Financial modeling, TA, efficient markets, etc.

Three questions we're frequently asked are: What exactly do we do here at Mantic Software? How is financial modeling related to, or different from, technical analysis? And what sorts of investment strategies do we favor?

Mantic Software isn't in the investment advisory business, but our attitude toward investing is essentially conservative, in the sense that we try to understand what an investment is really worth, and how its value will change if circumstances change. Our modeling tools reflect this goal.

This definition of conservatism doesn't preclude speculation; a speculation may be very attractive if you understand it accurately. We just like to quantify investment ideas, and if possible estimate the probability-weighted payoff. Even though such estimates are far from perfect, they provide a consistent way to compare alternatives, and temper any tendency to excessive optimism or pessimism.

We don't put much faith in "efficient markets" theories, at least in the short term. Markets do generally, eventually, correct themselves when they diverge too far from the true value of things. But this process can take a long time -- years, sometimes -- and usually overshoots when corrections finally occur. Sometimes a minor correction never really is evident, because circumstances change. Markets are subject to communication delays, investor avalanche behavior, mutual fund managers, lies, government interference, panic, misjudgments, hyperventilation and AOL.

Charles Kindleberger's book, Mania, Panics and Crashes: A History of Financial Crises is the classic text and a fast read on this subject. In more modern terms, we think stock price movements in particular exhibit fractal properties on large and small time scales, and self-organizing "crowding" behavior when new information arrives.

Historical information helps you understand the limiting relationships that eventually force major market turns. History is particularly valuable for assessing interest rates, bond market behavior, and the current relationship between stock prices and lower-risk investment alternatives. We believe it's less useful for forecasting individual stock prices, except for the very important observation that tomorrow's price will usually not wander too far from today's. On the other hand, we've met a very small number of people whose investment performance seems to contradict this position.

Of course, what you really want to know is when that price will suddenly move a long way, which is when self-organizing avalanches come into play.

Technical analysis is mostly about internal market processes and current relationships. The analyst looks for situations where prices and market cash flows behave unusually, and tries to anticipate short-term future behaviors of individual securities, commodities, indexes, or market segments. In other words, TA seeks to accurately describe future scenarios from the relationships among indicative measurements. Most current TA tools are variations on a few simple ideas. A real breakthrough will come if and when we find ways to directly model those internal market processes rather than just guess at them from price and volume numbers.

Financial modeling uses the mathematics of probability distributions, valuation models, hedging theory, and the time value of money to provide quantitative assessments of future scenarios. It's about risk exposure and reward potential. This activity is qualitiatively different from technical analysis, but the two fit together. Insights from technical insights and historical parameters can help create scenarios to run through modeling tools. This is what we concentrate on at Mantic Software. We've been very busy developing a general-purpose modeling infrastructure (i.e. software support) to facilitate this sort of analysis.

Statistics, unfortunately, throw away much more information than they capture. But lacking crystal balls (other than our own logo and a flawless Baccarat sphere we bought at Harrod's), statistics provide some of the best tools for evaluating future scenarios. And of course, the most widely used financial analysis techniques, such as option price theory and modern portfolio theory, depend fundamentally on "random walks" and related ideas.

If the subject of random walks interests you, take a look at the Binomial Market Model article and software at this web site.