Home Forums Main Forum Basics: What is priced into each Fed meeting?

  • Author
    Posts
  • Curve Advisor
    Keymaster
    Post count: 612
    #1019 |

    After you feel comfortable with trading further out the curve, you should feel comfortable looking at what is priced into the very short-end futures: ED serials (non-quarterly months) and Fed Funds futures, and most importantly what Fed meetings are priced into each ED contract.

    It always amazes me when people who trade EDs for years have no idea what Fed meetings they are getting long or short. This is not as important when you trade further out the curve, but this can be critical for trades within 2 years. Just as a self-quiz (you can approximate to the nearest half meeting):
    * when you buy EDM5, which FOMC meetings do you have a bullish rates position on?
    * when you buy EDM5-U5 spread, which FOMC meetings do you have a bullish rates position on?
    * When you buy EDM5-U5-Z5 fly, which FOMC meetings do you have a position on? Which meetings do you have a bullish rates position on, and which meetings do you have a bearish rates position on?

    [cue Jeopardy music… to be continued]

  • Curve Advisor
    Keymaster
    Post count: 612

    The answers to the above questions are:

    Q1: When you buy EDM5, which FOMC meetings do you have a bullish rates position on?
    A1: You are long rates for all meetings up to and including the June 2015 meeting, plus 54% of the July 2015 meeting.

    Q2: When you buy EDM5-U5 spread, which FOMC meetings do you have a bullish rates position on?
    A2: You have NO bullish rates position for any meeting. You only have a bearish rates position for: 46% of the July 2015 meeting, and bearish rates for 99% of the Sept 2015 meeting and bearish rates for 54% of the Oct 2015 meeting. You are basically short rates for 2 full Fed meetings.

    Q3: When you buy EDM5-U5-Z5 fly, which FOMC meetings do you have a position on? Which meetings do you have a bullish rates position on, and which meetings do you have a bearish rates position on?
    A3: You have a bearish rates position for 46% of July 2015 + 98% of Sept + 8% of Oct. You have a bullish rates position on 100% of the Dec 2015 meeting and 54% of the Jan 2016 meeting. You are basically short rates for 1.5 Fed meetings vs long rates for 1.5 Fed meetings. Note that while the spread was a position on 2 Fed meetings, the fly cancels out half a meeting for each spread, so you have a position on 1.5 vs 1.5 meetings.

  • Curve Advisor
    Keymaster
    Post count: 612

    If you didn’t get all three correct – don’t worry. I would say most people – even those who have traded EDs and ED flies for years – haven’t really thought about exactly which meetings are involved in each fly. When you are further out the curve, it is not as important to pay attention to those small details. However, as you get (closer) into the whites, the details become a lot more important. A hike is typically on the order of 25bps. So being just 20% off on the number of the Fed meetings is going to be 5bps. That may exceed your profit target on the trade!

    On a side note, I will start a new thread on some other items on the ED curve that you can ignore further out the curve, but you should start paying attention to within 1 year (and sometimes 2 years).

    [coming next… how to fish]

  • Curve Advisor
    Keymaster
    Post count: 612

    Now that the Fed is close to removing “patient”, you are going to want to start looking more at the front end of the curve. If you plan on trading the front part of the curve, you should know *exactly* which meetings you are going long or short. The process is fairly simple – you can do a rough approximation, or you can create a spreadsheet to calculate the percentages exactly.

    I. APPROXIMATION
    In recent years, the FOMC appears to have decided to have the second day of their “quarterly” meetings on the IMM dates (third Wednesday of Mar, Jun, Sep, Dec). This was not always the case. But this makes valuing ED contract much simpler. So in general:
    * EDH will price 100% of the March meeting and approximately 50% of the April meeting. If you buy an EDH contract, you will be long rates for all meetings through and including the March meeting, and also long 50% of the April meeting. This is because the April meeting occurs about midway through the 3 month period covered by the EDH future. So if you are long EDH and the April meeting goes from being 0% priced to 50% priced for a 25bp hike, you will lose 12.5bps (as an approximation). Note that EDH expires before the March and April meetings, so what matters is how many bps are priced into those meetings at the time of the EDH contract expiry.
    * EDM will price 100% of the June meeting and approximately 50% of the July meeting.
    * EDU will price 100% of the Sept meeting and approximately 50% of the Oct meeting. As an aside, the Sept meeting this year is on Wednesday & Thursday, so EDU will have slightly less than 100% – know the details! But this is the “approximation” section so, 100% is close enough.
    * EDZ will price 100% of the Dec meeting and approximately 50% of the Jan meeting.

    [Next time: Calculating exactly]

  • Curve Advisor
    Keymaster
    Post count: 612

    II. CALCULATING
    Estimating has it’s place, and you’ll find that it comes in very handy when you want a quick sanity check. However, there may also be times when you want to calculate the numbers exactly. After all, just 4% of a Fed meeting is 1bp, and that could be the difference between a decent trade and a good trade.

    It’s actually quite simple:
    * Get the Fed meeting dates. For meetings covering multiple days, use the day where the decision is made (usually the last day).
    * Determine the interest rate period that your futures contract covers. In the case of ED, it will just be the 3 months starting with the IMM date. In the case of FF, it will just be the 1 month starting with the first of the month.
    * To calculate what percent of a Fed meeting a contract contains,
    – it will be 100% if the meeting occurs before the first day of the contract,
    – it will be 0% if the meeting occurs after the last day of the contract,
    – use linear interpolation for find the percentage in between. If the Fed meeting occurs 1/4 of the way through the contract, the contract will have 75% of that meeting. If the Fed meeting occurs 3/4 of the way through the contract, the contract will have 25% of that meeting, etc.

    [next: accounting for the day of the Fed meeting]

  • Curve Advisor
    Keymaster
    Post count: 612

    The rate to use for the day of the FOMC decision can be tricky. Back in the 2004-6 hiking cycle, the day of the Fed meeting was usually priced closer to the new rate. But part of this may have been because the Fed was “measured” – so the markets were mostly priced for the new rate, because the Fed just hiked 25bps every meeting. However, when creating a spreadsheet, I believe in making the spreadsheet as flexible as possible. So have a cell where you can enter a fraction between 0 and 1, that accounts for the day of the Fed meeting. This way, you have the flexibility of being able to “tweak” your spreadsheet as you see fit.

    [at some point, I will go back to the last hiking and easing cycles and post what the FF effective rate was for those days]
    [next: how to set up a “useful” spreadsheet]

  • Curve Advisor
    Keymaster
    Post count: 612

    Before you start setting up a spreadsheet, you should always think about how you are going to use the spreadsheet. It’s nice to have a table that tells you what is priced into each Fed meeting, for each contract/security you trade. However, what would really be nice is some way you can construct trades and have the spreadsheet tell you what meetings you are long or short. Whenever I think about various structures on the front of the curve, I always want to make sure that I double-check the Fed meeting exposure of the trade – because those are the points where the structure can move the most.

