ACADEMY / Strategy generation

Quantitative trading firms: what they do and what retail can copy

Quant firms earn in different ways, but they share one research process: a written hypothesis, tests with realistic costs, out-of-sample validation and risk limits. An individual cannot copy their speed or capital. The research discipline is free to copy.

By Soren Huang · Published 2026-10-20 · updated 2026-10-20 · Editorial policy

Contents

What does a quantitative trading firm actually do?

A quantitative trading firm turns market ideas into rule-based trading programs and runs them under risk limits. Market makers, prop trading firms and systematic funds differ in whose money they trade and how fast they trade. An individual futures trader cannot copy their speed or capital. They can copy the research discipline: realistic costs, out-of-sample testing and portfolio thinking.

"Algo trading company" is a looser term. Sometimes it means a firm that trades with algorithms, and sometimes a company that sells trading software or signals. This article is about firms that trade.

The useful question is not which firm is biggest. It is how these firms decide that a strategy deserves real money. That process can be copied, and most of it costs only time and patience.

What types of quant trading firms are there?

Most quant trading firms fall into three groups, and the main differences are whose money they trade, how they earn and how long they hold positions.

A market maker works like a currency exchange booth at an airport. The booth always shows a buying price and a selling price, and it earns the small difference many times a day. A market maker is a firm that continuously offers to buy and sell a contract, earns the gap between the two prices (the spread), and manages the risk of the positions it is left holding. Many exchanges run market maker programs. These programs lower fees or pay rebates to firms that commit to quoting prices.

Proprietary trading is trading with the firm's own money instead of client money. A prop trading firm keeps its profits and carries its own losses. Its strategies range from market making to arbitrage, which is trading the price gap between two related instruments, to statistical strategies held for minutes or days. Some businesses that use the name "prop firm" sell evaluation programs to individual traders. In those programs, the trader pays a fee to trade an account under set rules. That is a different business model from a firm that trades only its own capital.

A systematic fund is an investment manager that trades outside investors' money with rules and charges fees for doing so. In futures, these managers are often called CTAs (commodity trading advisors), a term that comes from US futures regulation. Many CTAs trade across dozens of markets and hold positions for days to months.

Type	Whose money	Main source of return	Typical holding time	What matters most
Market maker	The firm's own	The spread and exchange incentives	Seconds to minutes	Speed, fees, inventory control
Prop trading firm	The firm's own	Its own strategies, from very fast to multi-day	Seconds to days	Execution, risk limits, research output
Systematic fund	Outside investors	Trends and other patterns across many markets	Days to months	Diversification, capacity, risk control

How does a quant firm research a strategy?

Quant firms treat every strategy as a claim that must survive several separate tests before it gets capital, and most ideas fail along the way. The order of steps matters more than any single tool:

Hypothesis. Write down why the pattern should exist, for example "prices tend to keep moving after a strong open because large orders take time to fill." An idea with a reason is easier to test honestly than a pattern found by searching.
Backtest with costs. Test the rules on historical data with commission, slippage and realistic fill timing from the first run.
Out-of-sample validation. Check the rules on data that was not used to build or tune them.
Risk limits. Decide in advance the maximum position size, the daily loss limit and the drawdown at which the strategy stops.
Small live size. Trade a small size first and compare live fills with the backtest.
Monitor or retire. Review the strategy on a schedule and switch it off when it breaks its limits.

The research funnel: each stage removes ideas, and only the survivors reach live trading.

Firms also count how many variants they tried. This matters because the best result out of many attempts looks good partly by luck. Bailey and co-authors (2014) show that when enough variants are tested, a strong backtest can appear by chance alone. Harvey and Liu (2015) argue that reported results should be reduced to account for the number of strategies tried.

What can't an individual trader copy?

Individuals cannot copy the infrastructure, the speed or the capital of large firms, so the sensible choice is to trade where those things do not decide the result.

Latency. Latency is the time between a market event and your order reaching the exchange. Fast firms use co-location, which means placing their servers inside or next to the exchange's data center. A home connection cannot compete with that.
Fee structure. Market makers often pay lower fees or receive rebates through exchange programs. A retail account pays standard commissions, so very small edges per trade disappear.
Capital. Large capital lets a firm trade many markets at once and survive long losing periods. A small account may only fit a few contracts.
Data and staff. Firms buy detailed order-book data and employ teams to clean it, build tools and watch risk all day.

Think of a Formula 1 team. You will not beat its pit crew on a racetrack. You can still choose a long road trip, where pit-stop speed does not decide who arrives. For an individual, the road trip is bar-based strategies that hold positions for many minutes, hours or days. In those strategies, a few milliseconds of delay cost very little compared with the size of the move.

Individuals also have advantages. A small account can trade where a large fund cannot fit its size, needs no investor reports and can wait as long as it wants for the right setup.

What can an individual trader copy from quant firms?

Realistic costs, out-of-sample testing, risk limits and portfolio thinking cost nothing to adopt, and together they catch many strategies that only looked good because costs were missing or the same data was used to build and judge them. AlphaPrime (Valynox) is built around these habits. AlphaPrime evaluates each candidate with fees, slippage and next-bar execution in its backtesting, so costs are part of the result from the first run.

