I had the honour to attend, and present, at the 50th annual conference of the CMT: Chartered Market Technicians. While everyone in finance uses charts, Market Technicians are the pinnacle of professional financial chart users: charts are core to their analyses, recommendations, actions and trading strategies.
Here’s a few comments on some of the sessions. There’s a few snapshots too: visualization people who aren’t financial professionals may be interested to see some differences and perhaps a few insights.
To start, technical analysis is the notion that price patterns in markets result from changes in supply and demand. But demand isn’t necessarily rational: people have fear of losses, fear of missing out, fear of not following trends that all their neighbours are talking about at dinner parties. We’re social, we tend to have herd behaviour in markets, whether bitcoin, meme stocks, houses (2007), railroads (1840), tulips (1634), etc. While these examples are market bubbles, there are other similar patterns in stock prices, bond prices, house prices and so forth. Patterns may similarly exist in inflation and other economic data too. The notion of trends and patterns are a feature of markets, and these trends and patterns can be observed, analyzed, communicated and traded. The Chartered Market Technicians (CMT) is the association that formalizes technical analysis and provides training and certification for professionals.
Many timeseries
David Keller – chief market strategist at Stockcharts.com – hosts a daily TV show on technical analysis. I’ve had the opportunity to work with Dave in the past, and one thing of note is that timeseries analysis is rarely done in isolation: timeseries are compared to other timeseries, events, derivations and so on. Here’s one of Dave’s charts from his show which he broadcast live at the CMT event:
Hand-drawn charts
David had an opportunity to interview legends of technical analysis Ralph Acampora (who painted a 70 foot chart of the Dow Jones Industrial Average on his barn), and Louise Yamada (who pulled out a paper chart during Dave’s interview). Both talked about their early careers when they had to manually calculate averages, plot their charts by hand and learned to get a sense of the markets from the physical charts.
Annotated charts
Many technical analysts annotate charts: drawing trend lines, support and resistance lines, indicating peaks and troughs (sharp or gradual), other patterns (head and shoulders, wedges) and textual notes such as prices or key events. Whereas the visualization community is questioning whether or not 2-4 annotations are too many, the technical analysis community is well inclined to have 10 or more annotations (such as Ralph’s barn, or Louise’s markup on her charts).
Paul Ciana – chief of fixed income, commodity and currency technical analysis at Bank of America – presented some work analyzing trends in unemployment. Here’s a fantastic chart looking at employment filled with annotations explaining unemployment trends:
Statistical analysis
Many technical analyses use statistical approaches to assess data. A simple example is Ned Davis use of overlaying a mean or regression line on a timeseries chart. In a trending chart, whether economics data or price data, it can be clearly seen when a series is above below the regression. A significant divergence from the regression line indicates at a minimum a warning. Here’s a sample of one of Ned’s charts, from his book:
John Bollinger – inventor of Bollinger bands – provided a historical perspective on bands in timeseries charts going back to the 1800’s. John made a special point that these financial calculations are no longer technically difficult to implement – he provided Python code for each historic example, and commented on ChatGPT’s ability to generate code. Here’s one of the charts, with Python code. More importantly, John stressed how these technical analysis studies are used, for example, a trivial understanding of a Bollinger band is to derive a signal when the raw data (price), crosses the band. John explained that a signal on a Bollinger band is created when the trend weakens, the price crosses to the opposite side of the band, touches the band, then fails to break back across the moving average line (I’m paraphrasing, my notes might not be exactly right on that).
Trading
Technical analysis can be used to understand markets, and also to underpin trading strategies used to make money in the markets. To use technical approaches to trade markets requires backtesting the technical analysis model against historic data to validate the model’s effectiveness. In addition to making a profit and other metrics, one key metric is max drawdown, an indication of how much money a model might lose at any time. In at least two presentations I saw models that had drawdowns of greater than 50% — which will require a significant commitment on the part of the trader to believe in the model and have confidence that the original model was not overfit to the test data.
There were a couple of great interviews interviews, including Jerry Parker. Parker’s story goes beyond backtesting to one of the gutsiest validation tests ever done. Parker was one of 20 people without a trading background, trained in a trading system, given a $1million to trade, and succeeded. It’s an incredible story, here’s a basic outline.