This week we received an email from an advisor that echoes some calls we’ve been receiving lately. We thought it would make a great topic for us to cover in our commentary this week.
The email read:
We’re seeing a lot of negative indicators in the market right now, and seeing commentary from other managers changing from “pessimistic” to “dire warnings of impending doom”. We greatly appreciate the systematic approach you take and would value any thoughts you might have on current risk levels, beyond your blog post from two weeks ago. I’m a little worried that the combination of the index tranching […] will result in insufficient action in the event of a significant correction- could you speak to that?
Also, do you have any insight on the statistical value of Hussman’s most recent analysis? http://www.hussmanfunds.com/wmc/wmc150727.htm I always find it suspect when someone manages to produce analysis that perfectly predicts the worst disasters with no false positives…”
On Rebalancing & Tranching
When we designed our rebalancing approach, the first question we needed to answer was “how often do we rebalance.”
Some tactical strategists rebalance as frequently as daily, some weekly, and others monthly. We firmly believe in the consistency, lower trading, and lower turnover afforded to us by a monthly rebalance.
Well if we look at the last 20 or so years, we really have gone through 5 market regimes, three bull markets only interrupted by the tech bubble bursting and the global financial crisis.
So if we had perfect hindsight tactically, we would have made four trades over the last 20 years (get out before the tech bubble, get back in at the bottom, get out right before the financial crisis started in 2007, and get back in in March 2009).
However, any tactical model that hopes to protect on the downside - no matter how consistent it is - is going to generate significantly more than four signal changes. This means in turn that it is far more likely that a signal change is a head fake as opposed to a correct signal. The more frequently we allow a rebalance, the more of these “bad” signals we will trade on - at a significant cost to the portfolio.
We believe in monthly rebalancing precisely because we believe it is the best balance of staying flexible while filtering out these unnecessary signal changes.
The next question we get sometimes is, "well won’t a monthly rebalance lead to less downside protection because it is slower to get out?” My response:
- We are trying to protect against large market dislocations. Large market dislocations tendto be multi-month or even multi-year events. Typically, waiting a few weeks here or there to get out when you are talking about a huge correction isn’t meaningful. It may feel like getting out of the market quickly is the right move at times, but that is our emotions talking, which is why we use a rules based approach. Our research and experience show that trades driven by human emotions tendto be the wrong moves for the long-term, which is why we rely on our rules and data driven signals – both of whichare constructed at a time when emotions are not high. Investors should keep in mind the goal of the investment strategy.If the strategy’s goal is to dually “protect and participate,” then we believe that our approach is optimal to both protect from significant downturns and participate meaningfully in bull markets.
If the goal is simply to “protect,” then there are plenty of investment strategies offered by other firms that will protect you in bear markets by selling quickly when signs of weakness appear, but they likely will not participate in markets such as the bull market from 2009 to 2015.
For example, say you called the top of the tech bubble perfectly in August 2000, but waited to sell one month until September 2000. You would have only lost about 5% compared to 45% for the S&P 500. Similarly, say you called the top of the financial crisis in October 2007, but waited to sell until November 2007. In this case, you would have lost a bit less than 4% compared to almost 50% for the market.
By being a bit more patient, we gain far more by avoiding incorrect signals with a monthly rebalance than we lose by being a few weeks late de-risking.
- People often only talk aboutwhipsaws in up markets. Butwhipsaws can just as easily happen in times of market crisis.With a full weekly rebalance, you may get fully back into the market due to a rapid recovery only to see the market resume its crash.
Just as the monthly rebalance can protect you against whipsaws in up markets, it can also do so in down markets - keeping you consistently out of equities even in the face of short-lived pops that soon reverse. The perfect example of this type of environment was late November 2008 to early January 2009, when the market recovered by almost 25%, only to lose close to 30% more.
The data backs up that a monthly rebalance is optimal. The chart below shows the risk-adjusted performance of a simple momentum strategy with different rebalance frequency. The monthly rebalance has consistently offered higher returns and lower downside risk than a weekly or daily rebalance.
So we want to rebalance monthly, but before we do we just need to make a very simple decision – which day of the month do we rebalance?
Should we rebalance on the first trading day, the last trading day, or maybe the 2nd Thursday of the month just because we feel like it. Such a trivial decision won’t matter anyways… right? Unfortunately, the data bears out that this assumption is wrong.
Momentum investors that use a pure monthly rebalance can either get very lucky or very unlucky depending on which day of the month they rebalance. We call this timing risk, and we think this is very real.
Unfortunately, this timing risk performance difference is entirely random – it is impossible to know ahead of time.
As risk managers, we are ok with conceding that we won’t get lucky in return for eliminating the possibility of getting very unlucky. That’s all that tranching is, a way to follow a monthly rebalance schedule while controlling for timing risk. So tranching will have no difference in downside protection than a pure rebalance on average.
In summary, we believe strongly in a monthly rebalance. We believe that there is significantly qualitative and quantitative data to back up this decision. Tranching is simply a way to implement a monthly rebalance in the most efficient way possible.
The Current Market Outlook
As a firm, we do not make economic forecasts or predictions. We simply rely on what our models are telling us in the current market environment.
At a high level, we think that a lot of the noise in the market is more about driving eyeballs to websites/TVs and assets into defensive strategies.
At the time of writing, SPY is in only a 1% drawdown from all-time highs.
Since March-09, SPY has been in a drawdown greater than 1% almost 60% of the time. In other words, this is perfectly healthy behavior in the midst of a huge bull market.
So there is nothing that has happened that is causing our models to hit eject on equities quite yet. However, we are definitely seeing waning momentum.
For example, in our Risk Managed U.S. Sectors portfolio, four of the nine sectors are “off”. This week industrials joined materials, energy, and utilities in the negative category.
Of the five remaining “on” sectors, financials and technology are pretty close to turning off. We would expect that if either lose more than 3% or so, those signals would switch “off.”
The other three sectors (staples, health care, and discretionary) would probably need to see about a 9-10% pullback to turn off given their strong momentum profile.
Putting this all together, we would expect us to see some cash introduced with a ~3% loss and a move to all sectors “off” with a ~10% loss. This is right where the models need to be to keep drawdowns limited to the historical range we have seen since Newfound launched.
In other words, we feel the model is very well positioned to quickly react to a more significant downturn if one does materializes.
Personally, we think that Hussman’s analysis is a great example of data mining.
It seems he had a process that he used that has clearly not worked (down -28% since the end of the crisis). He then changed his rules so it would have worked with hindsight.
We get particularly concerned when we see “magic numbers” in analysis like a “CAPE of 18,” “60% stocks above moving average,” “S&P high in prior 8 weeks.”
We also get concerned when we see analysis that is always right because then you know it is unrealistic and you have no idea where/how big the risks are of relying on it.
We prize consistency above all else. We would rather have a model that has worked most of the time in real time than one that has worked all of the time in a backtest. We have seen our model react very well to unprecedented market movements over the last 7 years or so.
This is crucial because one forward looking forecast we will make with confidence is that the next crisis won’t look like the last.