    The other thing you should always do with your spreadsheets is that you should always have it in a form that requires minimal updating. Preferably NO updating. You may not look at a spreadsheet for a while. But when you do, you want to just open the spreadsheet and go. You shouldn’t have to tweak a bunch of formulas to get the thing running.

    I generally like people to make up their own layout, so think of how you would set up a spreadsheet to do all of the above. Since this spreadsheet is harder to describe, I will post a screenshot of a sample basic sheet and discuss the details and features next time.

  • Curve Advisor
    Keymaster
    Post count: 612

    Enclosed is a screenshot of a section of a basic Fed Meeting worksheet. The rows contain different futures contracts. The columns contain the fraction of each Fed meeting contained in each contract (where 0 = no exposure, and 1 = 100% exposure). One way to think about interest rates is as a function of Fed move probabilities at each meeting. The Federal Reserve controls short term interest rates by adjusting the Fed Funds rate. It is possible for the Fed to change rates in between meetings, but this is rare (except in emergency situations). As the probabilities of a change in rates at a meeting increase or decrease, the rates implied in the interest rate futures change as well. That is why it is important to understand how many meetings (and which particular meetings) are imbedded in a futures contract.

    The table of meeting fractions are a useful instructional tool. However, the real value of this spreadsheet is in what you can do with the table. In the bottom section, you can set up an area where you can construct trades, to take a view on the various meetings. In the hypothetical example, we are buying 2 FFQ5 contracts vs selling 1 FFV contract. This may be appealing if you thought the first Fed move was going to be in September. Note that you are getting long rates for all meetings up to and including the July meeting (designated by a “+1” in the blue “total” area), and short rates for the Sept meeting and 13% of the Oct meeting (designated by a “-1” and “-0.13” in the blue “total” area). I’m not saying this is a great trade, but I hope you can see how having this kind of tool can help you express specific views you may have on rates and Fed meetings. It helps you to dissect your views and trades with precision.

    Note: You should be setting up the sheet to auto-calc. The only cells you should need to change are the light yellow ones – everything else calculates itself, even when the dates change.

    [next 1: some interesting results from the sheet]
    [next 2: looking at FF and adjustments]

    Attachments:
    You must be logged in to view attached files.
  • Curve Advisor
    Keymaster
    Post count: 612

    You can use the above tables to answer the questions I had in the first post in this thread (see attachment for the M5-U5-Z5 fly). You can type in the quantities associated with the M5-U5-Z5 fly, and see *how* the numbers I presented as the answer are correct. More importantly, you can develop the building blocks to analyze other front-end trades.

    Here are some interesting things to note about this example:
    * The net number of meetings you are long or short on the 3 month fly is close to zero, but not exactly zero. This is mostly a function of the FOMC meeting dates in this example. But when we get to late-2016 to 2017, this will be more important to pay attention to, as the number of days between IMM dates will vary from 91.
    * Now that the FOMC has removed “patient”, they can presumably hike at any meeting. You may have a view on which meetings the Fed is more likely to hike on. For example, “the first hike is going to be on a quarterly meeting, but after the first hike, all other meetings are equally likely.” In that case, the 46% of the July meeting that the M5-U5-Z5 fly has may not be relevant, while the 54% of the Jan meeting may still be relevant.
    * While a 3 month spread is a position on 2 total meetings, a half meeting from each spread (M-U and U-Z) get cancelled out when the two spreads are combined into a fly. So the meeting in the middle of the fly (in this case Oct), has close to zero exposure.

    You can make similar observations on other structures.

    Attachments:
    You must be logged in to view attached files.
  • Curve Advisor
    Keymaster
    Post count: 612

    You can also use the above tables to take spread views between ED and Fed Funds futures (FF). You can replicate the FOMC meeting exposure of the ED contract using FF, so that you cancel out all of the Fed meeting exposure, and are left with just the Libor to Fed Funds spread. See the attachment for an example. This may also allow you to capture any value you see in a particular FF (or ED) future. The way the meetings are scheduled during the year, most of the time, ED(X) (where “X” is a quarterly month) can be *approximately* replicated by 50% DV01 of FF(X+1) and 50% of FF(X+2). In our example, 10 EDM contracts is replicated by 3 FFN and 3 FFQ contracts. Note that you only need 6 FF contracts to hedge 10 ED contracts, because of how they are defined (1bp in EDs is $25, but 1bp in FF is $41.67). You can adjust what the “quantity” units are in your spreadsheet to reflect what makes most intuitive sense to you.

    The more tools you have in your toolbox, the more potential ways you have of making money.

    [Next: Looking at FF contracts]

    Attachments:
    You must be logged in to view attached files.
  • Curve Advisor
    Keymaster
    Post count: 612

    One colleague of mine who sat next to me traded a lot of FF futures, so I was able to learn a lot about how the futures were priced.

    From the NY Fed web site: “By trading government securities, the New York Fed affects the federal funds rate, which is the interest rate at which depository institutions lend balances to each other overnight. The Federal Open Market Committee establishes the target rate for trading in the federal funds market.” Below is a table from the NY Fed web site (http://www.newyorkfed.org/markets/omo/dmm/fedfundsdata.cfm). The important columns are the date, and the daily effective rate. You can also see what the target rate is on the far-right column. The FF futures settlement is the average of the FF daily rate over each of the days of the month. For weekends and holidays, the prior business day’s rate is used. If you take the average of the daily rates over all days, and assume 0.11 for 3/30 and 0.06 for 3/31, you get 0.112 as the average FF rate for the month (rounded to the nearest tenth of a bp, as per CME final settlement convention). FFH5 is currently trading 99.8875 (trades in 0.25bp increments), so this is fairly close to the projected rate. I will discuss how to model this in a separate post – I would prefer not to get into minutae right now (even though it is important).

    Attachments:
    You must be logged in to view attached files.
    • Mike G
      Participant
      Post count: 20

      <cite>@Curve Advisor said:</cite>
      One colleague of mine who sat next to me traded a lot of FF futures, so I was able to learn a lot about how the futures were priced.

      From the NY Fed web site: “By trading government securities, the New York Fed affects the federal funds rate, which is the interest rate at which depository institutions lend balances to each other overnight. The Federal Open Market Committee establishes the target rate for trading in the federal funds market.” Below is a table from the NY Fed web site (http://www.newyorkfed.org/markets/omo/dmm/fedfundsdata.cfm). The important columns are the date, and the daily effective rate. You can also see what the target rate is on the far-right column. The FF futures settlement is the average of the FF daily rate over each of the days of the month. For weekends and holidays, the prior business day’s rate is used. If you take the average of the daily rates over all days, and assume 0.11 for 3/30 and 0.06 for 3/31, you get 0.112 as the average FF rate for the month (rounded to the nearest tenth of a bp, as per CME final settlement convention). FFH5 is currently trading 99.8875 (trades in 0.25bp increments), so this is fairly close to the projected rate. I will discuss how to model this in a separate post – I would prefer not to get into minutae right now (even though it is important).

      I’m really interested in the modeling approach here. Can you elaborate more on how you are using the websites effective rate to model FF futures? How do you use this information?

