Tagged: FAQ Questions
Curve AdvisorKeymasterSeptember 23, 2014 at 2:33 pmPost count: 612
For anyone who has general questions about anything related to eurodollars, but does not want to start a new thread, post questions here. I may post in this thread replies to general questions I receive via email. I will attempt to post a reply to most questions, but I may refrain from posting if I do not have a strong view, or I just don’t know the answer.
Curve AdvisorKeymasterSeptember 24, 2014 at 4:10 pmPost count: 612
As a start, enclosed is what I call the Curvature Primer. It’s draft of something I wrote earlier in the year to explain why one would want to trade butterflies, and what some of the basics are. Let me know if you have any questions or comments.
William SloanParticipantSeptember 27, 2014 at 6:12 pmPost count: 12
Since EuroDollars are a rather complex futures contract, I sometimes make or lose money and I don’t know exactly why. So my question is about a trade from about this time last year (long 5yr futures, short greens (Z15 to U16)). This particular trade I let get away from me and I really got burned on it. Greens just kept looking more and more expensive and I kept selling them and buying ZF against em until I ended up having to buy greens at an even higher level. At this particular time in the market, ZF sold of and established a decent looking downtrend channel on a daily chart but greens just didn’t follow it, the correlation just went right out the window and I have always wondered why this trade behaved like that.
I am well aware that the correlation should breakdown in flight to quality trade that is causing credit spreads to widen. But in this case swap spreads traded to like all time lows in the 4 and 5 year sector when 5 yr yields were actually rising. This losing trade has really bothered me because I still don’t understand why I lost on it.
So if anyone would have some insight to why last year swap spreads narrowed so much so fast while treasury yields rose, I would greatly appreciate it.
Curve AdvisorKeymasterSeptember 27, 2014 at 7:29 pmPost count: 612
I have a big deadline Monday so I have to be brief, but I will expand next week. My key points regarding your post:
* You should only trade products where you feel like you have some sort of “edge”. Or something where you are at least not at a disadvantage. Remember, trading futures is a zero sum game. I would be shocked if there weren’t at least a dozen “major” bots jobbing the spread between EDs, treasury futures, treasuries, swaps, and pretty much any other related product. So I’m not sure where an individual trader’s edge would be.
* I did not trade 5 year futures (except for maybe the occasional punt)… mostly because I felt like I had no edge. So I am not as knowledgeable on the subject.
* However, it is pretty clear to me that the reason you lost money on the trade is because you had a duration mismatch between the EDs and FV. You probably only needed to hedge with ED9 (approximately). The fact that you tried to hedge with ED9, 10, 11 and 12 means you went too far out the curve. For example, I see that ED9-12 spread dropped over 10 bps from 9/17 (day before the FOMC) to 9/24. Also. I’m guessing swap spreads in general came in as well over that period, which would exacerbate your losses.
* And since the ED contracts probably just rolled (mid-September), you probably had an even larger duration mismatch than usual, as Z5 had much longer to expiry than a “typical” ED9 contract.
* It is remotely possible there was some sort of change in the deliverable security for FV (not sure if this would work for you or against you – again, this is not my area of expertise), but I strongly suspect most of your loss is probably the duration mismatch.
Let me get back to you on the “why swap spreads narrowed” question next week. I hope this helps. Let me know if you have any follow-up questions, or if you want to provide more details (numbers, dates, etc). Thanks.
William SloanParticipantSeptember 29, 2014 at 11:16 amPost count: 12
Thank you for the response.
To add a few more details on the trade, the majority of the position was in ED-9 and the dates were from the end of October to the end of Nobember 2013. When I regressed the position against the 4 and 5 year swap spreads they fit very close. That led me to draw the conclusion that the “problem” was how directional swap spreads became. I guess I was just surprised that spreads would narrow that dramatically when treasury yields were rising.
Curve AdvisorKeymasterSeptember 29, 2014 at 2:06 pmPost count: 612
In looking back, swap spreads did come in a lot. Again, this is not something I look at very often. My original thought was that this was related to the government shut-down, but the timing of the largest dip appears to be closer to December, so maybe this was something related to expectations of tapering?
Let me think about it some more.
ABC67ParticipantSeptember 29, 2014 at 2:52 pmPost count: 1
Last year, there was very heavy selling in the 5 yr sector by real money and foreign central banks. At the time, it was uncertain exactly why there was such heavy selling going on. I looked back and found this from Credit Suisse on 11/18/13 “There is little clarity on what is driving the latest move in spreads and
certainly little has changed on the fundamental side – budget or Fed purchase
prospects, while the flows have been biased toward receiving but not of the
order that we would expect to see behind this move. Positioning in spreads
themselves are probably still offsides given the amount of questions we are
getting.” Thus, heavy selling of old 5 yr paper led to spread compression well below what would fundamentally be expected. In addition, as spreads came lower, spec accounts began fading the move and creating large positions in wideners. As the Real Money selling persisted, specs got stopped out and 4.5 yr swap spreads briefly traded 0 (or slightly negative for some off the run issues). It was a perfect example of the market being irrational for longer than the specs could hold the trade, with stop outs in conjunction to the heavy selling bringing the market to levels that would be considered fundamentally impossible.
Hope that helps
Curve AdvisorKeymasterSeptember 29, 2014 at 6:51 pmPost count: 612
Thanks for the assist ABC67!
