Curve AdvisorKeymasterJanuary 16, 2015 at 11:56 amPost count: 612
If you subscribe to the newsletter and see me discuss the yield curve, I will occasionally say that some thing is “rich or cheap”. This is almost always a relative description. That is, it is possible the entire curve is “rich” (rates too low). However, when I say something on that curve is “cheap” (rates too high), I am saying this in relation to the rest of the curve. It’s very possible that a contract could be “cheap” relatively (compared to other contracts on the curve), but “rich” absolutely (if the rest of the curve is richer).
So here are the various ways you can look at to determine rich/cheap:
* rich/cheap to historicals – once you’ve identified the relevant historical comparison period, you can look at the current levels compared to the period’s high and low, as a percentile, as a regression residual, as a standard deviation, etc. You not only need to look at the statistics for the structure, but for the nearby structures and other related structures.
* rich/cheap to the current curve – how is the level of the structure, compared to other nearby structures and other related structures
* rich/cheap to a future view – in your rate view, what do you expect to happen to ALL parts of the curve?
Generally, the best trades tend to have all three of the above factors, although it is possible to have a great trade with just one or two.
[to be continued]
Curve AdvisorKeymasterJanuary 22, 2015 at 5:26 pmPost count: 612
RICH/CHEAP TO HISTORICALS:
* Make sure you are looking at a “relevant” historical comparison period. For example, if the central bank just switched from a hiking bias to an easing bias, the historicals will not be useful.
* Typical comparison periods are: 3 months, 6 months, 1 year, 5, years, etc. A reasonable comparison period may vary depending on the type of structure (some direction-neutral structures further out the curve are better candidates for longer-term analysis).
* The basic snapshot is just how the structure has traded over time (a “standart” price chart). You can look at highs/lows, support/resistance, trend, percentile observations and basically any traditional “technical” measure. This type of analysis is useful if the later points are more important.
* You can also do an analysis of the price of the structure vs the level of rates. This type of analysis is useful for understanding the relationship of the structure to the level of rates (bullish, bearish, etc). You may even be able to see some nonlinearity in certain structures. Typical analyses you can do are things like regression, and various measures of rich/cheap can be calculated (like standard deviation, residuals, etc).
* Finally, you can do an analysis of the price of the structure vs the price of other structures. The possible level of a butterfly is a function of the slope of the curve, or other butterflies on the curve. You can also look at other comparable structures on other curves (whether it be in a different currency, or different securities).
Curve AdvisorKeymasterJanuary 23, 2015 at 4:35 pmPost count: 612
RICH/CHEAP TO THE CURRENT CURVE:
* The simplest thing to do is to plot a 6mo fly chart (or 3mo, 9mo or 12mo), and ask yourself if it passes the “eye test.” That is, does the curve look “smooth”? From a theoretical standpoint, the curve should be somewhat smooth, if we are to take a probabilistic interpretation of the fly curve. But this also makes sense logically, as any large steps up in yield at a particular point on the curve would provide slightly outsized returns to investors. Any fly that is “too high” on the fly curve could be thought of as being rich (as a fly) and any fly too low could be considered “cheap.” Sometimes, you can get a bad close mat makes a fly look too high or low. HOWEVER, THERE ARE CAVEATS AND EXCEPTIONS TO THE EYE TEST, INCLUDING:
– What central bank moves are expected? Large positioning of a particular hiking (or easing) view can distort a part of the curve.
– Are there (calendar) constraints? There may be certain things which would make certain meetings more likely to hike (or ease) than others. One such example would have been the FOMC’s time-based “mid-2013 guarantee.” There may be a high probability of a budget impasse, an important meeting/deadline, etc.
– Are there any demand/supply issues (in a particular section of the curve)? A big example would be QE targeting a particular part of the curve.
* You should also look at a particular fly structure relative to other curve structures. In particular:
– A fly can be rich overall, but it may be cheap when compared to even richer flies on the curve.
– A equally-weighted fly is a function of the slope of the curve. A fly can be rich, but it may be cheap when compared to the slope of the curve.
– A fly can be a function of the level of rates. A fly can be rich, but it may be cheap when compared to the level of rates.
Curve AdvisorKeymasterJanuary 26, 2015 at 4:31 pmPost count: 612
RICH/CHEAP TO A FUTURE VIEW:
* You can translate most market views into: (1) directional view, (2) slope view, and (3) curvature view.
* Consider what will happen to the yield curve (in terms of direction, slope and curvature) in the various market scenarios. When considering scenarios (ie hawkish, dovish, neutral), also consider what are the most likely causes of a move – a bullish move because of FOMC dovishness is going to result in a different curve than a bullish move because of global crisis.
