..very true, if only strict risk management was applied, on any given trade, one can tell when a trade goes beyond the threshold of being viable. When trading outrights.. you can almost sense the underlying bias and the direction/orientation for entry. When volatility is present its even easier.
It's been 3 years and I'm back!
- Thread starter s0mmi
- Start date
..very true, if only strict risk management was applied, on any given trade, one can tell when a trade goes beyond the threshold of being viable. When trading outrights.. you can almost sense the underlying bias and the direction/orientation for entry. When volatility is present its even easier.
Hey s0mmi.
If it isn't too much trouble could you elaborate on how you recommend to start researching a new spread?
If I go back and look at each day and how the spread has traded what kind of things do you think I should be taking note of?
Also, should I choose a particular session and only look at that session or should I look at the entire 24h?
Thanks in advance.
I can make you a profitable strategy from scratch if you decide to do some research and put in the hard yards. If someone else is reading this message and they're struggling (like I was, for a long time) then they probably deserve this juice straight out of my nuts because we're onto Page 12 now.
Anyway if you want to start researching a spread, first pick if you want to specialise in Bonds or Equities. You can do both, eventually, but it's better to bang the same prostitute a few times before you're ready for drunk orgies.
Step 1: Pick two products that correlate with each other or fundamentally should link up over time. Let's start with a super easy one, Australian 10yr bond vs. Canadian 10yr bond. The CQG code is HXS for Aussie, and CB for Canada. Both cash baskets are 9.5 years so they're exactly the same duration matches. No toggling needed.
Similar economies, similar GDP, similar population, both "commodity linked" nations and their currencies are very similar too... so it's a good starting point.
Step 2: Plot the chart ratio up that you prefer to trade. If it's a DV01 neutral strategy, get the DV01 of both and match. The Aussie's can be found on SFE website in an excel sheet (it's like $101 atm), and the DV01 for Canada is $104 Canadian. The currencies are the same so it's about 1 for 1.
Chart: HXS-0.096*CB
How to get 0.096 co-efficient:
HXS DV01 = ~$100 (Aussie Bucks)
-> HXS move in 1 basis point increments (quoted in half-tick though). So the numerator is [1].
CB DV01 = ~$104 (Canada Bucks)
-> CB is $10/tick. So ($104/$10) = ~10 ticks in a basis point. So the denominator is [10]
Chart: HXS-(1/10)*CB
Ratio: Currency (DA6/CA6) = 1.007 (Basically parity)
If you made 1bp on Aussie 1-lot, and lost 1bp on Canada 1-lot, it would be $101 versus -$104 on the other leg.
(104/101) = 1.02 Aussie contracts needed for every 1 contract in Canada.
So the ratio would be 1 to 1. So if you traded it, it would be 10 versus 10. Perfect and easy.
Step 3:
Now that you have a chart code and ratio, you can do research. You want to know how much money is made/lost on extreme moves as well, so you can work out risk/reward and probabilities for yourself. I will give you two different approaches to this. You can play with this however you want and you can specialise in one or the other. It doesn't matter what you choose... all the matters is that when you click the button you are not a sitting duck.
Style A - Levels Trading
Go back 12-months. Bring up a 30-min chart of both Aussie and Canada outrights. What you need to do is go through every single day and write down, at the Session open (canada time) how close or far both products were from levels. Mark a level and write it down in excel.
Then you need to write down how far away it was, if it bounced or not, how much it bounced, and of course you need to write down where it finished at the end of the session.
This is a dynamic version of trading. If you start the session and the XT is near a level (or kind of near it), you may think that covers may happen around there, so you want to be 10 by 10 in the middle, then if it goes down there you want to be 12 by 10, or 10 by 8 (You're actively managing a trading as it approaches or goes away from levels).
Anyway you need to figure out what information you'll need to also change the ratio if the levels break. Maybe you want to be 8 by 10 if the Aussie XT level breaks, or vice versa.
