Our past trading experience, in both bull and bear markets, has allowed us to formulate a set of "rules" (or perhaps guiding principles) which influence the way we build systems and trade today. These rules are as follows:
1. Don't trade "manually"; always use computer-based automated systems.
Our basic premise is that, if you can't explain to a computer how your trading system works then you are unlikely to be able to trade it reliably and profitably yourself. Also, there are some systems that can only be practically implemented by machine; these are typically when you are looking for rare distortions (setups) that occur so infrequently that it is necessary to scan thousands of securities in order to make sufficient trades for the system to be profitable. Given rules 3 and 4, most of our trading systems currently fall into this category.
Another fundamental benefit of fully automatic trading is the ability to sidestep the dangerous emotions of "fear and greed". Such emotions are generally held responsible for the shocking (but likely true) statistic that over 90% of all discretionary, retail traders consistently lose money.
2. Go Long wherever possible.
Conventional wisdom has it that the best way to profit in a bear market is to go short. It is our experience that this is often not the case. Short trading has a number of disadvantages (e.g. adverse percentages, negative dividends etc) that mean we like to avoid taking short trades wherever possible. At the time of writing all our active trading systems only take long trades although short positions in currency or index futures are sometimes used to hedge positions when it is deemed propitious to do so.
3. Look for systems that exploit distortions in the market.
We believe it is important to search for distortions or inefficiencies in the market that other traders have not yet found or, at least, have not yet found a way of manipulating. For example: the price pressure build up at market open, coupled with human emotions of fear and greed can very often result in tradable distortions in the market as stocks gap up or down.
4. Trade portfolios rather than individual issues.
Because the sort of distortions that we look for in a trading system normally only occur rarely, it is usually necessary to scan a system over a large portfolio of securities in order to increase the number of setups available to trade within any time period. If one is not taking enough trades the temptation is to dangerously increase position size (and therefore risk) in order to boost profitability.
5. Keep systems simple.
We backtest all our systems on historical data in order to evaluate potential profitability. Obviously, as a system is being developed, we adjust a number of its parameters to maximize profitability or some other appropriate aspect. If the system has a large number of independent variables then there is a grave possibility that one might "curve fit" the system to address only the historical data that one has available and that it will be unlikely to work well into the future. There are a number of technical ways to address this (e.g. performing multiple backtests and forward tests) but most importantly we like to keep the system simple to reduce the number of independent variables.
6. Don't believe your analysis tools.
If we uncover what appears to be a successful trading system then we try relentlessly to break it. Unfortunately, if one doesn't try to do this then at some point the market surely will.