Curve AdvisorKeymasterDecember 18, 2014 at 12:23 pmPost count: 612
The first order of business should be for you to set up a spreadsheet that is going to help you look at the Eurodollar curve in a more meaningful way than just looking at the prices and changes on the screen. I generally do not believe you should ever use someone else’s spreadsheet. It’s the old “give a man a fish vs teach a man to fish” philosophy. So instead, below are some of the basic things you need to do to set up a reasonable spreadsheet. You should have a minimum of the following columns:
1. List of contracts by row. Start with EDH5 (which in this case is ED1) and go all the way down to EDZ0 (ED24). Most people generally look at the first 20 contracts, but there are times when a fly structure may extend a little further or times when you may want to see how a similar structure looks a little further out the curve.
2. Yesterday’s settle price. For P&L purposes.
3. Today’s current price and settle price. You should have a “checkbox” of some other method of switching between the current prices and the current closes (for end-of-day P&L).
4. Some way of turn-adjusting the prices. For simplicity, you can use 0.7bps for the year-end turn. This means that for each contract that has the letter “Z”, you need to add 0.7bps to smooth out the curve. (see the “Turn Adjustment” thread for more information on the turn) THIS IS THE COLUMN OF DATA WE WILL BE USING MOST OFTEN TO LOOK AT CURVATURE.
5. SPREADS. Calculate the prices for the 3 month, 6 month and 1 year spreads, in three separate columns. For example, in the ED1 row, you would use “=ED1-ED2” to calculate the 3mo spread (use the data from column 4). You can multiply by 100, if you find that easier to read.
6. BUTTERFLIES. Calculate the prices for the 3 month, 6 month and 1 year butterflies, in three separate columns. For example, in the ED1 row, you would use “=ED1-2*ED2+ED3” to calculate the 3mo butterfly (use the data from column 4). You can multiply by 100, if you find that easier to read.
AND THAT’S IT! (for now)
I will discuss each of the columns further, and what types of things you could look for. I will also tell you want other kinds of columns you may want to look at.
Curve AdvisorKeymasterDecember 19, 2014 at 8:19 pmPost count: 612
Let me start with number 6. BUTTERFLIES. Let’s take a column of 6 month butterflies.
You can plot them and get a roughly smooth curve. In this environment (where there will be a series of hikes and hikes aren’t “imminent”), you can think of the flies as indicators of Fed landing and liftoff probabilities. So the lower the negative fly, the more likely that “center” (of the fly) will be near where hiking starts, and the higher a positive fly, the more likely that center is near where the hiking ends. Of course this is a gross simplification, but that is one interpretation of curvature.
You can rationalize this by saying the following: Say for example, everyone “knew” the Fed was going to hike at the Jan 2016 meeting and end hiking in the December 2017 meeting (and say it was known that they will go 25bps per meeting). Everyone is going to want to buy ED spreads that encompass those meetings (roughly EDZ5-Z7 spread – I will discuss FOMC meeting assignment to EDs in another thread, as there are a lot of half meetings in 3 month ED contracts). So all 6 month spreads not including those meetings will be be zero (ie H5-M5 spread). Butterflies which are before those meetings will be zero (ie H5-M5-U5 fly), as will butterflies after those meetings (ie H8-M8-U8 fly).
Say you have a butterfly with legs a-b-c. As more of Jan 16 to Dec 17 is included in leg b-c, the more negative the butterfly will become. Eventually, you will get to the point where 100% of spread b-c includes hikes, while 0% of spread a-b includes hikes. That is where the lowest fly point will occur. Conversely, the highest fly point will be where 100% of spread a-b has hikes priced in, while 0% of spread b-c has hikes priced in.
However, we have no idea when the FOMC will start hiking, when they will stop, and in what increments they will hike. But if you picture the basic “complete knowledge” example in the previous paragraphs and assign a probability distribution to the various scenarios, you can start to see how we can get areas of negative and positive curvature.
DISCLAIMER: Of course, all of the above is still a simplification, because you have a lot of other factors involved, such as discounting for uncertainty, pricing of skews, convexity adjustment, and I am sure there are others I cam not thinking of right now.
But the basic takeaway is, the more negative a butterfly, the more hikes are priced into the back half of the fly (i.e. where hikes will probably begin), and the more positive a butterfly, the more hikes are priced into the front half of the fly (where hikes will end).
