One of the more interesting things from last week was that FFV-X traded up to 2bps.  For most of the year, we said it was highly unlikely for the Fed to hike less than a week before the election, and it has been close to 0.5bps for several months.  However, the “lottery ticket” effect that we have discussed for some time now has appeared in full force.  As the Fed meetings get closer, people feel the need to take a 10+:1 shot, get protection for their books, close out shorts in the spread, etc.  And with the current Fed, people may also want to play for a “language change” at the first meeting (Sept) for a move in a later meeting (Nov).  This is why if you have the excess capital and have ease protection, I like owning these non-quarterly meetings.  This spread has been between 0 and 1 for over a month, and finally broke out the past two days.  This brings up two interesting things I want to discuss:

  • 160821 FFEWhy the non-quarterlies can be “zero”[1] in the first place. A few weeks ago, most of the non-quarterlies were priced at “zero.”  Fed Funds futures aren’t incredibly deep (especially relative to Eurodollars).  The open interest on FFM7 (less than a year away) is just over 9K contracts – LOL.  The open interest on FFN7 is 21.7K contracts.  This is still not a lot, but relatively more.  The participants in the markets use FFs to target certain meetings.  FFN7 makes a lot of sense as a “June 2017” meeting play, as it captures 100% of the June meeting (and the prior meetings).  Most people want to take positions on quarterly meetings.  For example, when was the last time you heard an economist call for a Jan, April, July or Nov hike?[2]  It’s rare.  For the same reasons, most traders think about the quarterly meetings (and ignore the non-quarterlies) – they do have the Summary of Economic Projections and a press conference, after all.  If you were to take a view on a quarterly meeting, the best way to position is to buy: FFG-J spread for March, FFK-N for June, FFQ-V for September, and FFX-F for December.  The tendency is for the markets to buy {FFG, FFK, FFQ, FFX} and sell {FFJ, FFN, FFV, FFF}.  If you take a look at the pattern, you can see that because of this tendency, the non-quarterly meetings {FFF-G, FFJ-K, FFN-Q, FFV-X} will have a downward pricing bias, below “fair value.”  When the markets want to exit out of their quarterly meeting position, the pricing for the non-quarterlies could pop.  But the non-quarterlies will also pop as the meetings get closer, as the near-term lottery value materializes.  There is value in getting paid a little to provide some market liquidity in thin markets.  You can also get an occasional off-market fill in volatile markets.
  • The markets don’t seem to be fully factoring in the number of month-ends for further out FFs. On the right is the chart of the FF effective rate.  The main thing to note is the month-end day (the huge spike down in the blue line).  It’s not like the old days where you had to factor in things like weekends, holidays, maintenance periods, and whatever.  The rate is basically the same every day, and you need to factor in about 8-10bps lower for month-end (you can adjust for quarter-end and year-end if you want).  All you need to do is count the number of month-end days in each contract.  Each FF contract will have between 1 and 5 month-end days.[3]  So sometimes, buying a non-quarterly FF spread at 0.5 could be like buying at “zero”, if you are getting 2 additional month-end days in the contract you are buying.

 

In retrospect, not buying the snot out of FFV-X @ 0.5 after Dudley’s comments was a mistake.  FFV has 3 month-end days and FFX only has 1.  So that is probably over half a bp in that spread right there.  And we should have known people would be willing to pay up to 1.5 bps for the lottery/insurance effect.  This isn’t going to make your year, but like I always say, the little bits add up.  Let’s be more alert for the next such opportunity.

[1] Sometimes, a FF spread could be 0.5 or even above 1, and still be “zero” when you factor in the month-end day count.

[2] Note that some of the non-quarterly meeting months have changed for 2017 to: Feb, May, July and Nov.

[3] Don’t forget the holidays – in particular New Year’s Day, Memorial Day and Labor Day, which can all affect month-ends.