Now that I have a little more time,
* I suppose the reason I never actively followed the swaps or treasury markets is that going by the “zero sum game” logic, I felt like there were so many people with an information advantage over me (ie market makers who actually see all the flows for both swaps and treasuries, bond basis traders who keep on top of everything (supply, financing/repo, etc)), and it was not clear what my advantage was. I sat next to two bond basis guys in my career and they would always be complaining that their edge was getting eroded away. And it’s because you could program a machine to take in all the market inputs and automatically do any attractive trades – instantly. And even if something looks a little off, you can forget about being able to leg into it faster than the bots. I may just look at those markets once in a while as a sanity check.
* I’m guessing you must be doing that trade because you wanted to have a spread widener vs something else? Because the location on the 2 year graph isn’t screaming “buy me,” and the fact that the FOMC was going to buying less treasuries “soon” with tapering was probably not good for swap spreads.
* so how many basis points are we talking about? It looks like the swap spreads may have come in about 6bps in that 1 month? If you want to present your analysis, I’d be happy to look it over.
William SloanParticipantOctober 1, 2014 at 3:30 pmPost count: 12
Thank you for all of the replays. I do find some satisfaction to know I wasn’t the only one that got hurt on that. Thanks for that color ABC67.
With regards to my analysis I was trading CBOT treasury spreads (TUF and FYT) against EuroDollar spreads and just trying to leg in and out of them. I did weight the 5 yr leg of it a bit heavier because the 2 year is usually more volatile and I was punting a lil on direction because the 5 year leg looked “cheap” and I just didn’t think it could come in as much as it did as fast as it did. I had always figured that if those things were gonna run, it is to the upside. But as ABC67 said when everyone needs to get out at the same time, the speed and magnitude of a move can become exaggerated.
I find the Treasury vs EuroDollar trade to be interesting. I am enjoying what I have learned about butterfly’s from the forum and when I get some time beging looking at some flys vs treasuries.
Curve AdvisorKeymasterOctober 2, 2014 at 9:02 amPost count: 612
It’s interesting you mentioned flies vs treasuries… because the trade I just put out to clients today involves selling a fly vs a blues contract… the rationale being that the ED fly is too high (pricing in hikes for Q2 of next year), but that is a disconnect with tens at 2.40. I really can’t see the FOMC hiking in Q2 of next year and tens be at 2.4 – so one of the two things have to give. Because I feel like there is not much difference between trading an ED vs trading a treasury future (since the two are generally kept in-line), I suggested just selling a blue contract. I will post when clients are out of the trade.
AnonymousOctober 3, 2014 at 11:45 pmPost count: 2
Your methodology of finding trades really appeals to my trading philosophy, and so I have a few questions relating to the Curvature Primer document.
When I try and recreate the scatter plot that you show, I am not getting the smooth arch of data points like you. I suspect this is because you are using constant maturity Eurodollar prices, whereas I am using the individual contract months, and simply rolling forward at the quarterly expiry. Is this the most likely reason?
The second question is how are you identifying and monitoring opportunities? Do you create the two charts you describe for all of the different structures available? That’s a lot of scatter plots to watch! I have previously attempted to look for trades through creating curves of spreads and flys, but didn’t find it very useful as the movements appear so miniscule from day to day, and I gave up on the idea and returned to historical charts of the structures. So I am interested to hear how you do it, to see where I went wrong.
Thanks, and I look forward to bombarding you with more questions.
Curve AdvisorKeymasterOctober 4, 2014 at 7:54 amPost count: 612
Well, I see this might be the best $25 you’ve every spent… 🙂 I will answer your questions, but I will answer a little every day, just to give others a chance to chime in if they want (eh hem). I’m actually surprised some of the people newer to ED trading haven’t asked more questions.
1a) Yes. The biggest issue with people looking at Eurodollar historicals is that they do not use constant maturity. To get the most accurate picture possible, you should try and compare apples to apples. The errors will be magnified over the contract roll if you look at something smaller, like a 6mo double fly. And even trading something like a 1 year fly (that would not be affected by the contract roll) could still give you a bad picture, because you may be comparing a current point with a point as much as 2-3 months different. Not rolling the contract forward could be fine if you are trading technically, but it is not a good way to look at trades from a long-run perspective. Also, being aware of how far away a contract is from maturity is always important in EDs – and not just for historicals. Whether you are hedging, or trading EDs vs other securities, you should be aware that you could have a risk mismatch by not understanding that the risk parameters on a ED changes every day – because the interest rate maturity implied by the contract changes every day. There is a “Data Set Methodology” thread that discusses how to set up your spreadsheet/model, in case you are interested in looking at constant maturity and do not have a vendor that supplies such data (I am assuming such vendors exist – I do not know what is currently available). I’m not a big fan of paying extra for this, but it is built into certain platforms.
A “possible” second reason the scatter plot may not look right is if you do not factor in the year-end turn. This may only affect things like certain 6 mo flies or double flies. This used to be a huge problem when the turn adjustment was 6+bps, but is a much smaller factor now.
I will discuss the following early next week:
* Shortcuts for if you want to look at butterfly trades but do not want to spend time building a massive model/spreadsheet.
* Answer to your second question. And I will not say it is “proprietary.”