Here are some examples of how the above could help you with some trades:
* In this environment, say the EDH8-U8-H9 6mo fly was zero. That would be a screaming buy, because even if you thought hikes end in 2017, there is always a chance it leaks out into early 2018, in which case it is much more likely to leak into H8-U8 than U8-H9. The H8 6mo fly is not zero, but the H8 6mo double fly (buy H8-U8-H9 fly vs sell U8-H9-U0 fly) is, and that has similar properties.
* H5-H6-H7 fly is currently positive. If you thought that “patient” meant an average of 5 meetings (the minimum is 2, but say for whatever reason you suspect the average is 5), that basically means that they would not hike until September at the earliest. That could make the fly a good sell, because the H5-H6 spread would not have as much room to go higher (5.5 meetings) as the H6-H7 spread (8 meetings). I’m just throwing it out there as an very simplified example. However, as mentioned in the Disclaimer paragraph, there are other things you need to consider.
* Say for whatever reason you think the hikes end in Dec 17 (maybe you find out the FOMC dot plot medians were 100% accurate – LOL). Then buying the Z6-Z7-Z8 fly at the current market values makes sense, since that is definitely not what the market is pricing in with 100% certainty, and you expect Z6-Z7 to be 100% priced, while Z7-Z8 spread will be almost 0% priced.
As you look at more trade threads (I will add more going forward), think about why the fly structure is low or high, and see if it makes sense with the theory.
meParticipantDecember 20, 2014 at 4:09 amPost count: 27
There was an earlier thread which discussed data-set construction methdology of a timeseries of constant maturity curves. Here, in the basic analysis, it looks like we’re enumerating the primary spread combinations across the curve at a snapshot in time — either basis today’s price or yesterday’s price.
Should we be trying to construct one master workbook that does all of these from the outset? That is, you can feed a curve, based on constant maturity, for any given date into the spread/fly creator, for lack of better words? Or, is it typical to run separate workbooks with different cuts of the data. I guess the risk is that the adjustments/changes to one are not replicated in the other.
Or maybe you’re asking us to walk here before we run. 🙂
Curve AdvisorKeymasterDecember 22, 2014 at 7:53 amPost count: 612
You could combine into one workbook, but I find it easier to have as two separate workbooks, mostly because the historical sheet can get very large, depending on how many years of data and how much analytics you do on it. I don’t think you save much repetition by having it in one gigantic sheet. But it’s just personal preference.
The other reason I had it on separate sheets is that I used the “current curve” sheet (as described above in the first post), as part of my P&L sheet and trade blotter (list of all trades I did). This workbook was also very large for me. So my positions would feed from the trade blotter to the current curve sheet, and I could calculate my live P&L (and closing P&L) from this sheet. I could also assign a strategy number to the trade and get my daily, monthly and ytd P&L, as well as see the contract breakout by strategy. It’s just a bunch of sumif and sumproduct functions off of the trade blotter. Maybe after I finish talking about the basics, I can talk about keeping on top of your positions (since having a large book of flies can sometimes be confusing).
Curve AdvisorKeymasterDecember 23, 2014 at 8:20 pmPost count: 612
The most intuitive spread is probably the 1 year spread (ie EDH7-H8). In this environment, you can consider a year spread to be roughly equal to the amount the FOMC is expected to hike in a given “year.” Keep in mind this is only an approximation.
Spreads generally won’t benefit as much from nonlinearity, but looking at spreads may allow you to develop trade ideas. There are a number of things that could be useful when looking at spreads:
* does the magnitude and location of the highest year spread make sense? Keep in mind nearer term hikes will always be priced in more than later term hikes.
* does the magnitude and location of the ramp-up to or the dropoff from the largest year spread make sense?
* does the magnitude and location of the lowest year spread make sense?
* if any of the above do not seem “right” with respect to your view, you may want to consider a trade in: (1) curvature trades (ie butterflies) can be used to express some slope views (and may provide better risk-reward), (2) outright spread trades (buying spreads that look low and selling those that look high), or (3) options trades (low spreads may be attractive as conditional steepeners and high spreads may be attractive as conditional flatteners).
But looking at the curve in 1 year (or 6 month) slope increments may provide you with a better view of the curve.
meParticipantDecember 26, 2014 at 8:01 pmPost count: 27
Q: On the interpretation of the spreads.
When you say that in this environment that the slope represents the amount the FOMC will hike, is this for all parts of the curve? For example, say, U7-U8 (arbitrary) currently is 32bps as of close on 26. A year ago, we were around 90bps. Correct me if I’m wrong, but wasn’t the prospects of a rapid hike less likely a year ago than today?