* When you analyze the various curve outcomes, you may find that certain structures may outperform (or not underperform) in most of the scenarios. Those are the structures that provide “value.” For example, it may be that certain butterfly structures are so low that on a rally, they won’t get worse, but on a selloff they will get better. Conversely, some trade may be so oversubscribed that even if the view materializes, there is no more profit to be had from such a position.
* If you have a (strong) view on the markets/curve, use that view for your probability weightings for the various scenarios.
Thinking about various scenarios and the effect on curvature structures takes some experience, but you will start getting a good feel for it, as you watch how various contracts trade.
Curve AdvisorKeymasterJanuary 28, 2015 at 5:32 pmPost count: 612
If you have convenient (and preferably free) access to recent historical data, a good project to give yourself to get a sense of ranges is to do the following:
* Create the “Basic Analytics” spreadsheet
* Graph the 6 month flies. The best option would be to use what I call a “size adjusted price” (to be explained later in the “Basic Analytics” spreadsheet). For now, you can use “last price” of the individual contracts, 6 mo spreads, or 6 mo flies. Turn adjusted would probably be better (or even better would be if you put a toggle to switch turn adjustment on or off).
* Take the historical data and also graph the 1 month low and high, and the 3 month low and high. Use the actual contracts (ie “EDH5”) rather than the generic contracts (ie “ED1”). You do not need to interpolate, as we will be looking at a short time horizon.
* You can get a good feel for if something is really low or not historically. And really, the past 1-3 months is probably most important.
* You can also get a feel for how a current point looks relative to the points around it – giving you an idea of rolldown, how it rolls backwards, excess kinking, etc.
* If the graph is not clear, you can create a table instead and calculate percentiles.
* You can add functionality to the graph (or table) as you see fit.
For people who need free access to historical data, some time early next month, I was going to look into Qandl.com (free data – they have a download tool), and was going to do a review for the Forum.
Curve AdvisorKeymasterJanuary 31, 2015 at 10:37 amPost count: 612
Keep in mind when you do the project in post 1002 (above), that the “historicals” for flies in the white pack will be higher than usual, because it is not constant maturity (and this is the part of the curve with the greatest rolldown). It does not affect the back months as much because that part of the curve is more stable (and 3 months of rolldown is not as material on a fly structure). So that’s why I also wanted you to look at a 1 month range (the shorter the range, the lower the distortion from not using constant maturity).
Because the 3 month data is not constant maturity, you may be comparing the current point (say flies around H7), with a point that is the equivalent of M7 now (which the H7 point 3 months ago would be). So when you look at the unadjusted historical range for a fly around H7, also look at what the unadjusted range for the same fly structure around Z6. The combination of looking at the flies around both Z6 and H7 will give you a more complete range of “possibilities,” and also give you an idea of how you can expect the structure to roll down the curve in the short term.
Also keep in mind that 1-3 month highs and lows get broken all the time. So this should be most a “guide,” and not any kind of “hard” range.
Curve AdvisorKeymasterFebruary 3, 2015 at 12:19 pmPost count: 612
Here is just one example of the project I was describing in Post #1002, and the goal is to give you a quick pictorial representation of the 6 month butterflies. You can also add the table (standard feature in Excel), if you want to see the actual numbers (note that the numbers/graphs are turn-adjusted). The orange shows the 1 month range, the yellow shows the 3 month range, and today’s data is in black. So when you look at this, you get a snapshot right away of what the 6 mo fly curve looks like now, and what it has looked like in recent the past. You can also get an idea as to what part of the curve is depressed – both based on other parts of the black line, as well as based on the historicals. If you like “flipping,” this will also give you an idea as to what a reasonable exit level would be (typically when we get around half-way, unless you had some other reason to hold longer or shorter). You can also construct something similar in other structures (year flies, double flies, etc), but this is good for new users to get an idea of what the ranges are/were.
Here are some of the things you can glean from this chart:
* you can see from comparing the 1mo high and the 3 month high in the flies centered around M8 that that has come down a lot (because of the curve flattening).
* since the Dec FOMC meeting (which was over 1 month ago, so the 1 month data is all post-meeting), the flies centered around M6 have found renewed support (since the FOMC is hawkish).
* You can also get a sense as to whether certain “kinks” in the current curve are just temporary, or chronic. The small downward kink around U8 appears to be temporary (as it does not show up in the highs/lows). The upward kink around U6 appears to be chronic (as there are 98 days, instead of 91, in the M6-U6 spread). This could also tell you when you are chronically wrong on a turn adjustment.
* as you would expect, there is more volatility in the flies around the reds and greens. So this could give you an indication of sizing
So as you look at this, think about what has happened in the various time periods to cause the curve to move the way it has, and try and explain why the current fly curve is where it is, and why the ranges have changed to where it is.
Attachments:You must be logged in to view attached files.
You must be logged in to reply to this topic.