      Thanks
      M

  • Curve Advisor
    Keymaster
    Post count: 612

    The reason I am talking so much about FF is that in the world of futures, FF is the best way to take a specific Fed view. There used to be 1 month libor futures, but that does not seem to trade as much.

    The big disadvantage of using 3 month libor futures is that they expire a few days BEFORE the FOMC meeting. For example, EDM5 expires on June 15. The Fed meeting is on June 17. While you can hope that the result of the meeting is fully priced by futures expiry, you are always going to be leaving some money on the table (assuming your view comes to fruition). But this is especially bad if you want to express a view on a “surprise” move (where everyone is expecting one result, and you want to take a position on the opposite).

    The other disadvantage of using 3 month libor futures is that you have about a 50% exposure to the interquarterly meeting. For example, EDM5 has a 54% exposure to the July meeting. So you could have some additional noise. While we are close to 50-50 for the June meeting, the July meeting will probably move in the same direction as the June. But in this particular hiking cycle, as we get closer to 0 or 25 for the June meeting, at some point, what is priced into the July meeting may trade OPPOSITE to the direction of the quarterly meeting. That is, there may be a perception that the Fed may not hike in consecutive meetings. The markets have been reluctant to significantly price in more than 100bps in a year. So as June gets to 25bps, July may start going to zero. So this is going to affect the sell-off potential of EDM5 (on a per meeting basis).

    [Next: Structuring FF trades]

  • Curve Advisor
    Keymaster
    Post count: 612

    The Fed Meeting worksheet you created earlier will help you construct various ways to take a view on a particular Fed meeting. I think this is best illustrated with an example. Think of the ways you can construct a bearish position for the June FF meeting only (ideally). Take your time. [Jeopardy music]

    Try to figure this out, before proceeding further.

    I have illustrated 4 ways, using just FF futures in the table below:
    Ex 1) You can buy the FFK5-FFM5 spread. Buying FFK5 insulates you from anything in the April FOMC meeting, and just leaves you short 47% of the June meeting. So if the Fed hikes 25bps, you will make 11.75bps. So if you wanted a structure that will pay you 25bps, you will need to do 2.13 units of the FFK-M spread. This method has some drawbacks, in particular “fixing risk” (to be discussed in a later post).
    Ex 2) You can buy the FFM5-FFN5 spread. This will get you short 53% of the June meeting, but will also get you short 10% of the July meeting. This is less attractive than Ex 1 since it has an exposure to the July meeting, but it’s possible it fits your view better. This alternative is there in case you ever need it.
    Ex 3) You can buy the FFK5-FFN5 spread. This will get you short 100% of the June meeting (interestingly, you have no position in June FF), but will also get you short 10% of the July meeting. This would probably be my favorite alternative, as it is brokerage efficient (you have to trade relatively fewer contracts) and the fewer contracts you need, the lower your “fixing risk” (to be discussed later).
    Ex 4) You can buy the FFK5-FFN5 spread, and sell 11% of the FFN5-FFQ5 spread to get rid of the exposure to the August meeting. This will get you short 100% of the June meeting. This may seem like overkill, but depending on your view, this may make sense to do.

    You can also mix both ED and FF futures to structure a position. I may discuss this later.

    [Next: Oops – what is wrong with the pricing?]

    Attachments:
    You must be logged in to view attached files.
  • Curve Advisor
    Keymaster
    Post count: 612

    The settlements from Thursday, April 2 2015 were:
    FFK5 99.875
    FFM5 99.860
    FFN5 99.835
    FFQ5 99.805

    Let’s try and figure out what is priced into the June meeting:

    If we take Ex 1 from the previous post, we get:
    FFK5-M5 spread = 1.5bps
    So if 47% of the June FOMC meeting is 1.5bps, then 100% of the June meeting = 3.19bps.

    If we take Ex 4 from the previous post, we get:
    FFK5-N5 spread is 4bps,
    FFN5-Q5 spread is 3bps,
    So FFK5-N5 spread less 10% of FFN5-Q5 spread = 100% of the June meeting = 4 – 0.11 * 3 = 3.67bps

    If we take Ex 2 from the previous post, we get:
    FFM5-N5 spread = 2.5bps
    53% of the June FOMC meeting plus 10% of the July FOMC meeting is 2.5bps, and let’s assume the July meeting is 3.3bps. Then 100% of the June meeting = (2.5 – 0.1 * 3.3)/0.53 = 4.09bps.

    If we take Ex 3 from the previous post, we get:
    FFK5-N5 spread = 4bps
    100% of the June FOMC meeting plus 10% of the July FOMC meeting is 4bps, and let’s assume the July meeting is 3.3bps. Then 100% of the June meeting = 4 – 0.1 * 3.3 = 3.67bps.

    Consider the above results and try to explain the reasons for the similarities and differences.

  • Curve Advisor
    Keymaster
    Post count: 612

    There are three potential reasons why your calculations could be off:

    * You may have had a bad settle. The contracts generally trade in 0.5bp increments, with the exception of the first contract, which trades in 0.25bp increments. So it would not be unusual for a calculation to be 0.5bps off. Since it typically takes 2 contracts to isolate a meeting, it is possible the calculation could be as much as 1bp off (if the two contracts settle very poorly relative to each other).

    * “Extrapolation” could result in an even larger error. In Ex 1 and Ex 2, you are taking a spread that could have up to 0.5 to 1 bp of “noise” and roughly doubling that error (because each of those spreads only has half a meeting, you need to divide the spread by 0.5 to get the result for a full meeting). So it is generally preferred to extrapolate as little as possible (as in Ex 3 and Ex 4, where there is 100% of the June meeting in the spread, so there is less room for extrapolation error).

    * There is a daily adjustment that needs to be made to each contract. Depending on the calendar configuration, you could have minor adjustments from contract to contract. [Later post]

  • me
    Participant
    Post count: 27

    This is an amazing post. I wonder if there’s a way to make it *sticky* / more prominent at the top.

  • Curve Advisor
    Keymaster
    Post count: 612

    Glad you enjoyed it. Very few people I’ve worked with actually look at their front-end positions this way. JPM risk management used to bucket FF futures into FUTEQ (IMM futures equivalents) buckets, but I told them they should look at Fed meeting buckets. At the time, the Fed meetings were not as regular as they are now, so you could have more variability between time to settlement and the actual meetings the futures contained. Even now, there will still be a difference between futures settlement and the number of Fed meetings contained in the front contracts. That means if you use a risk management system based on time to settlement, there could be a mismatch with the actual amount if risk you have (depending the timing of the Fed meetings).

    Also, the other thing I wanted to add to the post was for people who are used to trading something like Treasury “spreads.” When you talk about buying a 5s-10s spread, you tend to overweight the 5s to do some form of DV01 weighted spread. This results in a butterfly of sorts in futures space – you are generally long up to “5 years” and short between 5 years and 10 years. When you buy a ED5-ED9 spread, you are short the part of the curve after ED5 and through ED9. So a typical “spread” in ED space is going to trade differently than a typical “spread” in treasury space.

