One of the most challenging problems in systematic trading is the design of a profitable trading strategy that generates steady growth of capital. Markets constantly change, and changes tend to happen faster than ever; October 2008 is a good illustration of an unexpected drop in all major indexes. Wild jumps of volatility in the fourth quarter of 2008 resulted in catastrophic losses for many funds, and the whole financial sector was is crisis after a matter of weeks. In this perspective, capital protection becomes a sore spot. Comprehensive systematic analysis of the issue proves the demand for multi-layered protection that must deploy:

early detection of coming market correction and crisis,

adaptive position sizing as a function of volatility,

exit rules for big gaps and intraday moves,

automatic close of all positions if volatility or current draw downs cross the threshold,

real time monitoring of irrational market reactions from news.

Most trading systems unable to respond this specification, big players switch to alternative technologies that substitute risks with short-term, but steady profit. By the example of high frequency trading, these technologies require large capital commitment leaving the majority of the market overboard. Quant’s work becomes the core of success. However, this intellectual resource is expensive and not available to everyone. Still, our high-tech society is ready to offer an alternative solution – Artificial Intelligence. Unlike human brain, A.I. is faster, cheaper, and emotion-free. Analysis of A.I. implementation in financial markets with a perspective of its affordability as well as efficiency shows a few proprietary technologies are already in the market. These systems have several layers of capital protection: volatility based position sizing, intraday stops, and a draw down threshold. They use:

complex scenarios;

joint optimization of different types of trading systems;

automated intelligent trade monitoring.

Combining several patterns into a complex scenario obviously potentiates more profit in comparison with a single pattern. Joint optimization of different types of trading systems helps to reduce draw down. The process includes two steps:

a few trading systems are designed on different types of scenarios; for example, one system is based on large waves, another system uses momentum and geometry;

trading systems are organized into a group that is tested as a meta trading system.

Joint optimization implies more sophisticated decision making. Since a meta system includes more than one non-correlated or weakly correlated trading systems, profitability is increased. Automated intelligent trade monitoring employs a knowledge base of different types of exits. It protects profitable trades from hitting the stop-loss. As well, it may trigger exit signals on a spike, thus increasing the average profit per trade.

For robust trading strategies it’s a good practice to use ETFs that have lower risk than stocks. ETFs are liquid and their liquidity grows rapidly. ETF’s price is a weighted average of prices of many stocks (for example SPY includes 500 stocks from S&P list), thus price patterns on ETFs are more predictable than on single stocks. Low-risk high-return robust trading strategies are designed completely automatically.