- 5 Rocket Stocks for Gluttonous Turkey Day Gains
- Time to Sell These 5 'Toxic' Stocks
- 5 Earnings Short-Squeeze Plays
- 5 Must-See Charts
- 5 Stocks With Big Insider Buying
Build Your Own Trading System: Technical Primer - views
BALTIMORE (Stockpickr) -- Imagine turning on a machine each morning and watching it rack up gains by trading the markets. While that scenario sounds like something most investors just fantasize about, it’s become a reality for many traders.
In this technical primer, we’ll take a brief look at how you can join them by building your own trading system.
Today, the barriers to entry have fallen dramatically for trading system developers. You don’t need a Ph.D., a multi-million dollar mainframe or server space at the NYSE to achieve success as an algorithmic trader. While building a trading system may go over the head of the average casual investor, it’s something that’s finally within reach of traders who aren’t bankrolled by large financial institutions.
First, it’s important to think about how quantitative analysis relates to technical analysis as a whole.
For the most part, you can think of quantitative analysis as a subsection of technical analysis (it’s worth noting, though, that some quantitative analysis is actually fundamentally driven). While technical analysis is any analysis that’s driven by market data, quantitative analysis throws out subjective elements.
Quants use trading systems (a set of well-defined rules or algorithms) to make investment decisions. When most investors hear the word “quant,” they think about the scientists who work on high-frequency trading systems for institutions. That’s not what we’re talking about here -- instead, the algorithmic trading we’re talking about is better suited to longer timeframes (daytrading through position trading).
While we’re barely scratching the surface of algorithmic trading in this primer, my goal is to give you a glimpse at how trading systems work, and where you can take the next steps toward creating one of your own.
An Introduction to System Development
We’ll start out with system development, that is, figuring out what rules you want to use to trigger a trading signal. The first step is choosing the factors (or inputs) that you’ll be using to generate investment decisions. It’s important to keep in mind that any inputs need to be quantifiable, so criteria like “good management” or “ample trading liquidity” can’t be part of the system (instead, “ample trading liquidity” would need to be something like “average trading volume greater than 1 million shares per day”).
So how do you know what inputs you should use?
There are two main ways of building a trading system: by logically thinking up criteria that you’d like to use (trial-and-error) or by data mining (using computers to sniff out the best factors). Logical rules can be hit or miss, but data mining exposes traders to concerns like computing power and data mining bias. Simple trading rules actually work more often than many quantitative traders would like to admit (one of the reasons the industry has developed fancy words for otherwise simple statistical concepts) -- so I’d recommend starting off with the trial-and-error unless you’re fairly comfortable with statistics.
After you’ve got a system, you need to check whether it works. To do that, we turn to backtesting. Backtesting just means taking historical price data and applying it to your system to see what sort of results it would have generated in the past after transaction costs (and other costs, like slippage). When you’re doing that, one of the most important considerations is going to be that you’ve got bias-free data (the books at the end of this primer can help you find just that).
In a backtest, you’re not just looking for a high percentage return. Other factors, such as maximum drawdown, market exposure, and risk-adjusted return are also important considerations. In essence, you’ll only want to put your money at risk with a system that’s consistently producing good performance -- 99 losses followed by one huge offsetting gain does not a good system make.
In the past, you’ve probably heard the SEC’s favorite phrase: “Past performance does not indicate future returns.” That’s not exactly true. But like most investment myths, there is a thread of truth to it at least -- which is why backtesting isn’t the final step. After backtesting, you’ll want to see how your system performs independently (either through walk forward testing or out of sample testing) before you put your cash on the line.
With a market-beating system locked down, you’re most of the way to actually making money as a quant.
Putting it Into Practice
To put your system into practice, you’ll need a few things. Your equipment setup is important; you’ll need a reasonably fast computer, a quick internet connection, and a data feed to push your inputs through your trading system. You’ll also need software (such as TradeStation, Amibroker, or MATLAB) to actually apply the system to the data and tell you when a trade gets signaled.
Then comes the question of how hands-on your trading is going to be.
Unless you’ve got programming knowledge (or are willing to spend money to hire a programmer to connect your trading software with your brokerage account), you’re probably going to have human interaction with your system. It may just be a matter of executing a trade whenever the system tells you to. Having a fully automated system can bring up a slew of new potential nightmares (such as out of whack position sizing), so it’s rarely a bad idea to have some human oversight over each of your trades.
Before you start trading in the real world, other considerations like risk management and the statistical significance of your returns are worth weighing. Also, keep in mind that the market is constantly changing; if your system stops working, be prepared to stop trading it.
Obviously, this primer only scratches the surface of understanding how trading systems work. To take your quantitative trading to the next level, there are a few books that do a particularly good job of introducing retail traders to the more advanced trading system concepts.
The first is Quantitative Trading by Dr. Ernie Chan, a book that does a good job of showing traders step-by-step how to create an algorithmic trading business, including data vendors and software packages that are used by successful systems traders.
Two other books worth looking at for more advanced quants-in-training are Perry Kaufman’s New Trading Systems and Methods and David Aronson’s Evidence-Based Technical Analysis. Each of these books takes an empirical look at systems development -- including example rules and backtests. They’re definitely recommended reading for anyone who’s planning on putting money to work in an algorithmic trading system.
For more technical analysis primers, visit the Technical Analysis Hub on Stockpickr.
Jonas Elmerraji, based out of Baltimore, is the editor and portfolio manager of the Rhino Stock Report, a free investment advisory that returned 15% in 2008. He is a contributor to numerous financial outlets, including Forbes and Investopedia, and has been featured in Investor's Business Daily, in Consumer's Digest and on MSNBC.com.