  • Curve Advisor
    Keymaster
    Post count: 612

    [DRAFT] When doing a trade that mixes FF and ED contracts, there is the chance that the “spread” between the two contracts can change over a week or so. If you were to construct the same Fed meeting profile between FFs and EDs, you will find that the spread is currently around 15. However, this number can fluctuate a little here and there, depending on what is going on in the libor and fed funds markets. [Give an example on a day with “fair” settlements.]

  • me
    Participant
    Post count: 27

    So going back to the example on 5 April.

    Here we assume a value for the July meeting. However, in practice, before any trade structuring can begin, we would need to figure out what that is… So, we’d basically need a table of the generic front FF spreads (FF1-FF2,FF2-FF3,FF3-FF4, etc.), to identify specific segments that cover one meeting (to minimize extrapolation) to back out how many Bps are covered in each meeting. From here we could compare this against our views, and play around with the table to get an exposure to match.

    Something like this?

  • Curve Advisor
    Keymaster
    Post count: 612

    Ugh. I clicked “detach” thinking that would download the file, but apparently it deleted it. Could you post the attachment again?

  • me
    Participant
    Post count: 27

    OK Here it is again…

    Attachments:
    You must be logged in to view attached files.
  • Curve Advisor
    Keymaster
    Post count: 612

    I mentioned previously that you want to try and minimize extrapolation. The Fed started scheduling meetings “around” quarterly IMM dates. The quarterly meetings (March, June, Sept and Dec) occur in the middle of those months. So I think it would be slightly preferable to take the two contracts surrounding the quarterly meeting date. For example, for the March meeting, take FFG-FFJ spread. This will encompass 100% of the March meeting. Unfortunately, this will also capture a small piece of the April meeting. But there is generally much less priced into the non-quarterly meetings, AND typically there will only be about 10% of the non-quarterly meeting captured in the 2 month spread around the quarterly meetings (i.e. FFG-FFJ), so the overall error will be small.

    With regard to your sheet, I think using all THREE methods to look at what is priced in is good: (1) what you have set up, (2) fill in the empty lines – there is information there (for example for FFU-FFV spread, you can back out 13% of the 3.4bps from the October meeting to calculate the September meeting, and (3) Use the FFQ-FFV spread to calculate the September meeting. It serves as a good check on your calculation, and it will make you aware of potential settlement adjustment issues for particular months.

  • Curve Advisor
    Keymaster
    Post count: 612

    There will be a spread between the Eurodollar curve (libor) and the Fed Funds curve (central bank rate). In the US, that spread in the front of the curve will typically be around 10-15bps, and will be wider further out the curve. That spread can move from day to day, just as swap spreads (swaps vs treasuries) can vary from day to day. Now that you can deconstruct a Eurodollar into its component FF probabilities, you can use this information to take a position on the spread. You can sell EDs vs its component Fed Funds if you think the spread is too low, and buy EDs vs its component FFs if you think the spread is too high.

    For example, below are the settlements from 9/1/2015:
    EDZ5 99.535
    FFF6 99.68
    FFG6 99.65

    So from your sheet (from post #1149), you can derive that EDZ5 is equivalent to 55% FFF6 and 45% FFG6 (DV01 weighted). The weighted FF price is 99.667. So the spread between EDZ and FF is 13.2bps. Note that this is a much more apples-to-apples comparison of the ED and FF curves (than comparing a single ED contract to a single FF contract), because the Fed meeting components were set to be equal.

  • Curve Advisor
    Keymaster
    Post count: 612

    The libor cash fixings can also be modeled with this spreadsheet. The fixings and the futures eventually converge, so when there are a lot of hikes priced in, the fixings would be higher, and vice versa. Depending on the country, the number of days between the cash fixing and the effective start date will be between 0 to 2 days later. In the US, the start date would be 2 days from today. You can figure out what fraction of each Fed meeting the libor fixing contains and determine the appropriate FF hedge. Once you determine the FF hedge, you calculate that weighted FF rate and take the difference to the libor fixing to calculate the spread between EDs and meeting-matched FFs. Note that the libor cash fixings can be sticky – so it may take a few days to converge. Also note that the libor cash fixings are held at a specific time in the morning, so there may be a time mismatch if you compare to the end-of-day FF closes. However, this mismatch will be minimized in quiet markets.

  • Curve Advisor
    Keymaster
    Post count: 612

    When you take a view on a meeting, you need to factor in if the contract expires before or after the meeting. If the future expires before the meeting, you need to figure out how you are going to maintain the position going into the meeting (assuming that’s what you want to do). For example, EDU5 officially settled on September 14, but the Fed meeting was on September 17. The Fed has typically picked dates for their quarterly meetings a few days after the IMM dates. This makes trading the front EDs less convenient for taking a view on the Fed meeting outcomes. EDs cover a 3 month period, which lacks precision. ED fixings can also exhibit some stickiness and noise. All of these factors make EDs a less precise instrument for trading near-term Fed outcomes.

    If you want to take more accurate Fed views, a better instrument to use would be the Fed Funds futures. FFV5, for example, would cover 100% of the September meeting and 13% of the October meeting. The FF volumes have been very low the past few years because of an inactive Fed. Look for more volume to go through in the next few years. There is some noise inherent in the day to day pricing of Fed Effective rates, but starting Jan 1, 2016, some of this noise should be reduced because of a minor calculation change (going from a volume weighted mean to a volume weighted median). There can sometimes be some drift in the level of FF fixings within the target range, but over a short holding period, the drift should not be very large. One way to reduce some of this risk would be to trade FF spreads (like FFG6-FFJ6 spread).

  • Curve Advisor
    Keymaster
    Post count: 612

    You will find that when you FOMC meeting-hedge an ED future with FF futures, you will find that you need multiple FF contracts to meeting-edge the ED contract. For example, in post #1401, you can see that EDZ5 is equivalent to 55% FFF6 and 45% FFG6 (DV01 weighted). It is generally very cumbersome to try and trade three different legs simultaneously. Here is a tip: if you have a bullish or bearish bias, you can incorporate that view into the trade. For example,

    * If you are bullish and wanted to put on a trade that would profit if the ED-FF spread narrows, you can buy EDZ5 and sell FFF6 (DV01 weighted). This would get you long 38% of the Jan 2016 meeting. This is an “interquarterly” meeting, so this may not be as active. However, this meeting will still move with the rest of the market.

    * If are bullish and wanted to put on a trade that would profit if the ED-FF spread widens, you can sell EDZ5 and buy FFG6 (DV01 weighted). This would get you long 46% of the Jan 2016 meeting.

    * If you are bearish and wanted to put on a trade that would profit if the ED-FF spread narrows, you can buy EDZ5 and sell FFG6 (DV01 weighted). This would get you long 46% of the Jan 2016 meeting. This is an “interquarterly” meeting, so this may not be as active. However, this meeting will still move with the rest of the market.

    * If are bearish and wanted to put on a trade that would profit if the ED-FF spread widens, you can sell EDZ5 and buy FFF6 (DV01 weighted). This would get you long 38% of the Jan 2016 meeting.

