsed byMostly funds

Volatility arbitrage trades the gap between the volatility priced into options (implied) and the volatility that actually happens afterward (realized). Hedging away the direction, called delta hedging, turns an option position into a bet on volatility alone. Research by Peter Carr and Liuren Wu, and by Gurdip Bakshi and Nikunj Kapadia, found that implied volatility in stock-index options has tended to sit above what later happened, a gap known as the variance risk premium.

How quants approach it: compare implied and expected volatility across strikes and expiries, hedge the direction frequently, and size for the rare large move rather than the usual quiet one.

The backtest trap: selling volatility earns small gains most of the time and loses heavily in sudden spikes. A test that misses one big spike tells you very little. The VIX index itself can’t be traded, either. VIX futures behave differently, and because their curve is usually in contango, holding them long has usually cost money.

Market microstructure

These strategies work at the level of individual orders and trades. They’re the hardest to test with ordinary chart data.

15. Market making

Time frameSeconds to minutesNeedsFull order book, queue positionUsed byProfessional firms

A market maker quotes both a price to buy (the bid) and a price to sell (the ask), earns the spread when others trade against both sides, and tries to keep its own position small. Lawrence Glosten and Paul Milgrom showed why the spread exists: some counterparties know more, and the market maker loses to them. Marco Avellaneda and Sasha Stoikov modeled how a dealer shifts quotes to manage the risk of holding inventory while prices move.

How quants approach it: quote and re-quote at very high speed, adjust prices against inventory, widen the spread when volatility rises, and keep fees as low as possible. It needs speed and infrastructure that individual traders rarely have.

The backtest trap: a backtest on bars assumes your order fills whenever price touches it. In reality you wait in a queue, and you tend to get filled exactly when price is about to move against you.

16. Order-flow trading

Time frameSeconds to minutesNeedsTick data, order-book depthUsed byProfessionals and individuals

Order-flow trading reads buying and selling pressure directly: the imbalance between orders waiting at the bid and the ask, how much volume hits each side (often shown as delta or footprint charts), and how the book changes. Rama Cont, Arseniy Kukanov and Sasha Stoikov found that over short intervals, order-flow imbalance at the best bid and ask explains most of the price change in that same interval. Explaining a move as it happens isn’t the same as predicting the next one, and their study used stocks. Futures exchanges such as CME publish which side started each trade, along with the full order book.

How quants approach it: quantify the imbalance, act within seconds or minutes, and model fills and latency as carefully as the entry itself.

The backtest trap: an ordinary chart can’t test this. You need historical tick and order-book data, a realistic model of your place in the queue, and caution with large resting orders that may never trade. Placing orders you intend to cancel to mislead others, known as spoofing, is illegal.

17. Execution algorithms

Time frameMinutes to a dayNeedsIntraday volume and fillsUsed byFunds and brokers

Execution algorithms don’t decide whether to trade. They decide how to fill a trade that has already been decided, at the lowest total cost. TWAP spreads the order evenly over time, VWAP follows the day’s typical volume pattern, and implementation-shortfall algorithms aim to finish as close as possible to the price when the decision was made. Robert Almgren and Neil Chriss described the central trade-off in 2000: trade slowly and you move the price less, but you’re exposed to the market moving away from you.

How quants approach it: measure every fill against the price at the moment of the decision, then adjust speed and order types to reduce that gap. For most individual traders with a few contracts in a busy market, the useful lesson is that measurement.

The backtest trap: assuming you get filled at the average price, or ignoring how much your own orders move the market. Every strategy’s backtest depends on getting this part right.

Models and portfolios

The last three are less about a single trading idea and more about how decisions are made and combined.

18. Machine-learning alpha

Time frameDays to monthsNeedsLots of data, strict validationUsed byMostly quant funds

Machine-learning strategies use flexible models, such as decision trees and neural networks, to predict returns from many inputs at once. Shihao Gu, Bryan Kelly and Dacheng Xiu found, in US stocks, that these models can capture interactions between predictors that simple models miss, and that the most useful inputs were variations on momentum, liquidity and volatility. They also stressed how noisy market returns are compared with any pattern hidden in them. Research on futures has reported similar findings, with gains that shrink once trading costs are included.

How quants approach it: strict separation of training, validation and test periods in time order, simple models first, and costs included from the start.

The backtest trap: leakage, when information from the test period sneaks into training. Marcos López de Prado recommends purging (removing training data that overlaps the test period) and an embargo (a gap after each test period). With so many models and settings to try, overfitting is the default outcome unless the number of tries is counted.

19. Portfolio optimization

Time frameMonthsNeedsEstimates of returns and riskUsed byMostly funds

Portfolio optimization decides how much to put in each position. Harry Markowitz framed it in 1952 as a trade-off between expected return and risk, where combining things that don’t move together lowers risk. The idea is sound, but the inputs are estimates. Richard Michaud showed that it tends to magnify the errors in its inputs, and Victor DeMiguel, Lorenzo Garlappi and Raman Uppal found that none of 14 optimized models consistently beat simply splitting money equally in their tests. Fischer Black and Robert Litterman proposed one well-known fix: start from market-implied returns and adjust toward your own views.