Also, so if I just observe U7-U8 on a charting platform, effectively, since I’m not looking at constant maturities, the downward drift in the chart here is just effectively the roll-down on the yield curve, from blues, to greens. But I would need to look at constant maturities of the spreads in order to see the changes in expectations. Is that correct?
Curve AdvisorKeymasterDecember 27, 2014 at 8:14 amPost count: 612
This is why I always say things like “this is an approximation”, or simplification, etc. There are a lot of things that go into the shape of the curve. In particular, right now the long end of the curve is very rich. Some possible explanations are:
* longer term demographics support a richer longer end (i.e. aging baby boomers / pension funds should invest more in fixed income and annuities)
* FOMC owning 4 trillion of the longer end
* people who see value in the US as compared to lower European (and Japanese) rates
* low inflation and growth
* crisis buying
* foreign trade surplus buying
* people opting for flatteners as a bearish view, even though the level of rates is low.
Since something further out the curve takes much longer to converge to the FOMC rate path, you need to weight/discount whatever is priced in further out the curve. The overall level of rates is low. So you need to reconcile in your mind what distribution of rates makes sense… For example, does U5-U6, U6-U7, U7-U8, U8-U9 all make sense, with a ten year yield at 2.25? Or does the 2.25% not make sense?
As for what is priced into U7-U8, typically when hikes are far away, the spreads further out will be larger (ie. they could be 90). But as hikes get closer, the hikes that were priced in further out the curve get moved up. So typically you would see a spread like U7-U8 get much lower with looming hikes, because there may be nothing left to hike by the time U7 rolls around (ie the Fed could be done). However, if you are saying 32bps seems very low, you would probably be correct. But you need to consider how low the level of rates overall is. In previous hike cycles, we saw a lot of flattening further out the curve. In fact, there are may pieces out which state that historically, ten year yields don’t change much once the FOMC starts hiking. But that was because in the past the ten year yield was very high! So I think people are more comfortable being long the ten year than they should be. So whether it’s longs or flatteners, this view may be oversubscribed.
Looking at constant maturity is preferable to looking at a fixed U7-U8 graph. Another thing you could do is to look at ED11-ED15 spread. So ignore the sharp breaks at the contract rolls. If you just look at the main “body” of the graph, you can get an idea as to what the general portion of the graph should look like. Again, while Fed expectations are very important to any rate graph, the further out the location on the graph, the less Fed convergence matters and the more other supply demand forces may play a bigger role.
Curve AdvisorKeymasterDecember 29, 2014 at 8:25 pmPost count: 612
As for “columns 1-3”, it can be as simple or as complex as you would like to make it (for example, you can keep certain reference closes you can access in a drop-down menu). Think of various ways you think you may use the sheet, and as you keep looking at curvature, add features that you see fit. Some examples include:
* Do you need to keep track of YTD or MTD P&L?
* Do you like to keep track of how the curve has moved since a key date (FOMC meeting, payroll, month-end, etc).
* How does your sheet handle contract rolls? Ideally the fewer the manual changes you have to make the better.
* Consider formatting by 4 contract packs (whites, reds, greens, etc…). It will make reading your worksheet easier.
* when looking at the “current” prices, it may be superior to use a “weighted price”, instead of the “last price”. Because if you have a double fly, that structure could move 1.5bps each time one of the middle contracts prints the wrong way. Using some weighted prices based on the bid and ask sizes of a single contract (or spread or butterfly sizes for various structures, etc) may make more sense depending on the depth and trading characteristics of your markets.
Curve AdvisorKeymasterJanuary 8, 2015 at 7:06 pmPost count: 612
Here are some refinements you can make:
* looking at change-on-day is probably more helpful than looking at the settle/last price – not just in the contracts, but in the structures as well.
* at some point, you may want to add something that also looks at more fractional pricing for flies (so you don’t see 0.5+ bp moves in flies all the time, but where the midpoint may actually be).
* you can also use this to estimate your daily P&L (and do it by strategy)
But this is all based on what *your* particular needs are. So as you use it, feel free to add to it (i.e. structures you like looking at, certain legs you want to try and target to get thins on, etc., target profit/entry levels, etc.
Curve AdvisorKeymasterJanuary 14, 2015 at 4:33 pmPost count: 612
So more refinements you can make, for people who look at multiple markets. Depending on the complexity with which you watch a particular market, you can put the following on separate tabs, or on the same sheet.
