Request Information

Request Information

Investors Week

One of the biggest changes in financial markets over the past decade or so has been increasing interdependence. Securities that previously went their own sweet way now seem inextricably entangled, but predicting the practical effect of these relationships is by no means straightforward. Assuming that if Market A opens higher, Market B will do the same is unwise, as Markets C-Z may also have their own varying impact as well. While there are a handful of experienced traders capable of making sense of these multiple interrelationships, most of us mere mortals have to muddle along as best we can.

Perhaps not. As mentioned in the recent review of Ward Systems' NeuroShell Trader, artificial intelligence (AI) software is ideally suited for exactly this sort of task. However, with most AI programs the trick lies in knowing which markets to throw into the analytical melting pot. While most packages have short-circuited the hours (or even days) of data preparation that were previously necessary - the user is still left with the task of deciding what is relevant.

An exception to this rule is Market Technologies' VantagePoint suite of programs. Market Technologies has been researching inter-market analysis and AI for more than a decade, with the first version of VantagePoint being released in 1991. The program consists of a core application that can be coupled with a range of market-specific modules for predicting short term moves in most major markets.

Each market-specific AI module uses data from a range of other markets that Market Technologies has tested and found of predictive value. For example, the FTSE 100 future model includes eight other markets in its calculations, including various stock indices such as the Dow, Nikkei, NYSE Composite and S&P 500, as well as US T-Bills/Bonds and the US Dollar index.

User intervention throughout is kept to an absolute minimum - all you have to do is update the daily data (CSI and Genesis are the formats currently supported), generate the analysis reports and decide how to use them.

The software can display its output in three basic formats. A daily report shows actual and predicted data for a user definable number of days up to the current date. A history report, which is similar to an expanded version of the daily report, also gives the user the option to decide exactly which data is displayed. Finally, for those who prefer a graphic representation the output can be displayed in a chart window.

All three output formats can use either overlay or displaced mode. In displaced mode, predicted data is shifted forward the appropriate number of days. (So, for example, a 10-day moving average that predicts a value four days ahead would be shifted forwards four days.) In overlay mode, predicted data is overlaid on the date it was generated next to actual data for that same date. Needless to say it's vital to realise which display mode a report is in, so this is clearly highlighted at the top of each type of report.

There are now more than 20 market specific AI modules available for VantagePoint including stock indices, currencies, energy, and interest rates.

Perhaps unusually for a US vendor, Market Technologies includes modules for overseas markets, such as the FTSE 100 future and Brent crude. Each market-specific module consists of five separate neural networks, which are used to make various predictions about the future performance of the security. (Since VantagePoint is not an automatic trading system, these predictions are intended to provide decision support rather than outright buy or sell signals.)

These predictions are tomorrow's high, tomorrow's low, a five-day moving average (of closes) projected two days ahead and a 10-day moving average projected four days ahead. The fifth figure generated is a proprietary "Neural Index", which fluctuates between 0 and 1. A value of 1.0 indicates that the trend of the market over the next two days is expected to be up relative to the last two days - vice versa for a value of 0.

In addition to these core predictions, VantagePoint also uses them as inputs to produce two crossover oscillators, called PTS and PTM.

PTS is generated by calculating the difference (in ticks) between the predicted and an actual five-day moving average, while PTM applies the same principle to the 10-day average. These AI oscillators can be used in a similar fashion to conventional oscillators, by taking long and short positions based on crossovers. The most aggressive traders will probably be happy to do this on the basis of signals from just the shorter PTS oscillator, while those of a more cautious disposition, will probably want to wait for the longer PTM oscillator to confirm the signal.

An alternative possibility is to combine the predicted five- and 10-day moving averages into a moving average crossover tool, which (obviously subject to the quality of the predictions) won't suffer from the traditional lag problem of moving averages as it uses predicted values.

Other traders may prefer even more straightforward methods - for example short term option traders could use the prediction of the next day's high and low as simply a timing tool. A popular further alternative among VantagePoint users is to combine all the VantagePoint tools into a standalone trading strategy. For example, a short position would be opened only if the predicted high and low for the following day were both below those for the current day, the PTS and PTM oscillators concurred on a short signal, and the Neural Index had fallen to zero.

Nevertheless, the value of these strategies is obviously dependent on the quality of the raw predictions, and the VantagePoint modules provided for review (FTSE, S&P and Dow) didn't disappoint in this respect. For example, calculating the correlation between the predicted and actual highs and lows produced figures typically above 0.7 for all three markets over a selection of time-frames and contract months. That is impressive when you consider that stock indices and futures are notoriously prone to noise, which makes accurate prediction particularly difficult.

However, such accuracy comes at a price, with a single VantagePoint market module costing $2995, though there are volume discounts if you take multiple modules.

Nevertheless, a broad range of traders seem happy to pay that price and the Market Technologies' website has a large set of appreciative testimonials from named (and pictured) traders ranging from beginners to professional money managers.

Review Date: 
Thursday, January 1, 2004