Curve AdvisorKeymasterOctober 5, 2014 at 3:17 pmPost count: 612
To answer your second question first, here is Part 1 of my reply:
No – I do not create scatter plots for everything I am looking at, and you are right – that would be too much. It can be quite tedious to look at historical charts when the markets are not moving. And even if you find something at historical highs and lows, a lot of times when you get a big move, half your spreadsheet is saying something is too high or too low, so it is unclear what you should be focusing on. And even when you can isolate it to a few things, many times, it could just be a bad trade. I pass up many more structures that are at historical highs and lows than I actually use to develop into a trade idea. When I was a junior trader on the desk, one of the senior traders had me make up a spreadsheet regressing the various flies on various x variables (single contracts and spreads). I did an awesome job – fitted values, residuals, r^2, z score, percentile observation, conditional formatting, non-US futures as well, etc. I think the senior trader called it a “work of art.” But I haven’t used that spreadsheet since 2002 – it’s just collecting cyber-dust on my computer somewhere.
Because if trading was all about just following a spreadsheet, even a monkey could do it. I’m not saying a spreadsheet can’t be useful – it’s a very useful tool. You just have to realize that it is just a tool.
Part 2 coming soon.
Curve AdvisorKeymasterOctober 6, 2014 at 10:32 amPost count: 612
As there is no ideal way of looking at the markets, you need to make use of all of the following three components:
1) Understanding the historicals. A sheet that tells you when we are at the various highs and lows are (ie 3 mo, 6mo, 1 yr, etc) could be helpful. So could a sheet that tells you how high or low something is to some form of regression/modeling. But you also need to…
2) Understanding market positioning/dynamics. Things like… “what is that fool BG doing? and when will he be done” “what trades are people likely to do?” “which part of the curve are the market participants likely to go long or short?” etc. If you know some part of the curve may be bad news, you should stay away, even if the historicals say otherwise (or do much smaller size).
3) Understanding the current market environment. at the end of the day, the shape of the curve is going to be function of FOMC rate probabilities. You may need to make some adjustments, like spread, convexity, etc. But if the long term FF rate is only going to be 3.75, or if the FOMC plans to be on hold until the UR is below 6.5%, or whatever, that will affect the shape of the curve. More importantly, you need to keep this in the back of your mind as you look at historicals, because some point when the FOMC was in the middle of QE is not going to be as relevant if they are close to hiking.
I put “historicals” first in the above list, but that is probably the least important of the three things. I generally think if something was so obvious that you could just look at historicals to trade off of it successfully, someone’s already programmed an algorithm to look for those types of trades and trade off of it.
So when I look for trades, most of the time, I generally have an idea what looks rich or cheap in my mind, based on factors 2 & 3. I then use the historicals to corroborate. There are other times when I do not consider a certain trade, and I use the historical analysis to highlight something. But you basically need to have all three components working to come up with a good trade.
Curve AdvisorKeymasterOctober 7, 2014 at 8:28 amPost count: 612
But don’t think that the fact that historicals are limited is BAD. In fact, it is GOOD! Because this means it’s not as simple to have an algo rip off huge chunks of P&L from just the historicals. I’m sure the bots rip off tiny pieces of the ED market, just like any other market. But there is value to be had if you develop the right tools and spend some more thought on what is going on in the interest rate markets. More later.
Curve AdvisorKeymasterOctober 9, 2014 at 3:37 amPost count: 612
One of the reasons to have a good set of reference data is to highlight highs and lows. We’ve been looking to take profit on a few things, and the historicals can be a good reference point. Also, I suggested a new trade today. I knew it was a pretty good level, but when I looked at the historicals, that helped confirm it. So you should definitely work on developing good supporting analytics.
Curve AdvisorKeymasterOctober 11, 2014 at 4:00 amPost count: 612
Finally, I would say almost half of the traders I worked with over my career looked at butterfly trades actively. Very few of them actually built any kind of massive spreadsheet to process the data. Almost all of them were successful, to varying degrees. So you do not NEED to have your own massive historical spreadsheets to make money trading curvature. However, all things being equal, I think it is HELPFUL doing your own analysis of the data. Whenever I had a junior person do a lengthy rotation with me, I would have them build spreadsheets. It wasn’t for me – I don’t use other people’s spreadsheets. But I thought it was a good learning tool for them, and something they could use when they followed the markets and when they eventually traded.
The other benefit to doing your own data analysis is that you can do a lot of custom analysis. For example, JPM management used Value-at-Risk to monitor a trader’s risk. Many times, VaR could overstate the risk on a butterfly because you get bad closes from time to time on say a double fly. The next day, it usually reverts back. So most of what VaR did was to capture the 7 worst days of bad closes. I used the data I collected to build my own VaR estimator. I would do a trade in medium size, but I could calculate my own VaR, and see what risk management saw on a new position before I did it in massive size. And if there was something I disagreed with, I would go chat with my manager BEFORE I super-sized the trade. You just always look smarter when you do this, than just doing the trade first and then getting the tap on the shoulder because you exceeded your risk limits (or what the VaR calculation thought was exceeding your risk limits). Or I would find something that was more “VaR friendly” (like in options or something). Back to the topic at hand…
Curve AdvisorKeymasterOctober 11, 2014 at 4:26 amPost count: 612
However, if you are not exceptional with spreadsheets/modeling and managing large amounts of data, you may be better off not building a massive historical model/spreadsheet. Everyone thinks they are good at Excel – just like everyone thinks they are a good driver. But this is not the case. You may make a ton of errors. And actually, just one major error is enough to make the risk / reward proposition on your spreadsheet high. Your alternative would be to do what many people who trade flies do:
* have a good way to look at butterflies for that current day only. You could make a basic sheet in under 5 minutes. have a column of ED prices (you can even use live feeds), have a column for turn adjustments, and then the rest is just calculating fly prices. Obviously, you can easily spend days making an awesome spreadsheet.
