Curve AdvisorKeymasterJanuary 16, 2015 at 11:29 amPost count: 612
You can think of trades being classified on a scale of “rangeboundness,” where one end of the scale is something that is completely unbounded, and the other side is something that trades in a well-defined range. For the newsletter, I generally use a [-3,+3] range.
* -3 would be mostly unbounded (for example, S&Ps, gold, EM FX, etc. – until you get somewhat close to 0 (or some other valuation support), it could be anywhere). Even if we had a 200 point correction in S&Ps, are you any surer it couldn’t go down another 200?
* +3 is something that is mostly range-bound (for that particular longer-term environment). Examples of the latter would be: ED12-16-20 vs 4x ED14-16-18 fly. Or even something simpler, like a ED18-20-22 fly. I’m not saying these couldn’t trade out of the historical range if the environment changes significantly, or someone for whatever reason decides to buy or sell a ton of one of the contracts. But most of the time, there generally will be reasons the structure has traded in a particular range.
* 0 would be something that may have some kind of boundary, but it could easily trade like there was none. For example, EDM5-Z5 spread. It is only 29, so one may think that as long as the FOMC is talking about “mid-2015” hikes, it should stay somewhat supported – you get 4 meetings in there and there is a little over 1 hike in there. That seems like there should be some support for that. But that thing could easily go to zero if the Fed rhetoric changes. And more disturbingly, that thing could get negative, if we get another libor blowout, or if the Fed has to ease to negative rates for whatever reason.
Generally it doesn’t make sense to scale into trades when a trade is not range-bound. When you get a trade that is more range-bound, it starts to make more sense.
Curve AdvisorKeymasterFebruary 4, 2015 at 7:52 pmPost count: 612
I’m sure someone out there is going to ask, “But what if you get a six-sigma event?” Well, I would say most of the time, you are looking at something closer to a 3 sigma event but you grossly underestimated the potential move. It happens to everyone, since it is mostly a judgement call. That EURCHF move on Jan 16, 2015 was statistically a 180 std dev move. It was NOT a 180 sigma event. If you factor in policy (error), no one should have thought that was more than a 3 or 4 std deviation move. The history was just misleading. So don’t put too much credence on historicals.
But that doesn’t mean that you shouldn’t scale in. I’m sure someone could come up with something statistically robust. If Robert Almgren can write a paper on how to maximize getting filled on single lot contracts, figuring that out for flies would be a breeze. Of course, even with modelling, you are still replying heavily on history, which may or may not be representative.
[NOTE: I did not specify that this was for a more “rangebound” trade. Please see discussion later in the thread for unbounded (directional) trades.]
Once you find a trade you like, what I prefer to do is the following:
* Get on 25-50% of the trade. Nike (Just do it). Because you’ve already identified the “good” trade, so just get some on. You may like a trade, even though it’s basically at the mean historical value, because of a market view. But typically, you probably want to wait until we get 1.5-2.0 std deviations from some target, and do something closer to 50% of your size.
* determine some threshold of what a reasonable “large move” could be (“max entry point”). You may use historical std deviation as a guide (and pick a max std dev to get to), or some kind of historical high/low, or maybe even some key technical level.
* figure out how you want to do your add to your position between your initial entry point and the max entry point. Because I am a simpleton, I just split it evenly between intervals.
[to be continued]
Curve AdvisorKeymasterApril 8, 2015 at 4:05 pmPost count: 612
Unfortunately, sizing trades is not as “automatic” as in technical trades. Neither is setting stops or take-profit levels. However, I think of this as an “advantage” rather than a “disadvantage.” Because if it was that easy to determine when to get in and out, then you could easily program an algo to do it, and there would be less of an “edge”.
But this is not to say that you can’t determine reasonable entry/stop/take-profit levels. If there is no fundamental “regime change” in the way the curve is trading, you can use historicals to set your various levels. However, as the shape of the curve changes, you may have to slightly lower or raise your levels. For example, you buy a fly as a bearish trade that you think should be 5bps higher. This trade has a 20% beta to the direction of rates.
* We rally 25bps but the fly is unchanged. You need to unwind the trade, even though you did not make the 5bps you thought you would. This is because after the rally, the fly is “fair” to the curve. If there is no value, you need to unwind.
* We sell off 25bps and the fly makes 5 bps. You can take profit on some of the trade, but it is very possible/likely that fair value is now 10bps higher from where you put on the trade. So even though you got to your tale-profit level, you can increase your profit target. This is because after the sell-off, the fly has moved 5bps, but it is still “cheap” to the curve.
* Rates in the long end don’t move, but the central bank just states that hikes will occur much slower than previously thought. This may be a regime change. There could be less curvature than previously thought, so your historicals may not be as important. You should exit, even though rates did not change, because going forward in the new environment, the fly may now be “fair” or even “rich” (when we thought it was previously “cheap”).
