Once you have a stock prediction model the Trading Rules tell you how to implement that model. Some rules are independent of the model and are optimised independently of any optimisation done in the model. And some rules are concerned with detecting when the model does not apply to the current trade.
This section of our expertise is mainly about implementing quantitative models rather than being about technical analysis. But some applications are useful for technical analysis. It is difficult to optimise trading rules in the absence of a quantitative model, however. The practice of optimising rules based on backtesting is dangerous. Regression to the mean will certainly apply which means that future trading using those rules will be less profitable than in the past.
Optimal Trade Execution refers to minimising the cost of trade execution where the major costs are: liquidity costs and brokerage. Liquidity costs are quoted bid-ask spreads, effective spreads (which allow for executions within the quotes), and realized spreads (which measure price reversals after trades).
Optimisation methods commonly include breaking up trades into smaller parcels and timing execution of those parcels. This requires a quantitative model of the market and access to market depth. Models for the market already exist such as the Madhavan, Richardson, and Roomans model.
A filter is a calculation done on a time series which gives a new time series such as a moving average. Technical analysts use filters such as trend filters (which engineers call low pass filters) and oscillators (which engineers call high pass filters). We use filter design software from engineering to design this style of filter.
But more powerful filters come from using the quantitative models that we are basing our trading on and from time series analysis. We can fit State Space Models using the Kalman Filter and more advanced Bayesian models using Particle Filters. When you have a quantitative model and filter you are able to calculate probabilities and measure of goodness-of-fit for models. Technical analysts not using quantitative models have to use backtesting to get these quantities.
Regime Switching concerns any kind of change in market conditions that you wish to model. Perhaps the most profitable regime change to detect is the change from a bull market to a bear market. Or from recession to economic growth.
Models generally fit better and are more powerful when they are tuned for specific regimes. A simple example is the use of beta to measure how far a stock price changes when the market changes. Usually a stock that goes up faster than the market when the market goes up has an even larger downward drop when the market drops. So it would not be appropriate to use the same beta value for both bull and bear markets.
When regime switching is built into a model the model not only fits better, it has the extra output of predicting regime changes.
A specific application of Regime Switching is the stop loss. A stop loss is a rule that tells you to exit a trade when the model no longer fits the stock or when a regime change demands an exit.
The simplest stop loss rule is to exit a trade when a stock drops below a certain price. But this is a pretty crude rule. For a start, such a rule mathematically increases the probability that the trade will lose money. Secondly, just because the stock has dropped below a certain point it does not mean that the model no longer applies. And thirdly, a single drop may be due to noise. Statistically we would want some cumulative evidence that the model is invalid rather than a single point of data.
Once you have a model and all the trading rules you will want to combine all the rules into a trading plan and optimise the overall expected return of the plan.
But you won't be finished there. Once you have optimal trading how much do you spend on each trade? You will then want to do overall portfolio optimisation . This will let you spend more on trades that improve your return to risk ratio and less on trades that don't.
Volatility trading includes volatility measurement and prediction.
Check out our award winning paper on volatility
If you have a technical trading rule (i.e. one that trades based on a price series) you can boost the performance of the rule by incorporating fundamental information.
If you are interested in momentum trading and rotational strategies then you can do this more effectively by using pairwise asset analysis.