* whenever you need to look at a trade, just create a custom definition in Bloomberg or whatever you use for data analytics. Use generic contracts instead of actual contracts (i.e. “ED9” instead of “EDZ6”). They may or may not roll the contracts properly. Even if they roll the contracts properly, they may not turn adjust properly. But you just have to keep that in mind as one of the limitations when looking at certain flies and double flies.
* If you want, you should also look at the structure 3mos in front and behind it. So if you are looking at the ED9-11-13 fly, also look at ED8-10-12 and ED10-12-14. This may highlight any issues that may come up because the contracts were not constant maturity (or not done properly). And I find it useful to see how a contract rolls forwards and backwards.
* And I suppose you now have another option, which was not available to me, which is just to say, “hey Joe, what do you think of this trade?” Actually, this is a service I provide to Trades issue subscribers, but I don’t mind doing this for forum readers once in a while (like once a week or something). And no – I won’t front-run your trade. I’ve probably already seen it. 😉 But if there is interest in discussing trades, maybe I can look into setting up some private forum feature where we can discuss it for a day or two, and after you do it (assuming it is good), we can make it public for discussion. Not sure – but I am open to whatever you folks want to do.
Curve AdvisorKeymasterOctober 16, 2014 at 4:01 pmPost count: 612
I seem to have killed this thread. Was my answer terrible?
I was telling someone that you can use the data to create a VaR calculator. This is actually a great tool if you are “new” to trading a book. You can see what the correlation of different types of trades are, and you can see what types of days are good for your trade, and which are poor. You can also add a feature that allows you to enter scenarios to stress-test.
William SloanParticipantOctober 20, 2014 at 11:49 amPost count: 12
With regards to something looking rich or cheap based on historicals, would it make sense to somehow change the data sample? I am thinking what if the data that is being used for analysis is instead of the most recent history is data further back in time that would be more similar to where the market is in the fed cycle based on fed fund futures. So with fed fund futures saying that the first hike is not going to be until the end of 2015 would it make sense to sample the data from 03 and 04 to perform the regression on?
Curve AdvisorKeymasterOctober 20, 2014 at 9:26 pmPost count: 612
Some analysis of what happened in 2003-4 could be useful. However, I think the previous 12-18 months is probably more useful. Yellen is not Greenspan, the environment 10 years ago were not the same as it is now, the growth and inflation outlook are different, etc. One thing that is clearly different is the fact that back then, it was common for the markets to price in more aggressive hikes in the beginning. This is less the case now. So while a study of that time could be useful, I’m not sure it’s as relevant. So you always have to be asking if the historical comparison period is relevant.
AmitParticipantNovember 19, 2014 at 9:41 amPost count: 2
Hello, in one of your earlier posts, you mentioned:
whenever you need to look at a trade, just create a custom definition in Bloomberg or whatever you use for data analytics. Use generic contracts instead of actual contracts (i.e. “ED9″ instead of “EDZ6″). They may or may not roll the contracts properly. Even if they roll the contracts properly, they may not turn adjust properly.
I was wondering what you meant when you said “they may not turn adjust properly”.
Curve AdvisorKeymasterNovember 19, 2014 at 11:41 amPost count: 612
I you look carefully, all the Z contracts will appear about 0.5 to 1 bp cheaper than the rest of the contracts. This is the “Z turn” (there is a thread in the Forum with that name as well). You need to add a small adjustment to the Zs to take this into account. Otherwise, when you look at historicals of butterflies, you will get slightly misleading data. This used to be a much bigger issue, when the turn was 2-6+ bps. But even now, you should take this into consideration.
Curve AdvisorKeymasterFebruary 10, 2015 at 3:34 pmPost count: 612
Viewer Mail: Do you look at TED spreads (Treasury – Eurodollar)?
I do not look at TED. FUTURES ARE A ZERO SUM GAME (or more accurately a negative sum game, if you include transactions costs), and I believe most people have zero edge trading note/bond futures. Two of the quants I used to work with (I sat with a few), traded bond basis, and they would build these models, keep track of the deliverable basket, look at issue size, repo, etc. So basically, they would know every single detail about the future, and all related securities/futures (swaps, STIRs, Treasuries, etc). So when I think about what my “edge” would be over these players in the market, I can honestly say it is zero (and possibly negative). In fact, even those guys I sat with, I would say were at a slight disadvantage to the markets because they did not see customer/financing flow. Not sure if you remember, but a couple of months ago, a few people in the markets mispriced the new bond future because they assumed the deliverable basket was the same (it was not), so I think the new bond future contract traded like 10 points off of there it should have been. And that exemplifies why I stay away from note/bond futures.
Also, the mechanics of TED are so simple, I find it hard to believe a dozen banks/prop shops haven’t quantified everything and put out algos to trade it efficiently, so that would be the second reason to think my edge was zero. Because if I was one of the above guys (and there must be hundreds of people on the Street qualified to do this), this is what I would be doing. I do not think someone who trades the TED manually can beat a machine looking at the same thing.
I just stick to things where I have an edge. My main point when trading futures is… if you have no edge, you can’t make money in a zero sum game.
DINESH SINGLAParticipantApril 28, 2015 at 1:53 amPost count: 4
I found your articles very interesting which you posted on CME website.