* you can do a similar analysis for other markets (ie Euribor, short sterling, Euroyen, BAs, etc)
* you can also start looking at crosses for the individual contracts (i.e. EDZ6 vs ERZ6)
* you can start to look at crosses for various structures… spreads and flies (ie EDZ5-ERZ6 spread vs ERZ5-ERZ6 spread)
* you can also lag one structure in one country vs a similar structure in another country. For example, last year, it was popular to think of the US as following the UK, 6 months behind.
Some of the latter things require you to have a strong view of the curve in various parts of the world. However, you probably have started hearing a lot more the past 6-12 months of the effect that Bunds have on US tens, etc. So there are many traders looking at the relationships of various curves. If you have time, take a look. But don’t just look for random patterns – make sure you can explain why things in various currencies are moving the way they do, and understand when those relationships break down. Also, I am not saying to go through all the possible permutations of the above things – just the things that “make sense” to you, and that you want to look at.
meParticipantJanuary 14, 2015 at 7:27 pmPost count: 27
Just an idea – I am totally in the camp that everyone should build their own models otherwise they don’t understand the formulas numbers. But, I thought it might be helpful for some sample screenshots from an existing model as sort of a guidepost for setting this up. As they say a picture is worth a thousand words.
Curve AdvisorKeymasterJanuary 15, 2015 at 10:24 amPost count: 612
It’s not really a model – just a spreadsheet with the dozen or so columns I mostly indicated on the first post. And then depending on what you find useful, you can incorporate some of the refinements I mention on a few of the later posts. My sheet has a bunch of numbers floating on it that would require even more explanation, so I’d rather not go through it. Which again, points to how important it is for you to add the things *you* find useful in your trading. You are welcome to post a screenshot of what you have in the forum if you want some feedback. I can’t give feedback on every one, but giving feedback on a few could probably be helpful to everyone.
Curve AdvisorKeymasterFebruary 2, 2015 at 4:36 pmPost count: 612
As you build on your spreadsheet, and start looking at more complex curve structures, like condors, double flies, etc, you should consider using some kind of short-hand for the various structures. The columns in your spreadsheets aren’t going to be wide enough for you to to have “H5-H6-H7 year fly.” Consider developing some kind of shorthand for the structures.
I use the following nomenclature, but you should definitely weigh the merits, and come up with your own:
* I put a number after the contract to designate what kind of fly it is. For example, “EDU52” = “U52” = U5 6 mo fly. “U51” = U5 3 mo fly, “U53” = U5 9 month fly, “U54” = U5 1 year fly… Basically the number at the end indicates how many contracts wide the fly is. You can also use this for generic contracts: “ED011” = “011” = ED1-ED2-ED3 3mo fly. This way, I don’t have to write that much, but it conveys exactly the same amount of information, in a smaller, easier-to-read way.
* You can use the following letters to designate other fly structures:
– “S” = spread. “U52S” = U5-H6 6mo spread
– “C” = condor. “U52C” = U5-H6-U6-H7 6mo condor
– “D” = double fly. “U52D” = U5 6mo vs H6 6mo double fly, 024D = ED2-ED6 1yr double fly
– “E” = double fly 1 fly apart. “U52E” = U5 6mo vs U6 6mo double fly
You get the idea. Depending on what your particular needs are and the particular structures you like to trade, you should come up with your own nomenclature.
Curve AdvisorKeymasterFebruary 25, 2015 at 11:03 amPost count: 612
How best to look at butterfly prices? Most people use a column of “last price” and construct a formula like “= ED1 – 2*ED2 + ED3”. But this method is not one of the better ways to look at a fly price, as it will show a much wider trading range than the reality. The reason is that sometimes (especially for contracts further out, or if you trade in a non-liquid period like overnight), you can go 30 minutes or more, without anything trading in a particular contract. So you end up not comparing apples to apples. You can get large errors looking at the fly markets this way, since the markets may move, and one or more contracts get left behind. Butterflies generally do not move much day to day. If we get a high or low settle on one day, we will be back to the same bid and offer (as the previous day) the next day. Below are some of the ways you can look at where a butterfly is trading:
* Actual bid/offer/last on the fly. Butterflies trade in Globex as a product (ie. “GEH15:H16:H17”). So you can see what is trading there. You can also look at the bid and offer size of the fly to get an indication of where the true fly price is, within the bid/offer spread. This last point may not be as useful if the bid/offer spread is not 0.5bps wide.