Curve AdvisorKeymasterApril 9, 2015 at 9:26 amPost count: 612
It’s not just butterflies with a directional component where the entry/stop/take-profit levels can change. There can be many reasons to like a a fly trade. But you can get a material change in the curve that could affect the appeal of a trade. For example, one of the reasons you may have wanted to do a trade is for rolldown, and/or another reason you liked the trade was its level relative to other similar structures. These things can change for the worse (or better). So you will need to regularly monitor your trades to select the optimal add/remove levels.
Curve AdvisorKeymasterNovember 17, 2015 at 11:23 amPost count: 612
For most trades, the risk on a “rangebound” trade are not symmetric. In other words, the risk of being long at a historically low level will not be the same as the risk of being short from a historically high level. The risks will vary based on the market environment, and assumes the historicals are relevant. For many trades, we generally prefer it in one direction, for a few reasons:
* A trade will only roll positively in one direction. For example, if you are long double flies further out the curve, the trade may be positive roll, so over time, you generally would expect to make money. This is not the case when you short that same double fly, where the trade keeps rolling higher.
* Fundamentally, a trade may not be as attractive in the other direction. If you think it could take a while for the Fed to reach their target rate (higher flies), you may be more inclined to buy flies at low levels further out the curve. Selling flies would be more if you thought we could get a recession, or some kind of massive curve flattening.
* The risk of a large event may favor a trade in one direction or another. I call this “error skew.” If rate hikes are mostly priced in, if you take a short position, you may have more to lose on a rally than gain on a selloff. Think about the absolute moves, as well as the moves on an expected value basis.
You generally want two or more of the above factors in your favor on a trade, unless you have a very strong view.
Curve AdvisorKeymasterFebruary 4, 2016 at 7:54 pmPost count: 612
In the first post, I wrote, “Generally it doesn’t make sense to scale into trades when a trade is not range-bound. When you get a trade that is more range-bound, it starts to make more sense.” The reason I wrote this at the time is because I was thinking in the big picture, that an unbounded trade has a range that is too wide where using any kind of “adding” strategy makes sense. For example, take USDJPY. This has traded in a reasonably tight range between 117 and 125 for the past year. All other things being equal, buying USDJPY makes some sense at the lows. However, with FX, you never know when it will break out huge to one side or another. There’s no reason USDJPY couldn’t be 80 or even 200. So when you are thinking of adding or scaling in, you really need to think about the bigger picture.
Because if you thought it was going to go one way but it didn’t, then from a conditional probability perspective, given the markets are the opposite of where you thought it would be, the chances of you being wrong are higher. So why would you add to a directional trade? It turns out there are times when scaling into a directional trade makes sense…
Curve AdvisorKeymasterFebruary 5, 2016 at 1:54 pmPost count: 612
Here are some times when you might want to scale into a non-rangebound trade:
* You have a reasonable fundamental view, but don’t necessarily love the entry point. You may just put it on in small (even microscopic) size just to get it on. When we get to a better entry point, you can do more. The strength of your view will most likely be a function of the price level. So you will probably want to buy more when the price is low and vice versa. So if the price of something you like is high, you may not like it enough to want to buy your full size there.
* There are multiple technical support levels. You can do some at one level and add if we get to the next. You should get in the habit of looking forward at what you would like to do at various price levels – both up and down.
* Something happens to strengthen your view. Your assessment of the risk/reward should constantly be changing. Some of these things can be anticipated (for example, a central bank meeting) and some just happen. In either case, if your view can change, you should always leave some room to add.
[I may add to this list]
Curve AdvisorKeymasterApril 15, 2016 at 2:18 amPost count: 612
There are a wide range of scaling possibilities. Sometimes, is more optimal to NOT to scale. Other times, it is more optimal to scale frequently. And every thing in between. There are a number of factors you need to consider when determining whether or not to scale and at what pace you should scale:
* Your view of the reasonable potential “target” of the trade. The more you are looking to make, the less frequently you should scale.
* Your view of how choppy I think the trade could get. You need to make some sort of determination as to whether it is more +EV to hold, or to trade the “chop”.
* Your view of how volatile think the markets will be in the near-term (ie. are there catalysts for a large move on the horizon?).
* Your view of how your positioning compares to how you think the markets are leaning (directionally). There may be times you want to let something run with the market, and times when you don’t.
* How the book fits in with the rest of your portfolio. If a trade is bearish, and you need some “+EV bearish trade” to give some balance/protection to the book, you may want to keep longer. Or if you just unwind all your other bullish trades, you may want to scale out of the bearish trade tighter.
You must be logged in to reply to this topic.