This Is What A Contrarian Indicator Looks Like

One of the big themes here over the past three weeks has been the extent to which spec positioning as detailed in the weekly CFTC report (always out Friday afternoon and current through the previous Tuesday) can be used as a contrarian indicator.

For anyone who missed this discussion, here’s a quick summary:

What does a “contrarian indicator” look like?

Well, if recent history is any guide, it looks like “smart” money positioning.

Recall that in March, a massive net long in crude was hit with a reality check during CERAweek, catching a lot of folks offsides. And last Friday, when the latest CFTC data hit, we learned that specs had covered the entirety of their 10Y short on the way to building the biggest long position since 2008 on the heels of an extended rally in Treasurys.

Oh, and don’t forget what was perhaps the most hilarious example of extended positioning gone wrong: the Russell 2000 short which got blindsided by the risk-on sentiment that accompanied the market-friendly outcome of the first round of the French elections.

Another interesting aside with regard to the above is the extent to which investors are paying closer and closer attention to “smart” money positioning in the commodities space. Recall this chart we highlighted on Wednesday:



Well, underscoring the notion that a whole lot of folks are now wondering if they should trade in the opposite direction of the ostensibly “smart” money, BofAML is out on Thursday with a look at how things usually turn out for Treasury yields when positioning gets extended.

You’ll note that, as mentioned in the excerpted passages above, this is a particularly important topic in light of last week’s CFTC report which showed that specs flipped from short to long in the 10Y and not only that, their position is now the most long since the crisis.


Find the latest on this from BofAML below and do note that the takeaway here is pretty goddamn simple:

When positioning gets extreme, the market frequently moves the other way.

Via BofAML


Positioning in rates has reached an extreme 10-year positioning, as indicated by net TY speculator positions, has reached a 9-year high, with this segment of the market now leaning more long in duration than at any time since March 2008. An empirical analysis of positioning data indicates that historically when positioning becomes this long, rates have risen 70% of the time in the following month. Table 1 shows the results of our positioning study in which we tracked rate changes after positions were established. The main result is that spec positioning in TY can be a powerful contra-indicator. As an example, the most recent extreme reading was on 7 March, 2017 when the market became very short after several Fed speakers, including Chair Yellen, verbally guided the market into a March hike in the three days before the March FOMC blackout window. In the month following that extreme short positioning reading, 10y rates rallied 16bp as specs closed out almost the entirety of their shorts. 10y rates are lower by 30bp since their 13 March close. Positioning has now swung the other way, and we think this bodes well for bond bears over the next month.

What the data show us about extreme positioning

There are four main points to take away from Table 1 below:

  • When positioning gets extreme, the market frequently moves the other way. The two columns labeled “frequency of rally/selloff next month” show that rally and selloff frequencies are 50/50 over the whole period (not conditioning on positioning, top row), but these frequencies become close to 60/40 in the case of extreme short positions (z < -1.25) or 70/30 in the case of extreme long positions (z > +1.25). The z-score of current positioning from this week is z=+2.2.
  • The contra rate moves in the following month occur as positions get covered. This is shown in the column labeled “change in position next month.” For example, when z>1.25, we find that positions typically cover the following month by a net of 40k contracts.
  • The average rate move over the next month is larger when positions are larger. The 3rd-to-last column “average rate move next month” shows that when ignoring position size (top row), the average rate move is close to 0. But when positioning is highly skewed, the average move can be 7-8bp in the contra direction.
  • The contra moves tend to be larger in magnitude. As shown in the last two columns, when positioning is short, the rallies tend to be considerably larger than the selloffs, and the same is true, to a smaller extent, when positioning is long.

Rather than viewing extreme positioning as a predictor of rate moves, we prefer to view it as creating better risk/reward opportunities to trade duration based on macro or technical views. The asymmetry of rate moves following large position buildups is intuitive: profit-taking dampens moves in the “right” direction while position-covering accelerates moves in the “contra” direction. We think the current level of extreme positioning provides an opportunity to express a tactical bearish view, taking advantage of what history indicates is traditionally an outcome with an asymmetric distribution and a favorable mean. Two potential near-term catalysts for a move to higher rates include a possible upside surprise in Friday’s payroll report, or a positive surprise out of Washington on health care legislation that could increase market confidence around future tax policy.




2 thoughts on “This Is What A Contrarian Indicator Looks Like

  1. Not sure if you take (stupid) questions…but how do CFTC futures contracts on the TY correlate (or not) to positions in TLT?

Speak On It