In “curvature trading” you showed a scatter plot between a butterfly and outright. If I have found a trade in which the current price is a outlier and want to trade it, what should be the ratio of outright( pack) and butterfly and how do we calculate it.
Curve AdvisorKeymasterApril 28, 2015 at 9:54 amPost count: 612
It really depends on the fly and your view. Without knowing what the structure is, it’s hard for me to give specific comments on it. Let me start a new thread with some general comments. Here is the link: http://www.curveadvisor.com/forums/topic/to-hedge-or-not-to-hedge-a-curve-trade/
DINESH SINGLAParticipantMay 7, 2015 at 10:41 amPost count: 4
I spotted a trade in Eurodollar markets
buy 357 fly and sell 579 fly.
The difference is at -12 bps.
How should I assess that this trade has a value?
What should be the strategy if it goes against from here ? ( Should I build up the position or should I halve it or cut stop loss )
If the trade doesn’t convince you what are the reasons behind it ?
Or is there any hedge against this trade ?
Curve AdvisorKeymasterMay 7, 2015 at 12:37 pmPost count: 612
Here are some basic answers to your questions:
Q: How should I assess that this trade has a value?
A: The first thing you need to do is to turn-adjust the prices. There is a thread called “The Z turn” that discusses this in more detail. Let’s assume the turn is 0.75 bps. The price on that double fly would go from -12 to -9 turn-adjusted (since you have four Z turns in the structure). I’m not sure if you selected this double fly because this is the point of greatest “kink” on the unadjusted double fly curve. Adjusting for the Z turn will tend to smooth out the graph, so that point won’t look so extreme.
There is a thread called “Basics: How Rich or Cheap” that talks some basic ways of determining value. I don’t want to go over the entire thread, but here are some basic things you can look at:
* on a constant maturity basis, I see the 3 month range as being 5bps higher and 0.8bps lower, and in the 5th percentile of observations. The 6 month range 5bps higher and 3.3bps lower, and in the 37th percentile.
* I have enclosed a 3 month scatter plot. It does not seem to trade directionally (it would depend on the market environment). So the main reasons to do the trade would be if you wanted to trade the 3 month range, or if you had a view on the curve shape or Fed timing.
This part of the curve tends to have more volatility, so you should do the trade only if it fits your view (unless the historicals are compelling).
[to be continued]
DINESH SINGLAParticipantMay 8, 2015 at 6:09 amPost count: 4
In addition to the above trade i liked these two as well..
1. ED357 BUY@-0.5 with dec15dec16 sell@90 Logic being if so many hikes are priced in dec15dec16 then shouldn’t the ED357 fly be up in Positive.
2. EDcm3 sell 3lots and EDcm7 buy 2 lots( not very convinced about this ratio though).
Plz tell how would you have approached the two ideas if you had the same ideas in mind.
Also plz continue the above reply to the prev query..thanks a lot
Curve AdvisorKeymasterMay 8, 2015 at 2:18 pmPost count: 612
Q: What should be the strategy if it goes against from here ? (Should I build up the position or should I halve it or cut stop loss )
If the trade doesn’t convince you what are the reasons behind it ?
Or is there any hedge against this trade ?
A: There are some range-bound trades that I would be very reluctant to stop out of. However, trades centered in the whites and reds can be volatile and so you should always be thinking about the conditions where you should stop out. For example, if the Fed were to imply they would be on hold until next spring, the ED3-5-7 fly could go severely negative. The one year low on the single fly is -22.6, and it is possible for the ED5-7-9 fly to lag severely behind. And in the scenario I described, the fly may go even lower than that. There are also “lesser” ways something similar could happen – like the modal forecast move from a Sept liftoff to a liftoff next summer.
As for adding, there is a thread called “Scaling into Trades.” Because of the aforementioned volatility in this type of fly, some additional care needs to be used in scaling into this type of trade. I can’t really comment in general on halving the position or stopping out, as it will vary depending on the situation.
There is a misconception that with relative value trades, you pick a historical low and you just ride it out, because a larger move would be a “multi-sigma” event. No. You must always be thinking about the scenarios where the trade would no longer be attractive and when the historicals don’t make sense. The trade has not been directional recently, so I’m not sure why you would want to hedge. It is possible for the trade to become directional, so if that is your future view, a directional (or slope) hedge could make sense.
Let me know if you have any follow-up questions. I’ll answer the questions in your second post next.
Curve AdvisorKeymasterMay 11, 2015 at 7:05 pmPost count: 612
Q: Comment on: ED357 BUY@-0.5 with dec15dec16 sell@90 Logic being if so many hikes are priced in dec15dec16 then shouldn’t the ED357 fly be up in Positive.
A: I’ve enclosed the 1 year and 3 month scatter plots of ED3 6mo fly. I have the following comments:
* If you look carefully, there are two regimes in the plot – there is one set of dots that seem to be hovering above 0, and there is a second set of dots that seem to be trading like a bearish trade (rising on a selloff and going lower on a rally). Since the December ’14 FOMC, when the Fed effectively removed “considerable time” (and inserted “patient”), the markets have been anticipating a nearer-term liftoff. So the flies have been reluctant to go noticeably negative. You can see that prior to Dec ’14, the points were very negative (even below -20). So while the 1 year plot may not look attractive, buying the fly below zero does make some sense in the current environment (where “patient” was removed) to think that the fly should stay bid.