* Actual bid/offer/last on the spread legs of the fly. Spreads trade in Globex as a product (ie. “GEH15:H16” and “GEH16:H17”). So you can see what is trading there for the two component spreads. You can also look at the bid and offer size of the spreads to get an indication of where the true spread prices are, and then take the difference. This last point may not be as useful if the bid/offer spread is not 0.5bps wide, but spreads are more likely to have tighter markets.
* Actual bid/offer/last on the individual contracts of the fly. You can look at the bid and offer size of the contracts to get an indication of where the true prices are, and then do the fly calculation. You can weight the bid/offer prices by the bid/offer sizes to get a “weighted” price for a contract – this “adjusted price” is probably a better indicator of where a contract is, than the “last price.” This point may not be as useful if the bid/offer spread is not 0.5bps wide, but individual contracts are most likely to have tighter markets.
Experiment and use one or more of the above for your spreadsheets.
meParticipantAugust 6, 2015 at 8:08 pmPost count: 27
Something like this? Feedback requested.
I am thinking about trying to incorporate somehow the actual bid/ask/last on the spread legs of the flies. But, thinking it might get cluttered a bit. Currently I can switch between weighted / last / settle on the individual legs.
What I call ‘Offset Double Flies’ are Double Flies with extra space. So the 6 mo. offset double fly would be one in which the two flies were a year apart. Not sure if there is better nomenclature.
It takes prices from my BBG feed. Just need to figure out how to use it. 🙂
Attachments:You must be logged in to view attached files.
Curve AdvisorKeymasterAugust 11, 2015 at 10:46 pmPost count: 612
Apologies in the delay in replying… I have been travelling, but I will put a series of additional “make-up” Forum posts when I return later in the week.
As for you spreadsheet, it looks okay. I have the following comments:
* This is just my personal preference, but prefer to look at price feeds in its pure form. So I would make the turn adjustment in a separate column, and not in the columns with the bid and ask. I realize you can toggle off between turn-adjusted and non-turn-adjusted. But you may want to reference this sheet before you put in an order, and it may be confusing not to see the numbers match your trading platform. Also, there may be times when you forget you have the toggle on (or off).
* You are not calculating the weighted prices correctly. For example, with ED4, there is almost nothing on the offer, so the weighted price should be almost equal to the offer. This may be one of the reasons your 6mo fly curve is so jagged. Another check you should do is to make sure that the bid/offer spread is 0.5bps wide before trying to calculate the weighted price. If the spread is 1bp or more, you may want to consider another methodology.
* You may find that the spreadsheet extends more columns than can comfortably fit on a printed page, or monitor. Consider each column and its relevance – you can hide the columns that do not add much value (or are repetitive). You can also consider extending the analysis on to a second set of rows if you have a large number of things you like looking at.
Curve AdvisorKeymasterJanuary 19, 2016 at 1:50 pmPost count: 612
In my quest to “teach a man to fish” (and have you do your own analytics), I think perhaps I may have been a little too vague and have not highlighted the importance of certain posts. Case in point, post #1106 (http://www.curveadvisor.com/forums/topic/basic-eurodollar-analytics/#post-1106). The main point of the post is that: Last price prints are not the best way to see where the market is, when looking at a multi-legged structure like butterflies. This is just one measure. However, there is a lot of information in the bid/offer prices and quantities. Use this information! It is probably the most valuable price information you have.
I listed the three sets of bid/offer prices available to you (butterfly, spread and single contract). I discussed how you can weight the bid/offer sizes to calculate a more accurate “mid” price. This will give you SIX different ways to price a butterfly (for example, H7-H8-H9 fly):
1) last print on H7-H8-H9 fly
2) weighted bid/ask price/qty calculation on H7-H8-H9 fly
3) last print on H7-H8 spread and H8-H9 spread
4) weighted bid/ask price/qty calculation on H7-H8 spread and H8-H9 spread
5) last print on H7, H8 and H9 single contracts
6) weighted bid/ask price/qty calculation on H7, H8 and H9 single contracts
If you had to choose only one, I would probably go with #6. This is because it’s the most “consistent.” There may be times when the bid/ask on flies or spreads are more than 0.5 bps wide, which makes the last prints and weighted calculations less robust.
[to be continued]
Curve AdvisorKeymasterJanuary 20, 2016 at 1:14 pmPost count: 612
There are two reasons why looking at accurate pricing is critical for structured trades:
1) Accurately knowing where markets are. Having accurate information is critical to making good decisions. Sometimes, a fly may not trade all day. However, this does not mean it does not move. You need to know where each trade in your book is at all times. Having good information on where the butterfly and spreads are trading will also be helpful in selecting opportunities for manual legging. For example, the market on a fly may be 2/3, but it may be that one of the spreads are better offered (or bid), which would allow you to leg into the trade in less liquid markets.