  • Curve Advisor
    Keymaster
    Post count: 612

    From time to time, I will publish a table of what is priced into the various Fed meetings (see example below). This is something you could easily create yourself from the information above. But first you need to decide a few things, related to simplicity vs thoroughness:

    * Use all contracts or some? To calculate what is priced into the 8 Fed meetings a year, you only need 8 contracts. But you can use 13 if you want. The additional information in the additional 4 contracts may only be marginally useful.

    * Use daily adjustments or no? You can model the daily behavior of FFs, as per post #1174. I haven’t written the follow-up post because the method of calculating the FF effective rate will change some time early next year.

    Using more information is a little more accurate. However, in most cases, this will not vary the results by more than 1bp or so (and many times it will be much less). So I will mainly discuss the “simple” method of calculating the table.

    Attachments:
    You must be logged in to view attached files.
  • Curve Advisor
    Keymaster
    Post count: 612

    The first thing to note about the Fed calendar is that the “quarterly” meetings (Mar, Jun, Sep, Dec) typically occur on IMM dates (the third Wednesday of the quarter). There can be exceptions (the Sep 2015 meeting was on a Thursday). The “non-quarterly” meetings occur typically on the last Wednesday of the month following the quarterly months. There can be exceptions (next year, the meeting between Sep and Dec occurs in early Nov, and not late Oct). It’s possible for a central bank to decide on a different number or configuration of monetary policy meetings a year. Alter your methodology as needed. The basic steps are as follows:

    STEP 1: Determine the meetings with the “cleanest” calculation. With the Fed, the cleanest meetings to calculate are the non-quarterlies. You just take the [spread of the two futures between the quarterly futures] and divide by [the fraction of the meeting in the spread]. For example, to calculate the Jan meeting, you take FFF6-FFG6 and divide by 0.84 (which is the amount of the Jan meeting in the FFF6-FFG6 spread). This calculation involves the fewest contracts possible (two) and the divisor is as large as possible. Do a similar calculation for the other non-quarterly meetings.

    You now have half of the meetings. Think about the “cleanest” way to calculate the other half of the meetings. Hint, you may need to use the information from Step 1. [cue Jeopardy music]

  • Curve Advisor
    Keymaster
    Post count: 612

    STEP 2: Determine the quarterly meetings. In my opinion, the best method is to take the [spread of the two futures surrounding the quarterly month] and subtract [the portion of an non-quarterly meeting in the previous spread]. For example, to calculate the March meeting, you take FFH6-FFJ6 and subtract 11.7% of the April meeting (that you calculated in the previous step). We use 11.7% because that is the fraction of the April meeting that is in FFJ6. Do a similar calculation for the other quarterly meetings.

    Now you have all the meetings. However, keep in mind that the results may be off by as much as 1bp, and in rare instances, can be off by more. You also need to factor in potentially bad settlements that may affect your calculations. If you want to be more precise, you should consider the following:

    1) Make a daily model of FF effectives (I will describe in a later post). Depending on the calendar configuration, you may have more (or less) high demand days than others. This could affect the future from 0 to 1bp (or more in rare situations).
    2) Use the information in all twelve futures contracts. We only used the information embedded in eight of the twelve futures contracts. While the quarterly futures can be noisy, there is still information embedded in them.

    The above requires another level of work. For most people, I think just the basic sheet should suffice.

  • Curve Advisor
    Keymaster
    Post count: 612

    We recently had liftoff in the US, from a 0-0.25 target range to a 0.25-0.5 target range. Whenever you have a change in rate regime, you should monitor all of the assumptions you had been making in the previous regime. In particular:

    * how does the Fed funds effective trade to its target? After liftoff, we seem to be trading 11-12 bps into the 25 bp range (from 0.25 to 0.50), from 14-15 bps into the old 25bp range (0 to 0.25).

    * how does the Fed funds effective trade to other short-term benchmarks, like EDs? Libor has widened much more than the increase in the Fed Funds rate. The widening move started about a week before the Fed meeting, which could be an indicator of the markets pricing the move ahead-of-time.

    * how does the Fed funds effective trade on the day of the Fed meeting? For the purposes of how the Fed Funds futures trade, you should make a note of how the FF effectives trade on the day of the meeting. A 25bp move means about a 0.82bp move in the value of the FF future, just for that one day. The details can be the difference between a good trade and a very good trade. Interestingly, the NY Fed reported that on the day of the Dec ’15 Fed meeting, the rate was basically unchanged from the previous day… which would imply a 0% move factor for the day of the meeting. This seems suspect, and may not hold for future meetings.

  • Curve Advisor
    Keymaster
    Post count: 612

    The new year is upon is… and it’s usually the time we take an even closer look at the FOMC meeting calendar. I mentioned this a few weeks ago in the newsletter, but this week, we have two notable things about the scheduling:

    * The December meeting announcement is on Dec 16. This is notable because most quarterly Fed meetings occur after the ED options expire. For EDZ6, this is not the case, so those options (on EDZ6 and Z6 expiries), should all be worth more. Wouldn’t you want to pay more for an option that expires a few days AFTER a Fed meeting?

    * The meeting between the Sept and Dec meetings occurs on November 2, instead of later October. This is especially interesting in that there is a presidential election in the US which occurs 6 days later. I think if the Fed *has to* move, they will, but if it is a reasonably close call, they may be perceived as holding off. In any event, expect the Nov meeting to be underpriced, relative to the other meetings.

    The money is in the details. Whenever you look at the calendar, think about things one step further and see if there are any opportunities there.

  • amit
    Participant
    Post count: 5

    Hi,

    If we look at the ED-FF spread it is trading higher as relative to the historicals. Can you please explain.

  • Curve Advisor
    Keymaster
    Post count: 612

    My best guess is that it’s a combination of a few things:
    * An active Fed usually means more front-end uncertainty, so spreads will widen.
    * Brexit may cause some additional volatility – LIBOR is a London-based rate, after all.
    * We discussed this recently in the Canada thread, but the SEC implemented some new rules that go into effect for money market funds. This will cause less of a demand for commercial paper, so there may be an adjustment period for short-end fixed income.

  • Curve Advisor
    Keymaster
    Post count: 612

    I can’t believe it’s just been a week since we discussed this, but at the end of last week, the 3mo libor fixing was about 4bps lower to the Fed Funds strip than Eurodollars were. I noted over the weekend that over the past week, EDM6 sold off 8.75bps, but the 3mo libor fixing only rose 2.63bps (mostly after the Fed minutes). This situation lasted for a few days. So two days ago, I finally put out a recommendation to buy EDM6 vs 0.75x FFQ, and the trade is up over 2bps. If you constructed the sheet mentioned in this thread, you could have spotted the same pattern.

    To calculated the spread between the cash libor fixing and the FF strip, you calculate what the Fed meeting exposure of the current cash 3mo libor fixing is. You construct a weighted strip of Fed Funds futures to replicate the Fed meeting exposure of the 3mo fixing. The difference between the 3mo libor fixing and the weighted rate implied by the FFs is the spread between libor and FFs (approx 20bps). Similarly, you can construct the spread between EDM6 and the FF strip required to replicate the Fed meeting exposure in EDM6. The difference between EDM6 and the weighted FF strip is the spread between EDM6 and FFs (approx 24 bps). We only had two and a half weeks before EDM settled. That did not seem like enough time for for the two spreads to converge. So we put on the trade.