Put realistic costs in from the start

Slippage is the difference between the price your backtest assumed and the price you would actually get. Firms model it carefully because a strategy's edge per trade is often small. The detailed method is in the guide to trading costs in backtests: commission, slippage and the average trade.

Hypothetical example

A strategy averages $40 per trade before costs. Commission is $5 per round trip, and slippage is $10 on each side, so the total cost is $25. The net result is $15 per trade. Now double the slippage to $20 per side. The total cost becomes $45, and the strategy loses $5 per trade. That strategy depended on cheap fills.

Test on data the strategy has not seen

Out-of-sample data is history that was kept aside and not used to build or tune the strategy. Firms judge a strategy mainly on that data. Walk-forward analysis repeats this check across many periods, as the article on walk-forward analysis and why one backtest is not enough explains. AlphaPrime's strategy validation includes walk-forward analysis and lets you set your own thresholds.

Set risk limits before the first live trade

Firms decide in advance how many contracts a strategy may trade and at what loss it stops. You can do the same. Choose a risk per trade, then use the position size calculator to turn your stop distance into a number of contracts. Also write down the drawdown at which you will stop and review the strategy, for example a multiple of its worst historical drawdown.

Think in portfolios, not single strategies

Correlation is a measure of how closely two strategies' results move together, from moving in step to moving in opposite directions. Strategies with low correlation tend to have their bad periods in different months, so the combined account falls less than the sum of their worst drawdowns.

Hypothetical example

Strategy A has its worst drawdown of $9,000 in year 2. Strategy B has its worst drawdown of $8,000 in year 5. If both worst periods hit at once, the account would fall $17,000. Because they happened in different years, the combined account's worst drawdown in this example is $10,000.

Hypothetical example: two strategies whose bad years do not overlap give a combined drawdown far smaller than the sum of their worst drawdowns.

How to combine strategies and read the joint drawdown is covered in portfolio backtesting: combine strategies and check the drawdown.

Keep a research log

Write the hypothesis before you test it, and record every variant you tried. The log tells you how many attempts stand behind your best result. That number is what the research on data snooping asks you to account for (White, 2000).

Why is this not a list of top quantitative trading firms?

This article gives an overview of how firms work, not a ranking or recommendation, because no ranking would tell you how to trade your own account. Firms differ in size, revenue, staff and returns, and most of these figures are private or out of date. Lists of "top" or "best" firms use different criteria, so they often disagree. A firm's results also depend on its infrastructure and capital, which you do not share.

The same questions are useful for any company in this field, whether it trades or sells tools. Are costs included in the results? Was the strategy judged on data it had not seen? How many variants stand behind the result? What risk limits apply? Those questions work for your own strategies too.

What this means if you use AlphaPrime: you can apply the research stages of that funnel in AlphaPrime: costs and next-bar execution in every backtest, and out-of-sample checks with walk-forward analysis in strategy validation. The strategy databank keeps your results organized for review, so the log of what you tried stays in one place.

Frequently asked questions

What is a quant trading firm?

A quant trading firm is a company that trades with rules built from data and statistics, usually executed by computer programs. The main types are market makers, prop trading firms and systematic funds. They differ in whose money they trade, how they earn and how long they hold positions.

What is the difference between a prop trading firm and a hedge fund?

A prop trading firm trades only its own money and keeps its own profits and losses. A hedge fund or systematic fund manages outside investors' money and charges fees for that work. Both may use quantitative methods, but they answer to different owners.

Can a retail trader compete with algorithmic trading companies?

Not on speed, fees or capital, which large firms control through infrastructure and exchange programs. A retail trader can compete in bar-based strategies held for minutes to days, where a small delay costs little. Small size is also an advantage, because it fits into markets and setups too small for large funds.

How do quantitative trading firms avoid overfitting?

They write a hypothesis before testing, include realistic costs, judge strategies on out-of-sample data and count how many variants they tried. The more variants they tested, the more they discount the best result, because some of its strength is likely luck. Any individual can apply these steps.

Which are the best algo trading companies?

There is no reliable ranking, because firms' results, sizes and criteria are mostly private, and published lists measure different things. For an individual trader, the more useful question is whether a firm's research habits can be applied to their own strategies. Most of them can, starting with costs and out-of-sample testing.

References

Bailey, D. H., Borwein, J., López de Prado, M., & Zhu, Q. J. (2014). Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance. Notices of the American Mathematical Society, 61(5), 458–471.
Harvey, C. R., & Liu, Y. (2015). Backtesting. The Journal of Portfolio Management, 42(1), 13–28.
White, H. (2000). A Reality Check for Data Snooping. Econometrica, 68(5), 1097–1126.

What AlphaPrime does here

AlphaPrime's strategy generation turns research ideas into testable rules. Its backtesting includes fees, slippage and next-bar execution, and its strategy validation checks each candidate on out-of-sample data with walk-forward analysis.

See plans and pricing