2) Accurately knowing your P&L. The closes can be off from time to time. You will find that when a butterfly closes too high or low, the next day, this will revert. While the closes can vary from day to day, you will find that the bid/ask prices are generally much more stable. You should expect some P&L volatility when trading butterflies. The daily settlement procedure is more focused on the trading activity of the single contracts. However, the actual volatility is much lower. If you know what the weighted “mid” of all your flies are at 2:00pm CST (the daily settlement time), you will know exactly how overstated (or understated) your P&L is, and how much you should get back the following day.
Curve AdvisorKeymasterJanuary 20, 2016 at 7:51 pmPost count: 612
In post #1380 in this thread (http://www.curveadvisor.com/forums/topic/basic-eurodollar-analytics/#post-1380), I made reference to “If the spread is 1bp or more, you may want to consider another methodology.” When this is the case, make use of all the information you have available to you. Say you look at some illiquid fly and the market is 4/5 (bid/ask). The last trade from overnight is 4. Don’t just assume the price of the fly is 4 because it’s the last print (from 8 hours ago) and it’s within the bid/offer. You have information in the spread markets as well as the single contract markets. Try and make use of that information. You may be able to ascertain if the market is closer to 4.5 bid or 4.5 offer. And you may be able to estimate to within a tenth of basis point, where the “mid price” on the fly is. I gave you six pieces of information you can look at 2 posts ago. Develop your own methodology and/or weighting for what gives you the best results. There is no ONE “right” answer (although some are better than others). At the very least, you will have an estimate that is better than the “4” last print from overnight, and the wide 4/5 market you got at the end of the day.
Hmm… that sure is a wide market… I wonder where it will settle today?
Curve AdvisorKeymasterJanuary 25, 2016 at 5:06 pmPost count: 612
One of the more important uses of accurate pricing is to figure out where your P&L is each day. I generally discuss all the great things about trading curvature structures. One of the drawbacks is that you are subject to daily mark-to-market risk. So trade in sizes that are appropriate to your company’s compliance and oversight policies. This mtm risk is disproportionately large as compared to a single contract trade, because you will generally have more size on than a single contract trade. A curvature trade will not move as much as a single contract trade, so you need to do more size. Take a butterfly trade, for example. An equally weighted contract will have 4 pieces, each of which can settle on the wrong side of the bid or offer. So the close may be off by a bp or more. This is not common, but it is not uncommon either. The fly close may even be through the bid or ask of the fly market. This is because the algorithm that determines closing prices puts the bid/ask of the fly markets as a very low priority.
So it is imperative for your sanity to know where structures are trading more accurately, and where things are marked at the end of the day. You generally don’t have this issue trading single contract structures. Because if you are long EDZ7 at 98.60, whether we settle at 98.75 or 98.755 is mostly irrelevant. When you have an accurate measure of the curvature structures, get into the habit of taking a snapshot at 2:00PM CST. This way, you will know where things were trading, and you can compare to where things settled. If you got a favorable settle, you will probably get that back tomorrow (all other things being equal), and vice versa.
Curve AdvisorKeymasterFebruary 25, 2016 at 10:42 amPost count: 612
I wanted to expand on something related to developing robustness of your pricing. I’ve previously mentioned that you can use not just the information from the single contract market (bid/ask levels and sizes), but from the spread market and fly markets as well. Consider the following facts:
* if you are given the level of the first contract, and a series of all 3 month spreads, you can construct the entire curve. This may seem obvious to most of you. But I don’t think most people have thought about the ramifications of this. So knowing where the all the 3mo spreads are (with a starting point) is the same as knowing where the entire curve is.
* if you are given the level of the first two contracts, and a series of all 3 month butterflies, you can construct the entire curve. Most people probably haven’t thought about this. So knowing where the all the 3mo flies are (with two starting points) is the same as knowing where the entire curve is.
So you can make your intraday pricing model more robust by bounding the interpolated results from the single contract market. Keep in mind that you can use any of the 3mo, 6mo, 9mo and 12mo spread and fly markets. So while most people only think of only ONE set of data to construct the ED curve, there are actually many. While the single contracts are the most liquid, there is data you can use from the other curves.
Curve AdvisorKeymasterDecember 24, 2016 at 7:43 amPost count: 612
Viewer Mail: If you don’t mind, I did have a couple of questions.
