Several readers have requested that we evaluate the market timing value of Tony Caldaro’s “Objective Elliott Wave (OEW)” analysis. Mr. Caldaro describes OEW as incorporating missing tenets, such that: “Applying these newfound tenets to the market, …the waves were crystal clear. The turning points were precise, to the day, because they were quantitatively derived. There was no question when a wave ended and another began.” Based on the record of “MEDIUM TERM” outlook synopses in Tony Caldaro’s daily commentary blog and contemporaneous daily data for the S&P 500 index over the period January 2006 through January 2008 (523 trading days), we conclude that:
Our first step is to translate the daily medium-term outlooks into quantitative investment postures, via the following rules and assumptions:
- There are often multiple posts and outlooks for each market day. We use the earliest outlook for each day.
- When an outlook implies a change in investment posture, we assume action is taken at the prior close for the S&P 500 index (presumably, an approximation of the outlook could have been known at the prior close). This assumption may introduce a bias in favor of the OEW method for this review.
- When a commentary explicitly provides different outlooks for different indexes, we focus on the outlook for the S&P 500 index (SPX).
- We assume the following correspondence between market outlooks and investment postures:
- Bullish/uptrend = 100% Long
- Bullish/uptrend with reservations = 50% Long
- Neutral/cautious/mixed signals = 100% Cash
- Bearish/correction = 100% Short
- Bearish/correction with reservations = 50% Short
- We assume the return on cash while out of the market exactly offsets all trading costs/frictions.
- For two trading days on which there is no commentary, we assume the investment posture is the same as the preceding trading day.
Some outlooks do not neatly fit within the five types above, necessitating ad hoc judgments.
The following chart, based on the above assumptions, overlays investment postures derived from OEW medium-term outlooks (see the table in the comments appended below) on a plot of the S&P 500 index over the entire sample period. There are 72 changes in OEW-derived investment posture, an average of about one every seven trading days. Visual inspection indicates no obvious relationship between OEW-derived investment posture and future returns. The outlooks are clearly bullish, neutral and bearish 44%, 28% and 12% of the time, respectively, during the sample period.
For more sensitive testing, we relate OEW-derived investment postures to future S&P 500 index returns.
The next chart is a modified scatter plot that relates current OEW-derived investment posture to S&P 500 index returns over the next seven trading days (since the information content of the medium-term OEW outlook changes on average every seven trading days). To avoid test interval overlap, we use data from every seventh trading day, yielding a total of 74 observations. The chart orders this winnowed sample from most pessimistic OEW-derived investment posture (-100%, or 100% Short)) to most optimistic OEW-derived investment posture (+100%, or 100% long). The sample points on the horizontal axis are therefore not in time sequence.
The Pearson correlation between the two series is 0.00, and the R-squared statistic is 0.00. These results indicate no relationship between OEW-derived investment postures and S&P 500 index returns over the next seven trading days. By contrast, the Pearson correlation between current OEW-derived investment postures and S&P 500 index returns over the previous seven trading days is 0.31. Evidence suggests that OEW outlooks may be somewhat reactive but not predictive.
The following table presents an alternative way to analyze this same winnowed dataset by grouping data according to current OEW investment posture. It shows no systematic relationship between OEW-derived investment postures and S&P 500 index returns over the next seven trading days. It does tend to confirm the impression from the preceding chart that pessimistic OEW investment postures may to some degree predict relatively high S&P 500 index volatility over the next seven trading days. However, most of the subsamples are very small, so picking a different start date for sample winnowing could result in substantially different results.
What if seven trading days is too short a period for OEW medium-term outlooks?
The next table summarizes S&P 500 index returns over the next 21 trading days according to OEW-derived investment posture. To avoid test interval overlap, we use data from every 21st trading day, yielding a total of just 24 observations. Results indicate no systematic relationship between OEW-derived investment postures and broad stock market returns over the next 21 trading days. However, overall sample and subsample sizes are so small that picking a different start date for sample winnowing could produce dramatically different results. A much longer sample period is needed for this test.
The last chart summarizes a final test based on the performance of an initial $100,000 portfolio traded according to daily OEW-derived investment posture as defined above (72 changes). The chart shows for comparison the performance of a $100,000 buy-and-hold investment in the S&P 500 index. The OEW portfolio tends to underperform (outperform) buy-and-hold when the market is rising (falling), as would a portfolio that is non-predictively long, short or in cash.
In summary, evidence does not support a belief that the “Objective Elliot Wave” method successfully times the broad U.S. stock market. The method appears more to react to past returns than to predict future returns.
Tony Calardo sent the following comments on the above review:
Comment 1: “If you had contacted me before the analysis I would have saved you a lot of time and effort. OEW is all about trends, not short term trading. Every trend is a significant wave. During the period you examined: 2006 – 2008, there were only nine trend changes. Therefore only nine waves. If I was promoting trading based on these waves there would have been nine trades. Not the 72 short term trades that you somehow arrived at based upon your analysis. The blog is not for short term traders. This is mentioned repeatedly throughout the years.”
Response 1: The above analysis relies exclusively on the daily “MEDIUM TERM” synopsis from daily OEW commentaries, under the assumption that OEW blog readers may focus on this systematic and prominent outlook for guidance. The following table provides the underlying detailed outlook data (originally not included with the analysis because of its length). The first column lists the dates of the outlook, using (as noted above) the earliest outlook provided for each calendar day. The second column reproduces the 521 medium-term outlooks from the OEW commentaries over the sample period. The third column assigns hypothesized investment postures, how readers might react to the outlooks (focusing on SPX where broken out separately), as defined in the analysis. The intent of the hypothesized investment posture is to benefit from from all shifts in the OEW medium-term outlook. The approach does not impose any trading frequency. Rather, it lets the variation in the medium-term outlooks guide the frequency of action. It may be possible to interpret the daily OEW medium-term outlooks with alternative investment posture assumptions such that they indicate only nine trades. It is arguable that such assumptions would ignore some of the information in the daily medium-term outlooks. Readers can scan the table and decide how they might react.
However, to ensure that data drives the analysis, we have adjusted the original test of the relationship between the OEW-derived investment posture and the future return on the S&P 500 index such that the future return interval matches exactly the average frequency of changes in the OEW medium-term outlook (seven trading days). We have also added an analysis that relates OEW-derived investment posture to the future performance of the S&P 500 index over a longer period (21 trading days).
Comment 2: “I’ve attached a chart of my mid-2006 forecast for the SPX. As you can observe the market unfolded within 2% of the forecast, while it was rising 25%.”
Response 2: In general, CXOadvisory.com reviews of forecasting methods and outputs do not rely on data specified by the forecasters. Readers can investigate further as they desire.
Comment 3: “Your conclusion was correct based the data you used. The commentary discusses what the market has done, within the waves, and not the waves themselves. The MEDIUM TERM trend for the SPX, if comparing to the SPX, alone should have been examined. Or, the LONG TERM trend. There were only nine medium term trend reversals during the entire two year period, and only one long term trend reversal.”
Response 3: The above analysis relies exclusively on the “MEDIUM TERM” outlook synopses for the SPX as recorded in the OEW blog. See Response 1 above.
