Stocks and safe haven assets are rallying together. Bearish?

by Troy

It’s been a quiet week in the stock market as the 4th of July approaches. Today’s headlines:

  1. VIX fell more than expected
  2. Stocks and safe haven assets rally together
  3. Breadth divergence
  4. Alligator jaws
  5. JPMorgan’s AI model

Go here to understand our fundamentals-driven long term outlook. For reference, here’s the random probability of the U.S. stock market going up on any given day.

VIX

VIX fell more than -8% while the S&P rallied less than 0.3%. This is uncommon, because VIX doesn’t usually fall this much on such a small stock market rally.

Similar historical cases were mostly random in the short-medium term, but bullish 1 year later.

We can look at volatility from another angle. VIX fell below 13 for the first time in more than 40 days. In the past, this was almost exclusively bullish for the S&P 1-2 months later.

However, it was almost exclusively bullish for VIX 3 months later. VIX tends to be lower-bound, which means that as the S&P keeps rallying, VIX doesn’t go much lower.

Stocks and safe havens

I’ve seen a lot of these headlines over the past few weeks, from Bloomberg:

Stocks and traditional safe haven assets (gold, bonds) have gone up together on a lot of days over the past few weeks. 6 of the past 18 days have seen the S&P 500, gold, and the 10 year Treasury bond go up together.

One would assume that this could be bearish for stocks. Afterall, a lot of people are buying safe haven assets in anticipation that “something” could go wrong in the future (trade, macroeconomic data, Fed, etc).

But that’s not what the data suggests. Similar historical cases have not been consistently bullish or bearish for the S&P. If anything, this is slightly bullish for stocks over the next 2-3 months.

Breadth divergence

Traders love breadth indicators. A popular way of using breadth indicators is to monitor “divergences”. For example, fewer and fewer of the S&P 500’s stocks are above their 50 dma while the stock market makes new highs. The narrative is that this is bearish for stocks.

The problem with divergences? They almost ALWAYS happen before a significant market top, but not every divergence leads to a market top. In other words, hard to use for market timing because there are too many false signals.

Here’s what happened to the S&P when it rallied more than 5% over the past 90 days to a 1 year high, while the % of stocks above their 50 dma fell more than -10% (DIVERGENCE!!!!)

A lot of these cases are overlaps, so let’s examine some of them individually.

Here’s September 2003:

Here’s December 2009:

Here’s February 2014:

So why do these divergences exist? The % of S&P stocks above their 50 dma EXACTLY MIRRORS the S&P’s distance from its 50 dma.

  1. The more stocks above their 50 dma, the more the S&P will be above its 50 dma.
  2. The more stocks are below their 50 dma, the more the S&P will be below its 50 dma.

It’s just that simple.

The only reason the S&P is making a “divergence” from the % of stocks above their 50 dma is because the S&P has become closer to its 50 dma.

Copper

Speaking over divergences, a lot of traders believe that divergences always end up with “alligator jaws snapping shut”. For example

HOLY SHIT THE S&P IS DIVERGING FROM COPPER! ALLIGATOR JAWS ALWAYS SNAP SHUT!

This is far from true. Divergences between markets occur all the time. When correlation is not causation, correlation eventually breaks down. Alligators aren’t going to come and bite the legs off of bulls. (I honestly don’t know how people come up with these phrases.)

JPMorgan’s AI model

And lastly, Bloomberg posted an interesting article about JPMorgan’s AI model.

JPMorgan’s AI model is bearish for this month. Maybe they’re right. Maybe they’re wrong. I don’t know. Predicting the short term is very hard.

But generally, I wouldn’t use models that use 10-20 years of historical data. These models tend to be flimsy, because the next 10-20 years will probably be different from the previous 10-20 years. (Remember how different 1980-2000 is from 2000-2020).

A lot of financial firms create indicators and models as marketing tools (even better if they include buzzwords like “AI”). But just think about the simple logic. If these models were so good, why doesn’t JPMorgan just trade most of its own capital on them? My guess – they’re skeptical too.

We don’t use our discretionary outlook for trading. We use our quantitative trading models because they are end-to-end systems that tell you how to trade ALL THE TIME, even when our discretionary outlook is mixed. Members can see our model’s latest trades here updated in real-time.

Conclusion

Here is our discretionary market outlook:

  1. Long term: risk:reward is not bullish. In a most optimistic scenario, the bull market probably has 1 year left.
  2. Medium term (next 6-9 months): most market studies are bullish.
  3. Short term (next 1-3 months) market studies are mixed.
  4. We focus on the medium-long term.

Goldman Sachs’ Bull/Bear Indicator demonstrates that risk:reward favors long term bears.