I’ve been making a spreadsheet with the historical prices of the structures but I’m not really happy with the way I’ve priced them. The only daily historical data I can get access to at the moment is:
a) last trade of the outrights
b) last trade of the 3 month spreads
c) settlement of the outrights which for us is the offer price at close.
Assume my outrights of a fly close at 0/1, 0/1 and 0/1 (our tick size is 1 bp). Using a) my fly price can be anything from -2 to 2 depending on the last trade of each outright. Using b) my fly price can take a smaller range of 2bp depending on the spreads instead of 4bp like a). Using c) i will always have the mid point of a).
Which do you think is the best given I can’t get bid/ask size historical data?
Also, I’ve been trying to do a contant maturity adjustmant which works fine for all but my first contract because I can’t get access to the historical data for the spot rate which for me is the bank bill swap rate. If we are 40 days to maturity and I’m adjusting a first outright that is 60 days to maturity can I interpolate backwards using the first spread or is that too crude of an estimate? I can start to record the BBSW data daily but I can’t get historical data.
If i start recording accurate data now without using historical data, how long till I will be able to use it? I’m really not sure if it is worth the effort to start from scratch.
Curve Advisor reply:
I would use (c) because the issue with last price is that for some less liquid structures or in fast moving markets, you may be comparing apples to oranges, in that you could be comparing two prices that were “hours” apart. This is bad if you are going to use the pieces to construct a structure price. However, settlement does have the disadvantage of not being able to see where something is intraday. But if you don’t have live pricing, I’m assuming you aren’t trading? So maybe just looking at the end of day prices wouldn’t be so terrible.
As for what to do if you don’t have a spot rate, I suppose if you are not looking to extrapolate too far inward, your suggestion could be okay. I’m a little surprised that data wouldn’t be available. I don’t trade IR futures so I don’t know the details, but I would think the following interest rate data from the RBA may be close enough (you should verify): http://www.rba.gov.au/statistics/tables/ There is also a section on the RBA site with data even further back.
You don’t have to wait to do any analysis. Use a data site like Quandl.com (and the above RBA site). It is free and you can get historical data. Quandl also has “continuous futures” that you can use (if you don’t want to extrapolate or even do the calculations). I believe you can also auto-download the data to a spreadsheet.
NickParticipantJanuary 7, 2017 at 12:24 pmPost count: 8
When analysing flys and spreads do you generally find it more useful to look at the term structure roll (GE1-2*GE5+GE9) or yearly roll (GEH17-2*GEH18+GEH19) of historical data? I imagine it depends on what your thinking at the time/looking for?
Thanks in advance for any insights.
Curve AdvisorKeymasterJanuary 8, 2017 at 5:41 pmPost count: 612
It’s hard to generalize. Most of the time, I look at neither, since I prefer constant maturity structures (that I calculate). The general methodology is in the following thread: http://www.curveadvisor.com/forums/topic/data-set-methodology/
But there may be times where looking at the generics (GE1-2*GE5+GE9) or the actuals (GEH17-2*GEH18+GEH19) charts could also be useful. The three types of charts (constant maturity, generic or actual) all have their pros and cons. The generic and actual are easily accessible, and those also contain very useful information. In a nutshell:
* I like looking at generics when the structure is further out the curve (where any spike from the contract roll is minimized). This could also be useful if you just want to see what a possible range for a contract could be (over a longer period of time). If you are looking at ranges, you may also want to check a similar structure located nearby (for example, if you are looking at ED12 1yr fly, you may also want to look at ED10 thru ED14 1 yr flies).
* I like looking at actuals when I want to see something specific to that part of the curve. For example, if I wanted to see what the behavior of the year-end turn around EDZ8 was, I may look at EDM8 6mo single or double fly. Actuals are also useful if you are looking at a short term time frame around a recent contract roll. Around a contract roll, the generics may contain too much noise. When looking at actuals over a longer time horizon, you should also look at the surrounding structures, if you want an idea of how a structure moves over time. For example, if you wanted to see EDH18 1 yr fly, you may want to see what EDU17 1yr fly looked like 6 months ago or what EDZ17 1yr fly looked like 3 months ago. Each of these structures represent a similar structure with similar times to settlement.
There will be other cases where one type of chart may be useful over another. But I just wanted to give you some examples. The most important thing is to consider for each trade, what aspect of its history you want to examine, and choose the appropriate type of chart.
