How do you approach trading spreads like this?

If you have the $$$, then a server-hosted auto-spreader still seems like the best possible option.
I have not touched the energy spreads for a while, but my experience with other multi-legged spreads is that execution is very touchy. We have a fairly nice in-house HFT-based solution for it and still I am not happy with the quality of execution.
 
All,

My main goal currently is trying to reduce my true cost of spreads for non natives. Ie crude fly. Execution is fixed, we all pay based on our size.
Slippage and spreads are variable. There is a direct relationship between liquidty and these costs.

I view this as a double edge sword, I am rewarded more for making a book in a thinly traded market, but I need participants to sell my goods to.

To aid in spread selection, I am trying to ensure enough liquidty but not too much where the profit can't cover my costs.

Knowing that the back lots are typically the thinnest traded, What do you use to gauge liquidity and what ratio of the daily volume is your max for spread sizing on non native spreads. Ie min daily volume required to trade a ten lot fly on the least traded calendar? I have used an spread trader in the past to stand on the back leg of a pair to some success in terms of mitigating slippage, but have got stuck waiting for a order to come to market and not get picked off by another bot or have the other leg move.

Thanks for your time.

PS. Where is ET hiding the spell check button? Sorry for any typos on my phone.

If you are market making then its absolutely all about liquidity and low round trips. As you said if you market make a spread with a large b/o spread and there is very little liquidity its sub optimal. When I used to market make I would very rarely trade any legs that had average daily volume sub 1000 contracts, otherwise you end up stuck in a position waiting for fills.

What I used to do was find commodity products that traded enough for me to make a decent market in and get plenty of fills BUT there was not enough volume to attract the huge electronic market makers as it wasn't worth their while. That way you have a solid edge. You need $2 round turns ideally less to make this viable.
 
Here's an example of what I mean by these non-implied energy flies using CL M7/Z7/M8 as an example and why they present difficulties with native stops and limit orders:

IB quotes:
View attachment 171495

CTS quotes (sim, but still live data):
View attachment 171494

I took these screen captures within seconds of each other so aside from some minor differences they should be quite close. The main things to notice in the CTS capture above are the bottom row which consists of the following three instruments in increasingly preferred order:
  • Synthetic fly using M7, Z7, and M8 outrights (CTS tradesniper)
  • Exchange traded fly (Standard order book)
  • Synthetic fly using the M7:Z7 and Z7:M8 calendars (CTS tradesniper)
The top row is the outrights and the middle row is the calendars (with M7:M8 thrown in for the hell of it).

As can be seen the worst possible execution would be buying and selling the outrights individually. Never, ever do this unless your house is burning down and you absolutely must get out right that second. Based on the math of the outrights themselves and crossing the spread (hit the bid/ask for everything):
Code:
1 M7 @ 51.63 51.65
2 Z7 @ 52.54 52.57
1 M8 @ 52.53 52.58

A == 51.65 - 2 * 52.54 + 52.58 == -0.85
B == 51.63 - 2 * 52.57 + 52.53 == -0.98

This squares with what CTS shows with the synthetic outright based fly and we see a 13 tick bid/ask spread (which is complete garbage).

Using the native exchange based spread (remember, the liquidity seen there is a separate book from the outright book and a separate book from the calendar books without any cross-spread or outright implication [because the latter is disabled]) we go from a 13 tick spread to "just" a 10 tick spread:
Code:
A == -0.86
B == -0.96

Another thing to remember here is that the liquidity in these native fly books comes and goes (especially after hours). There are times I've seen my *target* order on a CL fly as the best bid or ask because all other liquidity on that side of the book disappeared and even had a friend confirm he saw my target as the best bid or ask using a different platform. That friggin' sucks and it especially sucks if one of these wide ass books gets lifted by a 1 lot unfairly triggering your stop in the process (which is why you should never use anything but a stop limit, period). The only positive these native flies have going for them is that they're guaranteed to fill in one order (but you'll never be filled at anything particularly advantageous compared to using the calendars for execution).

Now using the 3rd synthetic fly based off of the two underlying calendars results in the tightest spreads (because the calendars have tight liquidity and implication enabled) but is impossible to use with native exchange based stop/limit orders - which is the overall crux of this issue I'm getting at:
Code:
1 M7:Z7 @ -0.92 -0.91
1 Z7:M8 @ -0.01 +0.01
A == -0.91 - -0.01 == -0.90
B == -0.92 - +0.01 == -0.93

That's a 3 tick spread and also jives with what CTS shows. 3 ticks is greatly preferable over 10 ticks and especially preferable over 13 ticks. However, what sucks about it is that it requires a separate program to monitor said synthetic instrument, has the potential for leg risk (2nd calendar might not get filled), cannot use exchange based stop/targets whatsoever, and requires said program have constant internet connectivity to monitor the instruments. In the case of tradesniper while it's not an auto-spreader in the traditional sense (doesn't keep sending/pulling orders) and can actually manage risk on a synthetic it's still something that needs to be constantly on/present at all times.

I honestly don't know how people are doing risk management on WTI and Brent flies (or anything else beyond a calendar spread with a separate book that has implieds disabled) using exchange based stop/limits. The order books suck compared to the calendars and you cannot use the exchange to risk manage these flies using the underlying cals which have much tighter spreads. This is also another avenue where ICE could take advantage of the situation by out-doing CME but even they seem to be doing similar stuff with their own WBS and BRN flies although they have the technology to do it (https://www.tradingtechnologies.com...arket-window/implied-functionality-by-market/ https://www.theice.com/publicdocs/technology/Additional_Implieds_FAQ.pdf).