* I think we can ignore the dots before Dec ’14. In addition, some of the dots before the Fed removed “patient” (March ’15 meeting) may not be applicable. So as an approximation, let’s look at the 3 month chart as our “relevant” historical comparison. I don’t think it’s “wrong” to throw in a month+ of data from before “patient” was removed, since the data is a bit weaker now (so there is a small chance the current trading regime may change). It’s not great to lose 3/4 of our sample dots, but I’d much prefer to look at fewer relevant dots than a bunch of irrelevant dots. As you probably suspected, the ED3-5-7 fly still has a directional bias (although not as great as before).
* Just based on historicals (the 3 mo chart), we are about 2 bps from “fair value” (the fitted line), so there is some value here (assuming the historicals corroborate your view). But the dots do have a bearish lean to them.
* The next question is whether to hedge or not to hedge for directionality (or slope). If you are bearish, you should just buy the fly outright. If we sell off, then great. If we rally, there’s a reasonable chance we scratch the trade. That’s basically what you are looking for when doing a non-linear trade. Of course, you may prefer to just make a smaller amount of money on both a selloff and even a rally – just look for the dot to get back to the line, regardless of what happens in the markets.
[Next: how best to hedge]
Curve AdvisorKeymasterMay 12, 2015 at 1:19 pmPost count: 612
So how best to hedge?
There are two basic ways you can hedge a fly trade:
* Direction. You can buy/sell ED5 (or any other contract that makes more sense).
* Slope. You can buy/sell ED3-7 spread (or any other spread that makes more sense).
The two are similar (as this far up the curve, the direction and slope will move together. Which is best? It will vary case by case. For this particular trade, if you look at the 3mo scatter plot of ED3-7 vs ED5 (below), it does appear that the slope is high historically when compared to ED5. So ED3-7 looks like a better sell for our trade historically. More importantly, as your intuition suggests, right now selling ED3-7 spread makes more sense, because a lot of hikes got priced out of EDZ5, relative to the slope. Note that by selling 8% of Z5-Z6, you will be left with the Z5-M6 spread vs M6-Z6 spread in a 92:108 ratio.
So how much EDZ5-Z6 spread to sell? The 3mo historial beta on the simple regression is 8%. So that seems somewhat reasonable. Choosing a weighting for a hedge is a lot more art than science, because you need to be forward-looking. You can use all the sophisticated analytics you want, but at the end of the day, the past is the past and the future is the future.
Curve AdvisorKeymasterMay 13, 2015 at 10:15 pmPost count: 612
There are a number of other things you should consider before doing a trade like this:
* This trade is in the most volatile part of the curve. This can be good, if you are looking for a “choppy” trade you can range trade. But this can also lead to P&L volatility.
* As you can see from the 1 year scatter plot, an unfavorable regime change where the Fed is perceived to be on hold for a year could be devastating for this trade. However, it is also possible for us to get an inflation situation (aggressive bear flattening) where the trade breaks out of the range to the upside.
* This trade has unfavorable “rolldown.” If we were to be exactly in our current environment 3 months from now, the trade will be around 5bps lower (depending on which day you looked). That’s a lot of negative rolldown to capture 2bps of “value” on the trade. Keep in mind that the historicals are static (where the contracts are a certain number of days to expiry), but we will roll down the curve every day.
* This is not what I would consider a “rangbound” trade. The scatter plot looks rangebound. However you need to consider that the contracts will roll to a different part of the curve. The chart that looks amazing now will not be the same part of the curve 3 months from now. Below is what the same trade looks like 3 months closer to expiry (Buy ED2-4-6 fly vs Sell 8% ED2-6 spread), assuming the curve remains the same. As you can see, this chart does not look very balanced. Note that the dots are now about 5bps lower than before.
So when you factor in all of the above, the trade does not look particularly attractive. However, it may be attractive if it strongly corroborates your view. Whether and how you scale into or stop out of something like this will again depend on your view and your reasons for doing the trade.
Curve AdvisorKeymasterMay 19, 2015 at 12:18 pmPost count: 612
Q: EDcm3 sell 3lots and EDcm7 buy 2 lots( not very convinced about this ratio though).
A. There are two ways you can think about this trade:
* As a weighted (calendar) spread trade. Since ED7 should move more than ED3, it makes sense to sell (or buy) more ED3 than you buy (or sell) ED7. In the past 3 months, it looks like ED7 moves about 3.5x as much as ED3. So structuring as a 3:2 would severely underweight ED3. However, doing the trade as a 3.5:1 seems a little scary, but depending on your analysis of the scenarios, it may not be.
* As a weighted fly trade. If there was an ED0 (say cash libor), you basically have a weighted ED0-ED3 spread vs a ED3-ED7 spread trade. This comparison would make more sense if the trade was further out the curve (say, the ED0-ED7-ED11 broken fly). Teh analysis is similar, but you can look at the trade in “fly buckets.” and see if that is the curvature exposure you wanted.
In the following posts, I will discuss some of the things I will look at when contemplating a trade like this.
Curve AdvisorKeymasterMay 20, 2015 at 1:40 pmPost count: 612
The first thing one should note is that in the front of the curve can be extremely volatile. And if the STIR is in a market with an active central bank, genuine “relative value” trades present themselves less frequently (but they are there) than “directional” trades. However, this does not mean that there won’t be structured trades that offer a lot of value.
Here is a 1 year scatter plot of 1.5x ED3 vs ED7 (the preliminary stab at the hedge), and a scatter plot of 3.5x ED3 vs ED7 (what the 3 month historicals would imply). Here are the things to note:
* As with the previous example of the ED3-5-7 fly, there are clusters of dots representing various trading “regimes”. The lower set of dots in both plots are the more recent (and more relevant) dots.