    For ease of execution, We also incorporated a “skip June, hike July” bias in there. This happened to be my view. EDM6 settles before the June FOMC meeting, so it was unlikely for the ratio between the June and July meetings to diverge excessively. But it is much easier to buy EDM6 vs sell 0.75x FFQ6 (DV01 weighted) than to try and buy EDM6 vs sell 50% FFN6 and 50% FFQ6. I may post this trade in the Positions section next month. It’s not a huge trade, but I like it as an example of just staying on top of the basics.

  • Nick
    Participant
    Post count: 8

    I found this old CBOT FF reference manual (attached). It has a useful guide on calculating the probability of a rate move that the market is pricing in for any given meeting based on FF price. Perhaps it’s of some use to someone?…

  • Nick
    Participant
    Post count: 8
  • Curve Advisor
    Keymaster
    Post count: 612

    Thanks for that. The first link was dead, so I removed it. Let me look at it next week and I’ll post my thoughts. In general though, this gives a reasonable approximation of how to calculate the FF probabilities, but I think a better way is to have a model where you list your estimate for each day (to account for month-ends, etc). This uses “FFER()”, which can vary by over a bp, depending on the month.

  • Curve Advisor
    Keymaster
    Post count: 612

    The methodology in the CME paper is a reasonable estimation, but it has two problems:

    * It ignores “special days” – in particular month-ends, which tends to trade lower. The FF effective rate has been somewhat stable. However, you can see that at month-ends, the rate can be much lower (see the graph on the bottom: https://apps.newyorkfed.org/markets/autorates/fed%20funds#Chart12). Depending on the calendar configuration, a month can have up to SIX month-end days. That could easily be 1-1.5 basis points. Keep in mind the rate from Friday carries over to Saturday and Sunday. So when you look at a particular FF contract, you have to know how many month-end days are in the current month (and the surrounding months) and adjust accordingly. In addition, there may be cases where days other than month-end days may also exhibit a different rate characteristic. For example, in the past, Fridays fixed lower because of the weekend.

    * It assumes that the Fed Effective rate on the day of the meeting is the new rate. Typically, you would expect the rate on a Fed meeting day to be closer to the new rate, but this is not always the case. When the Fed hiked in Dec 16, 2016, the Fed effective rate for that day was the old rate. You are going to want some flexibility to modify the rate at the day of the Fed meeting.

    For these reasons, the methodology in the paper is not “ideal.” You have two choices: (1) make a model that lists each day, that allows you to make any necessary adjustments. This is actually fairly simple using Excel. (2) Use the methodology in the paper, but make an adjustment for the above two factors.

  • Mike
    Participant
    Post count: 2

    Is it normal to have 3 meetings in three consecutive months? I just checked out the 2017 schedule and it seems they have meetings for may June July. Don’t think I’ve seen that before.

  • Curve Advisor
    Keymaster
    Post count: 612

    It also happened Aug-Oct 2012 and Nov-Jan 2012. It used to be more common prior to 2013, but starting 2013, they seem to have made some changes where the quarterly meetings corresponds to the IMM dates (more often than not), and the non-quarterly meetings are somewhat more evenly spaced between the quarterly meetings. The meetings have also been more typically on Wednesdays (not always) since 2013 (the last exception was Thursday Sept 17, 2015).

    Recognizing patterns like the above may be helpful in estimating which futures contracts have which meetings in future years.

    The spacing between meetings seems to be 6 or 7 weeks. We have also had a gap of 8 weeks between the July and Sept meetings in 2016. But if you have a spacing of 6 then 7 weeks, that is 91 days, and since 3 consecutive months could have as many as 92 days, there is just enough room to get those 91 days into the “right” three months. And it helps that the non-quarterly meetings are around the end of a month or the start of a month. It would be much easier (like May-July 2017) if the meetings were 6 weeks apart.

  • Mike
    Participant
    Post count: 2

    Thanks for the reply, I must say this place is by far the best resource on STIR futures trading. I do have one follow-up question: in previous posts you touched on what kind of assumptions to make on the day of the Fed meeting when it comes to FF futures pricing. It does seem like for the two most recent hikes (Dec 2015 and 2016), the EFFR remained at the previous rate on the day of, then jumped by about 25 bps at T+1, but I saw in another one of your posts that it is actually the norm for the EFFR to adjust upwards the day of the hike.

    I’m curious as to how it actually all works. On the day of a hike, do banks lend/borrow their excess reserves as usual in the morning hours prior to the meeting, and then as soon as the hike is announced the rate simultaneously shifts by 25 bps (and the resulting official EFFR is just an average of those)? From my understanding the earliest you can obtain the previous day’s official EFFR (unless you’re an actual participating depository institution) is the next business day around 9am EST, according to the NY Fed website. This tells me the the banks actually have a “head start” than the rest, since they have a better sense of the EFFR the day of the hike while the rest of the market only knows what the actual rate was the day after, meaning they have a short time window to actually arb the futures (provided that the market expects the EFFR to jump only at T+1, not the day of). If you had any experience dealing with that kind of situation I would love to hear about it.

    Given that a 25bp hike is worth just under 1bp in futures (close to two ticks), it seems like the the day-of-hike assumption is a pretty significant aspect of futures pricing. Not sure if I got this whole thing enormously confused, would love to get some more details on it given your extensive experience in the STIR futures space. Thank you.

  • Curve Advisor
    Keymaster
    Post count: 612

    Thanks.

    In the past, the FFER used to be closer to the new rate on Fed meeting days. However, as you pointed out, in the current environment, the Fed has used the old target rate for the day of the meeting. The new rate didn’t take effect until the following day.

    I’m actually not sure why this change has occurred. Part of it is that the Fed is manipulating the rate via interest on excess reserves and the reverse repos. So they decided to trade the day of the meeting at the old rate, for reasons unclear to me. However, I would assume this policy would be consistent going forward, so we have adjusted our spreadsheets accordingly.

    You make an interesting observation about being at an information disadvantage. You can see some intraday fed funds effective rate flows on some platforms. However, it is never going to be as good as what a dealer would see. And this speaks to the competitive disadvantage we are at when it comes to certain types of futures trading. A good example would be trying to trade something like a TED spread. I never understood how someone who doesn’t see swap or treasury flows would think they had some advantage in trading that. In any event, the amount of disadvantage in FF futures is rather small when you trade the futures that are further out the curve. However, I rarely take a position on the current contract, for the information disadvantage reasons you just mentioned.

  • Curve Advisor
    Keymaster
    Post count: 612

    @mguan said:
    I’m really interested in the modeling approach here [post 1174 in this thread]. Can you elaborate more on how you are using the websites effective rate to model FF futures? How do you use this information?