The one piece of good news here is that there are relatively cheap non-autospreader based options (tradesniper) that allow you to get the job done if you're willing to ensure they're kept connected and monitored at all times. A backup native order could also be used but better make sure you have a way of killing it if/when you get filled on the synthetic. If you have the $$$, then a server-hosted auto-spreader still seems like the best possible option.

This is the sort of thing you will not see in any book/paper. In reality oil flys are traded as per your right-hand ladder, synthetically calendar to calendar. Say for example on that right hand ladder I wanted to buy the fly I would be bidding -91 and/or -92 depending on how badly I wanted the fill. You have correctly identified that stops are dangerous things if you are trading energy flys. As you say b/o spread can widen hugely outside the US session and a stop could be triggered causing you to get a very bad exit price or be legged. Bear in mind when you enter and exit flys you never pay up using a market order unless you are desperate to get out OR perhaps if you are swing trading and just want the fill.

Most traders I have seen who trade these markets do not use a hard stop. It’s about bankroll management and your average price position. The pros use TT or CQG, you can link to an excel spreadsheet that uses mid-point figures to calculate spread values no matter how wide the b/o out of hours.

To trade these spreads properly you have to decide what you average hold time is going to be. Are you going to hold them for days/weeks? Or are you going to hold them for a few hours to a couple of days?

If it’s the former you can get away with a cheaper front end, some offer free, you can also get away with $4 round trips as you are holding for bigger targets. If it’s the latter you likely need TT pro or CQG spreader which will cost $1200pm min. You also need $2 round trips so you have to get specialist clearing.
 
We leg exchange calendars versus exchange calendars in order to make up our butterfly and condor intramarket futures positions. I strongly discourage my clients from legging individual expiries under any circumstance. With respect to the bid/ask spread, and under the conditions that we are swing trading these spreads, if we can split the bid/ask the slippage works fine in terms of our price targeting.

If I can get my guys to generate great returns whilst avoiding the cost of TT AutoSpreader or CQG Integrated Client then that's a huge win.

I also suggest to my greener clients that they used a big Chicago FCM with plenty of institutional and spec spread clientele. I suggest to these less experienced clients that they have their FCM's 24 hour execution desk execute their spread orders giving them a couple tics DRT if necessary. Seems to work fine for us - keep in mind that we swing trading and that might be a completely different approach than yours. YMMV.
 
here is a teaser for you and showing how reading books and papers can get you only so far. people with specialist knowledge in this field are trading things like the below chart. This is a double butterfly spread away from the front of the CL curve. It shows the market profile of last 8 days. It's 8 round turns to get this spread on so retail will pay around $32 to get a 1 lot on at $10 a tick it will cost them 3.2 ticks to get in and out so it prohibitive for them. The lowest volume tier clearing through a prop firm will get you around $2.20 a round turn so $17.60 to get a 1 lot on. The largest spread traders will be getting it done for as low as $10 but you would have to be doing huge volume to get that.

It doesn't take a rocket scientist to work out a money making strategy, the daily range is around 10-15 ticks. The best keep their leverage low. On a 20 lot a trader with a $100k account can comfortably manage their risk. If they bank a 7 tick winning trade $1400-$352 = $1048. If well capitalised and using low leverage scaling in is often used, positions are usually only cut if weekly /monthly ranges are violated. Hold periods a few hours to a few days. You need to make sure you have liquidity in the back leg.

When I talk about opening a can of worms this is what I am referring to.

enjoy

View attachment 171087

here is an observation which my get a penny or two to drop. Why would you want to trade spreads like I posted above away from the front? As you can see from the 2 charts below outright oil has shat out aggressively last 2 days dropping the best part of $4. Contrast that with the daily price action on the above double fly spread I posted above. We stayed range bound albeit the POC moved down slightly yesterday but nothing out of the ordinary and very manageable in terms of risk. look at the price action of the daily profiles showing the last 10 days from 24th Jan, compared to the previous chart I posted earlier in this thread. not bad eh?

flatpricemay.png
dlfy.png
 
Yes the first chart does look like a mess (to me anyway). Looking at everyone's posts on spreads, mainly using a mean reversion range over multiple days (and longer), it does look much clearer. All of this started me reading all of Bone's posts last year, and yours and many of the others that volunteer their time to help others. Thank you.
 
Last edited:
... and yes the range looks much more manageable on the second chart. I don't know what the far history looks like on the left side of the the chart (continuous spread futures charts - I think Bone mentioned something ages ago?), but is it a more cost effective option for a retail person to hold this three days to a week each time? A 30 tick range absorbs a lot of those retail costs. At the moment it's mainly a learning exercise as I won't be touching these for a long while.
 
Last edited:
... and yes the range looks more manageable on the second chart. I don't know what the rest of the the chart looks like, but is it a better option to hold this three days to a week each time?

yes. hold times say 1 day to 2 weeks. working an average position.
 
... ummm ... ok ... what's the cheapest way to chart these double flys with even bar or line charts? Is esignal the cheapest option for longer term continuous spread charts?
 
... ummm ... ok ... what's the cheapest way to chart these double flys with even bar or line charts? Is esignal the cheapest option for longer term continuous spread charts?

yes cheapest way is esignal or you can get a TT web demo. You cannot plot spreads like these in a continuous format like you can a prompt month future.
 
Back
Top