How quants approach it: constrain the weights, prefer stable estimates of risk over guesses about returns, and compare every result with equal weighting.

The backtest trap: weights fitted on the same history used to judge them look excellent and rarely hold up. “Optimal” means optimal for the inputs, not for the future.

20. Risk parity

Time frameMonthsNeedsVolatility and correlation estimatesUsed byMostly funds

Risk parity gives each asset an equal share of the portfolio’s risk instead of an equal share of its money. Edward Qian pointed out that a classic 60/40 stock and bond portfolio looks balanced in dollars but gets most of its risk from stocks. Bridgewater’s All Weather approach, launched in 1996, spreads risk across economic environments instead of predicting which one comes next. Balancing risk means holding far more bonds than stocks, so to reach a useful overall level of return, risk parity portfolios usually add leverage, often through futures.

How quants approach it: estimate each asset’s volatility and correlations using only data available at the time, rebalance on a schedule, and account honestly for the cost of leverage.

The backtest trap: it relies on assets not falling together. In 2022, stocks and bonds fell at the same time, and leverage made that worse. A backtest dominated by a long stretch of falling interest rates can flatter the approach.

Where popular retail approaches fit

Many traders describe their approach in other words: smart money, price action, indicators. Written down as exact rules, most of these fall into the families above.

• ICT and smart-money concepts describe setups such as fair value gaps (a gap left between candles by a fast move), liquidity sweeps (price running past a prior high or low, then turning) and order blocks (the last opposing candle before a strong move). As rules, a sweep-and-reverse setup is a short-term reversal, and entering a pullback into a fair value gap in the direction of a strong move is a momentum-style continuation. Rules about which hours to trade act as filters.
• Price action traders read bars, swings and support and resistance without indicators. Their setups usually come down to breakouts, trading through a level, or reversals, fading one. Both become testable once the levels are defined precisely.
• Indicator strategies sort the same way. RSI and Bollinger Band setups are usually mean reversion. Moving-average crossovers and trend-strength filters such as ADX are usually trend following.

Whatever the vocabulary, the test is the same: exact rules, realistic costs, and data the rules were never tuned on. Kwaant’s strategy language includes terms like fair value gaps, swings, market structure and opening ranges, and when traders define a term differently, it asks which version you mean.

What every strategy has in common

Strip away the names and every strategy above faces the same questions. Did it work on data it was never tuned on? How many versions were tried before this one? Does it survive realistic costs and fills? Does it hold up across calm and volatile markets? Our guide to telling a real edge from luck covers six checks that apply to all of them, from walk-forward testing to keeping a sealed test you open only once.

A few questions also help you see whether a type suits you before you spend time testing it:

• What data does a fair test need? A strategy that depends on the order book can’t be judged on a price chart.
• How often does it trade? The more trades, the more costs matter and the more an edge must clear them.
• What does losing look like? Trend strategies lose often and win big. Reversion and carry win often and lose big. Know which kind of pain you’re signing up for.
• Can you explain why it should work? A reason the edge might exist, such as a risk being paid for or a behavior that repeats, is worth more than a good-looking backtest.

Whichever type you trade, what matters is whether your version of it holds up when it’s tested honestly. That’s the testing Kwaant is built for: you describe the strategy you trade in plain English, and it’s tested on 16+ years of historical CME Group data, with every variation checked for luck.

Sources and further reading

• Jegadeesh and Titman, Returns to Buying Winners and Selling Losers (Journal of Finance, 1993)
• Moskowitz, Ooi and Pedersen, Time Series Momentum (Journal of Financial Economics, 2012)
• Hurst, Ooi and Pedersen, A Century of Evidence on Trend-Following Investing
• The Original Turtles: the Turtle trading rules
• Gatev, Goetzmann and Rouwenhorst, Pairs Trading: Performance of a Relative Value Arbitrage Rule
• Avellaneda and Lee, Statistical Arbitrage in the US Equities Market (Quantitative Finance, 2010)
• Asness, Moskowitz and Pedersen, Value and Momentum Everywhere (Journal of Finance, 2013)
• Harvey, Liu and Zhu, ...and the Cross-Section of Expected Returns
• Koijen, Moskowitz, Pedersen and Vrugt, Carry
• Mitchell and Pulvino, Characteristics of Risk and Return in Risk Arbitrage (Journal of Finance, 2001)
• Hamilton, A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle (Econometrica, 1989)
• Carr and Wu, Variance Risk Premiums (Review of Financial Studies, 2009)
• Cont, Kukanov and Stoikov, The Price Impact of Order Book Events
• Gu, Kelly and Xiu, Empirical Asset Pricing via Machine Learning
• DeMiguel, Garlappi and Uppal, Optimal Versus Naive Diversification (Review of Financial Studies, 2009)
• Qian, Risk Parity Portfolios (PanAgora, 2005)