    The Fed Funds futures prices are calculated by taking the average of the FFER rates for each day. On days where there is no FFER, you use the previous day’s rate. So the Friday rate will carry over to Saturday and Sunday. For holidays. you use the rate from the previous business day. This practice makes the calendar configuration of the month-ends (and month-ends from the previous month) important, because those days tend to have rates that are noticeably lower than the other days. For example, FFJ7 contains FIVE months-end days, and so this will trade about 1.3bps higher than the surrounding FF contracts (FFH7 and FFK7), both of which only have ONE month end day in the contract.

    You just list every day on a spreadsheet, and put in a FFER for each day. You then average those rates over all days in that month, to get the expected FF rate. You can tie out to the FF futures contract for that month if the amount of basis point change for the Fed day causes your monthly average to match what is priced in for that futures contract. You can use this worksheet to estimate where the various FF contracts will settle in various Fed scenarios.

  • Mike G
    Participant
    Post count: 20

    I am not following your steps. So for April, we don’t have FFR rate for 4/1,4/2,4/14( good Friday), 4/29 and 4/30?

    You mentioned: “You just list every day on a spreadsheet, and put in a FFER for each day.” What FFER are you talking about? The published FFER is only for the front month futures contract, right? What rate do you use for other contracts further out? Can you give a simple example?

  • Curve Advisor
    Keymaster
    Post count: 612

    For April 2017, the five “month-end days” would be 4/1, 4/2, 4/28, 4/29 and 4/30. The rate from Friday 3/31 would carry over to 4/1 and 4/2. The rate from Friday 4/28 would carry over to 4/29 and 4/30. All of these FFER rates would be lower than the “normal” rates because they are either the last business day of the month (or weekend days following the last business day of the month).

    As a simple example of what to use for April, let’s assume the month-end days have FFERs that are 10bps lower than “normal” days. We know FFJ7 settled at 99.14. So the average FFER for April is priced to be 86bps.
    Let x = FFER for a non-month-end day
    Then x – 10 = FFER for a month-end day
    [25 * x + 5 * (x – 10)] / 30 = 86
    Solving, we get x = 87.7bps

    The current “normal” FFER (say on 3/2) is 66bps. Then we can estimate that 21.7bps are priced into the March meeting, since that is the only scheduled decision point for a change in the “normal” FFER. This is a gross simplification because the FFER can move for reasons other than the Fed changing rates (based on market supply/demand and what the Fed does to manage the rate). And after a hike, it is also possible that the FFER may be a different spread to the target range.

    Note that there are more FF contracts than meeting dates. So you have more data points than you need. For example, you can use FFH and/or FFJ to calculate what is priced into the March meeting. For an estimate using a single future, FFJ is generally preferred to FFH because it does not involve extrapolating half a month of FFH data to get the results of the March meeting. But you can also use both FFH and FFJ to make the estimate.

    A non-algebraic method of calculating what is priced in to the Fed meetings is to just list every day, and use some sort of “solver” to back out what meeting changes would result in matching the FFER strip to the FF futures strip.

    Let me know if you have any other questions.

  • Mike G
    Participant
    Post count: 20

    This month end pricing distortion seem to have started recently (after 2015). I think it existed before but at a much less noticeable way. Is it because something structural changed?

  • Curve Advisor
    Keymaster
    Post count: 612

    I think it was mostly a function of the zero lower bound, and rates getting compressed against “zero”. As we get away from “zero”, there is more room for things like the month-end FFER and possibly the year-end turn to get higher.

  • Mike G
    Participant
    Post count: 20

    Not sure if this question should be on a new thread. Feel free to move it where ever it belongs.

    I was comparing the prices of Eurodollar CS spreads from 2/28 when fed speakers started to hint on March rate hike until today 3/9 when the market has priced in a much higher probability of a hike. I wanted to share my observation to see if my thoughts make sense. I am looking for criticism mainly 😉

    As of today, the price of GEH7-M7 spread is 0.1975. A far from accurate estimated probability of March rate hike is 0.1925/0.25 = 77%. This assumes a 25 bps rate hike. If we look at fed fund futures hiking probability (from CME website) its at 90%. The difference between them represents the TED spread and risk premium.

    On the other hand, the GEM7-GEZ7 six month spread is current at 0.31. Following the above framework, this means the market is pricing in only 1 more rate hike after March. If I believe there is going to be at least 3 rate hikes this year, GEM7-GEZ7’s price should be greater than 0.50 correct? (March is a hike, there is 2 more after).

    What are your thoughts on this? Thanks again, Mike.

  • Curve Advisor
    Keymaster
    Post count: 612

    Keep in mind everything I am about to say is to help you. You REALLY need to make the spreadsheet that I’ve outlined in this thread (sample in post 1149). My philosophy is to “teach a man to fish” rather than “give a man a fish,” so rather than answer your questions directly, I am trying to help you to think about the markets and market pricing correctly.

    * You are incorrect in assuming that the .1975 is the pricing of a March meeting. In fact, the .1975 bps represents only 1% of the March meeting – the rest of the GEH7-M7 spread is something completely different. In other words, GEH7-M7 has almost no direct exposure to the March meeting. This is similar to the first post of this thread where I suggested that many people who have traded futures for years have no idea what meetings they are getting long or short.

    * We can have a discussion about TED and risk premium later, but when you look at a GE 3mo calendar spread and in the current environment, they are less relevant. We can also have a discussion about EDM7-Z7 when you make the spreadsheet.

    The spreadsheet previously described will help you tremendously. It’s fairly simple to construct. It can be done in Excel and does not require any programming.

  • Mike G
    Participant
    Post count: 20

    Thanks so much for the pointers!

    I’ve just created the spreadsheet and came to the same conclusion of the GEH7-M7 spread having no exposure to March. See attached(1.png).

    Using just Eurodollars futures, is it possible to back out how many rate hikes there are priced in between May and December? For example, according to the spreadsheet the GEM7-Z7 spread doesn’t have any exposure to the June meeting so as a spread it is measuring the pricing from July onwards (see attached 2.png) I didn’t use March contract since its expiring.

    It seems to me it is hard to get a clean cut view on hiking pricing only utilizing Eurodollars futures.

    Attachments:
    You must be logged in to view attached files.
  • Mike G
    Participant
    Post count: 20

    See this attachment for 1.png.

    Attachments:
    You must be logged in to view attached files.
  • Curve Advisor
    Keymaster
    Post count: 612

    The spreadsheet is a good start, but you have some things you can “fix”:

    * The IMM dates actually start on Wednesdays. The ED contract settles on the Monday before. By convention, the rate of the libor cash fixing on a given day starts two business days later. So the Monday fixing would actually take effect starting Wednesday. “The Eurodollar interbank time deposit rate is determined by the Ice Benchmark Administration Ltd. (IBA) LIBOR fixing on the second London bank business day immediately preceding the third Wednesday of the contract’s month of delivery. This value is then rounded to the nearest 1/10,000th of a percentage point per annum.”

    * The “libor 3mo rate” is a 3 month deposit. It does not go from one IMM date to the next IMM date. So four consecutive ED contracts do not give you even coverage throughout the year, as you will have some overlap or gaps.

