Predictions or Predilections
Every year we see a slew of economic predictions from investors, economists, traders and everybody else in between. There is usually a very rational basis for these predictions, based on macro-economic and fundamental analysis.
But the problem with making predictions is that they tend to be guided by human predilections based on emotions rather than a systematic process that adapts quickly to changing information.
At the start of 2012 the top predictions across the big media outlets were:
1. Hard Landing in China
A decelerating Chinese economy had several notable investors calling for a hard landing in China, including hedge-fund manager Jim Chanos and Marc Faber, author of the Gloom, Boom and Doom Report.
- DID NOT HAPPEN: That did not transpire as the manufacturing, export and housing numbers have picked up and the Chinese growth is forecasted to grow at over 8%. The Shanghai stock index instead of collapsing finished 2012 flat.
2. "Grexit" and a euro Collapse
May saw the anti-bailout party make big gains that led to strong speculation that Greece's exit from the eurozone was imminent. The widely cited forecast of the indebted nation's exit was even assigned a nickname: "Grexit" (a term coined by two Citi analysts). Prominent figures in the investment community, including Pimco's CEO Mohammed El-Erian, were looking for a Greek exit from the eurozone.
- DID NOT HAPPEN: However, the outcome of Greece's second election on June 17 followed by massive eurozone intervention supported by the IMF not just staved off a Grexit scenario, but also supported the euro, which has stabilized and trading at a 30% premium to the U.S. dollar.
3. Gold Rush
Every gold bug was predicting gold to hit $2,000 per ounce level by the end of 2012, but the highest we reached was $1,800.
Proponents of the precious metal based their calls on possibility of rising inflation stemming from massive quantitative easing by major central banks and increased demand from India and China.
- DID NOT HAPPEN: None of these scenarios played out and gold prices were pretty much unchanged for 2012.
Our 2013 Predictions
We take a different approach to the markets and let our models do the predicting.
Bottom line we are in a risk-on and risk-off environment. So, our goal is to catch the next risk-on wave and make sure we are invested during that period in high alpha equities and conversely in the risk-off phase, we want to be out of equities and into bonds.
Secondly, these waves change multiple times a year, so our models will make several predictions a year and not just once at the beginning of the year.
As of the start of 2013, our model is in a risk-on phase, and we have our clients invested primarily in European stocks through FEZ (SPDR EURO STOXX 50 ETF), commodities through DBC (PowerShares DB Commodity Index Tracking), real estate through VNQ (Vanguard REIT Index ETF and frontier markets of Africa through a mutual fund.
While our models look at many variables like moving averages, speed of movement, volume, etc. to determine the onset of the next wave, today I will give you a glimpse into one of the variables we use called, "price slope".
Price slope tells us whether we are approaching a risk-on or risk-off phase by measuring the slope of the latest price action of the S&P500. In a risk-on state, investors move into risky assets like equities. In a risk-off state, the opposite happens, investors move into less risky assets like bonds and cash.
Here are the 3 steps in using this indicator:
Step1: Determine the slope of the price action
Calculate the slope by picking a range of the closing prices and the dates. This can be easily done using the MS-EXCEL function, Slope.
Step 2: Determine a tolerance level for the slope
Next step is to determine a certain tolerance level. The tolerance level is important because it helps increase the accuracy of the predictions. If the slope is above that tolerance level, you make the assumption that the market is in a risk-on phase and vice-versa.
In this example, I have set that tolerance level to be 0%. So if the slope is greater than +0% the assumption is that we are in a risk-on phase. If the slope is less than 0% then the assumption is that we are in a risk-off phase.
Step 3: Test the Model
The next step is to see how this model would have done using real data on the S&P 500 going back 20 years.
The following are the results of running this simple one variable model and comparing the results to the S&P 500 over the last 20 years, from 1993-2012. In this test sample, we bought the S&P 500 during the risk-on phase and went to cash in the risk-off phase.
By just avoiding the risk-off phase as determined by the simple model shown above, we could have produced better results, with lower volatility and much lower losses. And that's using just ONE indicator.
The following chart shows the risk-on periods (dark bars) when the model would have us invested in the S&P500 and the risk-off periods (white bars) when the model would have us in cash.
Even this simple one variable model which looks at the slope of the price action has some challenges. The challenges are how many days to use in the calculation of the slope and what tolerance levels to use and most importantly, how to ensure that those parameters will work in the future as well.
To overcome these challenges, we use a multi-variable model which looks at not one, but several variables, and adjusts the parameters dynamically to ensure that the model adapts to the changing times.
It has worked well for us for several years and I will hang my hat on its predictions rather than be swayed by my feelings or predilections on any given day.