NickParticipantJanuary 27, 2017 at 11:23 amPost count: 8
Please could you give some suggestions on how the curves obtained from the sheet can be of more use? I understand the concepts above with regards to negative/positive flies and the rate cycle. However, many of the 3m, 6m and 12m fly (and spread) price points will be beyond a years time, where frankly it would seem almost a fools errand to predict what may happen or where rates may be for Eurodollars at least (I imagine Euribor, Sterling and BA’s may well be different).
e.g. Currently I’m getting a price of -0.5 for the Z18 6month fly. On first thoughts, that does seem very cheap, especially compared to the surrounding flies – but it seems out of context as I have no real view that far out.
Am I over complicating this? Perhaps it’s just a useful tool as so far as one has some kind of view on – which for me, is perhaps 18 months at the most for Eurodollars. So in my example above, the best I could do is lift the offer, and aim to flip it for say 1.5 bps in the coming week/s (not that I’m asking for investment advise of course!)
Thanks in advance.
Curve AdvisorKeymasterJanuary 29, 2017 at 9:11 pmPost count: 612
What we are trying to do is not necessarily pick where structures will be (although if you had an accurate crystal ball of where the curve would be in 1+ years, that certainly would be helpful to say the least). Even though the structures may settle 1+ years out into the future, what we are actually trying to do is to pick good risk/reward structures for the shorter term. For example, if we buy the Z8 6mo fly, we are not necessarily looking to hold for 23 months. Rather, we think the expected value of the gains outweighs the expected value of the losses over the near term. So when we look over the various scenarios in the next day, week or month, we think because of its location and relative value, we would make more on average on an expected value basis. This could be some combination of making more on a selloff than a rally, or having more scenarios where it makes money, etc. A simple way to think about it is think about what is on the calendar and think about how the curve would move on a 10bp rally vs a 10bp selloff. Think about the probability of the next 10bp move being a selloff vs a rally. After a while, you will get a feel for how the curve moves in different environments. The curve moves differently depending on the cause, and the environment.
Regarding the -0.5 price on the Z8 1yr fly, you did not turn-adjust the fly. I discuss the year-end turn in a number of places in the thread, including item 4 of the first post in this thread and in the following thread: http://www.curveadvisor.com/forums/topic/data-set-methodology/#post-522. However, the turn has widened recently, so you need to use at least 1.25bps in the whites, reds and greens, around 1 for the blues and about 0.5-0.75 further out. So if you se 1.25bps for the turn, the actual value of that Z8 6mo fly would be 2bps (= -0.5bps + 1.25 + 1.25), since there are two turns in the structure. 2bps could be still considered somewhat low, but it’s not as “crazy” as it was before we adjusted for the turn. You should always make adjustments for the year-end turn and keep in mind the turn value can change (one way of thinking about it).
It’s really hard to say in a vacuum what trade is good for you. It would depend on your costs structure, margin structure, your risk preference, your trading objectives, your views, what the rest of your book looks like, just to name a few factors. But you should know that the structure after adjusting for the turn is not as low as you may think. Let me know if I answered your questions.
NickParticipantJanuary 30, 2017 at 12:45 pmPost count: 8
Indeed you did. Thanks very much. Your answer also gave me a better understanding of why the Z turn is so important to consider – I also wasnt aware of the values as I believe I read somewhere that since the financial crisis, it was now only worth 0.5 bps. Perhaps not then!
Looks like I need to re-read the Z Turn thread again and the Burghardt paper.
It’s definitely good to see these issues put in the context of an example.
As I’m sure you can tell, I’m not working in a hedge fund or bank (Im mostly scalping Bunds & Bobl) – but am looking to learn more about spreads and flies, especially in ED. Coming across your site was a stroke of divine luck!!
Curve AdvisorKeymasterJanuary 30, 2017 at 8:19 pmPost count: 612
The year-end turn was closer to 0.5bps when rates were near “zero.” But lately, it has been closer to 1. One of the things I pointed out a while back was that as rates rise, you may see larger year-end turns. Way back when I started trading (2000), the year-end turn was closer to 6bps. I don’t think we see that again, but it does chop around, so when you look at your portfolio, it is prudent to look at your net exposure of “Z” contracts to make sure it is not absurdly large (unless you have a view on the turn).
Interestingly, the turn is negligible in Europe, so it is a little more straightforward to trade the European curves. If you trade Bunds and Bobls, you may have some familiarity with how the curve moves on various events. Also add the Schatz to the mix, and just see how many basis points the different parts of the curve move on various events.
Glad you like the site!
You must be logged in to reply to this topic.