    * I think your daycounts and/or the day of the Fed meeting is a little off on the FF calculations, as I get slightly different numbers. The meeting day of the Fed announcement was much closer to the new rate in the old days, but in more recent times, the FFER on the announcement day has been the old rate, and the new rate starts on the following day. I’m not sure if this will always be the case. I have a box where I can put the weighting of the day of the Fed announcement. I am currently weighting the announcement day as “10%” new rate and 90% old rate. But do whatever makes sense to you.

    Otherwise, your spreadsheet looks okay.

    EDs are a 3 month instrument, so it’s hard to get “monthly” precision with them. Most times, they also settle 2 days before the quarterly Fed meetings, so they are useless if you want to take a specific view through a meeting. And they trade at a spread to the Fed Funds rate, and there can be some variability in the ED-FF spread. But EDs can still be incredibly useful isolating groups of meetings – especially if it is a further out view that you do not need to hold through a meeting. In recent times, the non-quarterly meetings have not been as active, so having a fractional amount of a “lesser” non-quarterly meeting does not get too messy. This could change at some point, obviously.

    I’m not completely sure what you are asking, but there are also monthly 3mo Eurodollar futures. These are called “serials” – and for EDs, serial futures are available for the next six months (some other countries have the first three months). So you also can choose from EDJ7, EDK7, EDN7 and EDQ7 (in addition to EDM7 and EDU7). This gives you a little more flexibility.

    If you want a more accurate way to isolate the May through December meetings, you have a few alternatives:
    * FFJ7-FFF8 spread. This is probably the most straightforward way (assuming it didn’t have to be “exact”).
    * FFJ7-EDZ7 spread. This exposes you to libor widening out vs FFs, but this may fit your view. In the current environment, if the meeting is “fully priced” and there was no chance of a skip or a 50 (or anything other than a 25), you can just sell or buy EDZ7 (or FFF8) as a single contract trade.
    * EDJ7-EDZ7 spread. This only has 16% of the May meeting and 62% of the June meeting. But I suppose you could supplement with a little FFJ7-M7 spread or something.

    Let me know if this answered your questions.

  • Mike G
    Participant
    Post count: 20

    Thanks for the response. I will attempt to fix it and see if I have more questions. Currently, I am building the spreadsheet on google sheets which is convenient.

    I am failing to create the pricing of each meeting to get a holistic view. In the blog posts you occasionally write, you post each meetings pricing in bps. When I tried to replicate somethings similar, the resulting pricing doesn’t really match with what I am expecting.

    The way I am currently doing it is to use the spreadsheet to find a spread that is able to isolate a single meeting. For example, for today’s March meeting, I am using the FFH17-FFJ17 spread. If I enter this spread in to the spreadsheet I get a net exposure of around 0.47 for the march meeting (nothing else for other meetings). Using linear interpolation, I get the Marching pricing in bps: (FFH17-FFJ17) * (1+(1-0.47)). Using this, I derived a Marching meeting pricing of 0.1568 using 3/14/2017 FF settles.

    Now this is rather far away from the 90% probability from the CME rate hike website (15.68/25=63%). What am I doing wrong?

    FYI here are the other spreads I am using for the other meetings:

    May Meeting = J17-K17
    June Meeting = K17-M17
    July Meeting = N17-Q17
    Sept Meeting = U17-V17
    Nov Meeting = V17-X17
    Dec Meeting = X17-Z17

    Thanks,
    Mike

  • Curve Advisor
    Keymaster
    Post count: 612

    I think the best thing is for you to think about this more. Here are some hints:

    * Why you can’t back out the FF pricing for March. Take a step back. At first glance, it appears that about half the March meeting is 10.5 bps. Therefore, 100% of the March meeting would equal about ______? [Hint: it can’t be something as low as 15.68bps. Always do a sanity check.] The two other things you need to factor in are:
    (1) you need to factor in month-end and any other adjustments (year-end, special financing days, etc, but these may be negligible) before you start calculating probabilities. We discussed how FFJ7 has five month-end days. FFH7 only has one month-end day. How would that affect the Fed probability pricing?
    (2) there can be an “off-settle” in daily settlements. Since the FF contracts have to settle in half bp increments (the first FF in quarter bp increments), you will get a small degree of error that can be magnified when extrapolating.

    * What FF spread to use to estimate a meeting. You have 12 variables (12 FF contracts) and 8 unknowns (8 Fed meetings). There are many way to calculate the meetings, including statistical fitting methods. However, given (1) and (2) above, what would be best way to estimate the Sept meeting: (1) Q17-U17, (2) U17-V17, (3) Q17-V17? Why?

  • Mike G
    Participant
    Post count: 20

    Hi Joseph,

    I wanted to ask you about the relationship between short term interest rates and say other treasury futures. Is there a meaningful one that can be exploited from your experience?

    I read an article awhile back saying how the 2 Year Carry spread (2 Year Yield Vs Target FF Rate) represents feds hiking expectations after the third meeting. Have you ever explored this area trading FF Futures against other treasuries?

    Thanks

  • Curve Advisor
    Keymaster
    Post count: 612

    I am of the opinion that most people are at a major disadvantage trading treasury futures. Most people don’t realize that to properly price a Treasury future, you have to keep on top of things like the cheapest-to-deliver, repo/financing, issuance, etc. Most people just look at it as a punting (aka directional trading) vehicle. Over my career, I’ve sat with several (very smart) guys who would routinely look at all these factors (and more) and trade the treasury futures. And as smart as those guys are, they were still at a “disadvantage” because they didn’t see all the flows like the market-markers did. So I never bothered to look at anything in depth where I felt like I was at a disadvantage.

    Because trading futures is a zero-sum game, before you trade anything, you need to make sure you are not at a severe competitive disadvantage to the rest of the market. I fail to see how most people aren’t at a disadvantage when trading Treasury futures. A number of years back, I think a new bond future opened up, and it initially traded a few big figures off because someone didn’t realize the cheapest-to-deliver had changed. So if you aren’t going to spend a lot of time looking at Treasury futures, I’m not sure it’s worth touching. I know people trade TED spreads (Treasury vs EDs), but I doubt there is any value there that an algo hasn’t already completely captured.

    As for the article, I can’t comment without having looked at it. At first glance though, it looks like FF-2yr spread would probably be closer to the Fed hiking expectations for the next “three” quarters or so. I’m not sure how someone could say “after” the third meeting (since the spread starts with the current FF rate), or even “third” meeting (since that is dependent on how quickly the Fed is expected to hike).

  • Mike G
    Participant
    Post count: 20

    Please see attached for document.

    Attachments:
    You must be logged in to view attached files.
  • Curve Advisor
    Keymaster
    Post count: 612

    I think what you meant to say was, “I read an article awhile back saying how the 2 Year Carry spread (2 Year Yield Vs Target FF Rate) represents feds hiking expectations BEFORE the third meeting.” They are basically saying the 2yr “current” yield = expected funds rate at third meeting.

You must be logged in to reply to this topic.