* As you can see, the 3.5:1 structure utilizes the historical data better, so it looks more “balanced.” This is an artifact of using the data to come up with the hedge ratio. So of course, the hedged chart is going to look more “balanced.”
* In neither of these historical plots would the conclusion be that you want to aggressively SELL the structure (even if you only look at the more recent bottom dots). So for you to want to do this trade, you would need to have a STRONG view that overrides what the historicals are telling you.
This analysis may seem counterintuitive to what is going on in the economy. Because the Fed removed “patient” in March and ED3 seems unusually high all things considered. And that is the problem with the (longer term) historicals – they are mostly from a time period where “considerable time” and “patient” were in the statement. So those historicals are going to have ED3 much higher (Fed on hold) than in the current environment.
Curve AdvisorKeymasterMay 21, 2015 at 1:05 pmPost count: 612
Let’s just look at the three month scatter plot of the 3.5:1 ratio. I think EDU is a bit on the rich side – the September meeting is priced at under 30% of a 25bp move (note that there is still a fractional amount in the June and July meetings). Just for the sake of discussion, let’s say this was your view. So I would probably want to err on the side, of being short more EDU, rather than less (like the 1.5). I think there might be a small amount of value in doing the trade here (since the structure is towards the top of the 3mo rage).
You should weigh how strong the historical data has been for the trade (in this case it’s borderline). Ideally, you would want a larger sample than 3 months, and there are two points that are 8bps higher. So you would need a bit of a view to want to do this trade. Is our view that strong?
But then you also should consider rolldown – how the trade would fare if three months were to pass and nothing changes. As the 3 month scatter plot of 3.5x ED2 vs ED6 shows, the trade starts getting really ugly. We would have to overcome 38bps of negative roll. That’s almost 3bps a week. So you would need to have a REALLY strong view. I’m not sure that I do. Also note that the trade again becomes “unbalanced” – because our hedge was for a constant maturity ED3 vs constant maturity ED7, and not the constant maturity ED2 vs constant maturity ED6 that this rolled down to.
Curve AdvisorKeymasterAugust 27, 2015 at 11:45 amPost count: 612
I typically don’t follow swap spreads, because it is my belief that you should focus on products where you are not at a significant information disadvantage. You will probably be at a slight information disadvantage (to varying degrees) with most products. Trading most interest rate products are a zero-sum (or negative sum) game. Considering the market-makers at the big banks all see the big flows, and I don’t find out until much later (if at all), any skill I may have with respect to spreads will be completely wiped out by the information disadvantage. Getting to your questions…
I think a lot of the relative treasury selloff to swaps was from some central banks selling their holdings (China, Japan), of US securities. I believe this has been going on for months now. I don’t think it helps that the Fed is insistent on hiking with no inflation, which is going to put general downward pressure on long rates at the same time.
As for how to anticipate such a move, it helps to see the flows. The sales team at the bank you do the swaps with should be sending you regular color on the big flows. Other than that, I think just thinking about what is going to cause a change in the relative supply/demand of swaps vs Treasuries will help. Call me Captain Obvious. But again, this is not my forte, so my advice is very general.
I asked a few people who would know more than I, so if I get any meaningfully different replies, I will add it here.
bambamParticipantOctober 8, 2015 at 9:24 pmPost count: 24
I was wondering how you all found this thread? I had “googled” for years a place to talk about curve and flies yet I couldn’t find anything… YEARS!!!!
While discussing a “trade the down curve on BAs” with a friend I thought about googling “what happens to the curve on a hike cycle and bam! here I am!.
First thing that caught my attention CA, so much curve! why do you barely mention and use 3M flies?
What is it that drives you more toward longer term structures 6m 12m flies?
Curve AdvisorKeymasterOctober 9, 2015 at 2:57 amPost count: 612
I suppose since I do no advertising, the only way people can find the Forum is via word of mouth and from the three articles I wrote for the Chicago Mercantile Exchange.
Right now the curves are unusually flat, so there isn’t as much “juice” as there could be. Also, you need a very low cost structure to trade 3 month flies (the larger flies are more brokerage-efficient, all things being equal). I think when there is more activity, we can look at 3 month flies more. But most of the things in the Forum are fairly general, so they should will also apply to 3mo flies. If there is anything in particular related to 3mo flies you want to discuss, let me know.
bambamParticipantOctober 11, 2015 at 6:28 pmPost count: 24
You are on point with the low fee structure.
The move from full ticks to half ticks in BAs was a game change for me, insane actually.
Success rates unchanged, automated strategies just died.
“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).”
I think this explains why the 4-5th BA 3MF have been rolling negative.
presently the 5th trades 0bid 1 offer, as the curve is bit concave BA2-BA3 and then flattens.
So shorting the 5th fly 0 is a good strategy if I think that the odds for the bank to start changing their regime in q1-16.
as a higher probability of a hike would be priced into the 1st leg of the fly.
This trade has been golden since feb. Yet I had failed to understand the why.
Curve AdvisorKeymasterOctober 12, 2015 at 5:46 pmPost count: 612
I just want to check nomenclature, as I have found that different people use different expressions to refer to the same thing:
* When you say the “5th fly”, you are referring to BAZ6-H7-M7?
* Just to check that we are calculating the prices the same way, this fly closed at -1 the past 2 days?
* I am used to “shorting a fly” refer to selling the first and last legs, in this case selling Z6, buying 2x H7 and selling M7. But it sounds like you are saying you want to “sell” the belly? (which I would call “buy” this fly) * I am not sure if you meant Q1 2017 (instead of 2016)… since we are talking about the fifth fly.
* I usually don’t look at BAs. Do you have a historical chart of this fly readily available? Otherwise, give me a few days to download the data and look at it.
Thanks for the clarifications.
bambamParticipantOctober 12, 2015 at 7:08 pmPost count: 24
“When you say the “5th fly”, you are referring to BAZ6-H7-M7?”
Yes, fifth contract is Z6, so the 5th 3month fly is Z6 2H7 M7.
“* Just to check that we are calculating the prices the same way, this fly closed at -1 the past 2 days?”
“I am used to “shorting a fly” refer to selling the first and last legs, in this case selling Z6, buying 2x H7 and selling M7. But it sounds like you are saying you want to “sell” the belly? (which I would call “buy” this fly) * I am not sure if you meant Q1 2017 (instead of 2016)… since we are talking about the fifth fly.”
When I say Short the fly, i am short the first contract, so you would call it “long” the belly. So when you say on any other post, long you mean long the belly? correct?
I meant Q1 16.
BAs only trade until the greens, the rest is dead, and even the last two greens are mainly just market makers, very low liquidity.
Which is why most of the expectations are priced into whites and reds,
I will see if I can finish some work and upload a constant maturity chart on that fly.
Curve AdvisorKeymasterOctober 13, 2015 at 8:51 amPost count: 612
Thanks for the clarifications. Maybe my coffee hasn’t kicked in yet, but if you think the BOC becomes more hawkish starting in a few months (which I assume is what you meant, if you expect more hikes to be priced into the front of the fly), I’m not sure why you would want to sell the fly. Unless you are expecting a bull steepening (a steepening of the curve on a rally).
I haven’t been following the BAs, but would imagine this fly would probably have kept drifting lower for most of the past year? Last year, when the Canadian unemployment rate was drifting lower and Fed was in the midst of tapering, the curve was probably very steep (and the 5th fly was pretty high). As the economy shows weakness and Europe started to engage in QE, the curve probably flattened a bunch and the fly started going lower. In the US, the flies centered in the reds and greens have been uncharacteristically low, considering the Fed is about to liftoff. The markets are pricing in a severe degree of a slow hiking cycle, so the flies have been flatter than usual (there is not a material difference in the amount priced into the front and back legs of the fly). It’s funny how every 3mo spread appears to be around 7 bps (which means every fly is around zero).
[I’m going to move the last few posts into a new BAX thread]
Curve AdvisorKeymasterMay 12, 2016 at 11:25 amPost count: 612
I hear Bloomberg offers this feature (using CIXX), but not sure if you want to spend $2K a month on a Bloomberg terminal (and I’m not sure if it offers all features). I would think the simplest/cheapest thing to do would be to just get an add-in for whatever software you used to generate the data. Assuming you used Excel, you can Google “excel add in technical analysis review”. It looks like there are a few that are either free or $49, which seems reasonable. I haven’t used any of them, so I can’t say which is good. If you used Matlab or some other statistical package, I would assume you should be able to get technical analysis on there somehow. If there is one you end up liking, feel free to write a review for the Forum. 🙂
It’s funny you mentioned technical analysis because my next “Philosophy” post mentions why I don’t think technical analysis as a stand-alone can be an edge in your trading. But hey! It’s not like I have never been wrong before. I will put the link here when it is posted.
amitParticipantMay 12, 2016 at 11:55 pmPost count: 5
thanks for the quick reply.
I dont use technical analysis as the only tool for trading. It just helps me for good entry and exit points for putting in my view trades.
Curve AdvisorKeymasterMay 13, 2016 at 8:22 pmPost count: 612
I usually write a few posts ahead of time, and I forgot I posted a few thoughts on technical analysis already: http://www.curveadvisor.com/2016/05/taking-inventory/ Good luck with your project.
Curve AdvisorKeymasterJuly 27, 2016 at 11:02 amPost count: 612
I read mostly “primary” sources of information (Wall St Journal, Bloomberg, Fed speeches, etc) to get the facts. I’ll generally have my own views on what is going on, based on the evidence. There are only a few “secondary” sites I look at:
* Ben Hunt: http://www.salientpartners.com/epsilon-theory/ It’s a little egg-headed, but sometimes he’ll say something interesting.
* Zerohedge.com: I used to look at this more often, but all the ads on their site make it less readable.
* I like SeekingAlpha.com’s morning “Wall St Breakfast” email – just a few bullets on the economy (and companies).
A friend of mine likes MacroMan (https://macro-man.blogspot.com/). I think it’s pretty good… but not enough to look at it regularly, but maybe I’ll start.
Curve AdvisorKeymasterFebruary 21, 2017 at 1:54 pmPost count: 612
Depending on how you prefer to trade, you can have bids and offers in a bunch of structures. In that way, you are making markets – you are adding to the bids and offers and possibly even putting in bids or offers when there were none. Many of the marketmakers who used to work in the CME Eurodollar pit now work in offices making markets electronically, and this is effectively what they do – both manually and automatically (via an algo).
I did not make markets for clients at JPM. However, because I was so active, I probably was on the other side if some of the larger Eurodollar trades that went on in the markets.
You must be logged in to reply to this topic.