Style B - Volatility Trading
Go back 12-months and start plotting on an excel sheet the high and low of the session. Make another column with things like "reversion" (how much it came back in that session from the high or low), and also write the Rate and Number of Ticks (in Canada or Australia whichever you prefer) that it travelled.
By the end of it you should have median reversion rates and general homework on the highs and lows and how it finished at the end of session. If its too dangerous, you know its not a trade to fade and trade volatility around.
You can make an ATR chart and see what the 20-day ATR is. Plot these distances on your chart using an indicator or whatever and you'll have "value areas" based on volatility. Note you don't have to take the other side. You may find products that have great momentum when they go. For example you might see that in the past 3-months, a spread like the Bund v Buxl --> when it goes 30 ticks it has a 50% chance of going 60 ticks, and like a 80-90% chance of letting you scratch or take a -10 tick loss.
Its up to you how you want to think about it. Some things are clunky in their movement, and others like to run.
But of course, these are all great in a vacuum.. because there's always potential fundamental shifts happening due to data.. see below:
== Applies for Both ==
You want to definitely make a "Notes" column. go to ForexFactory and make sure you can explain every big move (if you can). Was there Red Tier data in australia, or canada? Was there red-tier american data? Was there roll-weeks? Very important. These days, you must be aware and should probably avoid if you do research and find that your EV is negative if you punt.
================
Canada only opens for a specific time thats why I used this as an example. You dont have to split the sessions. But of course for everything else, yes its important (usually Aussies since they open all day).
By the end of the homework, whichever style you decide, you won't be clicking somewhere and thinking "I hope it goes up". You will know that you're in the 50% ATR area, or 100% ATR area, or whatever.
And if you haven't figured it out yet, trying to make it all on 1 product is basically suicide these days. You can thank those secretive f*ckwits who browse this forum for contributing to that. Did you know that there are scum in London and Chicago who posed as new trainees in my inbox to try and get me to tell them how I trade? One of them sent their firm to Australias Bond market. I trade every major bond market yield curve now anyway so it doesn't effect me at all.
I've given you a blueprint now. You can do whatever you want if you go back and plug away. For future reference, all the kids I back are allocated this homework when they start. It gets creative after that, but this is the baseline.
The #1 goal is to get you to at-least "know" if a trade is Great or has STD's by doing homework first.
And I have to leave the most important note for last; Do not think you can make money in your product every single day. Overdosing on the trading heroin will kill you. Once again, secretive scum sent their f@ggot firms to slice 55% win rates just about everywhere so you can't get away with "low level" point and click strategies. Align yourself on high probability trades by first picking what you think is good enough for you. Is it 70%? Is it 80%? 90%? The higher you go, the more infrequent it becomes. And this is why you need to probably grab 3-5 of these. It's not glamorous but at least you won't need to waste your time reading those rubbish trading seminar/scam tools on this forum.
Last edited:
Thanks mate this is a great place to start. I'll work on this for the next month and see what I can find. I feel style B appeals to me more so I'll start with that.
What happens if the duration of the bond baskets differ? If I choose an equity spread like SPI v ES what's the best way to chart the spread?
If I start with a spreadsheet as follows is there anything else I should add? Open, high, low, close, reversion from high, rate of reversion from high, reversion from low, rate of reversion from low, notes on news, general notes on how the spread traded for the session, other random notes.
By "the rate ... that it travelled" you mean just a quick note on how it traded from the high/low?
One of my favorite trading dictums is:
"Don't confuse your account balance with intelligence."
s0mmi, you (and a bunch of others on this thread) would obviously understand that.
Best Wishes, and continued healthy outcomes to you.
Awesome thread.
"Don't confuse your account balance with intelligence."
s0mmi, you (and a bunch of others on this thread) would obviously understand that.
Best Wishes, and continued healthy outcomes to you.
Awesome thread.
1. Duration linking
For a strategy based on the relative change of ATR, you definitely want to be linking durations together as much as possible. If you don't, you open yourself up to the belly of the beast shifting and add more moving variables to the equation.
Take an extreme example; how do you think the spread between the U.S. 2yr note and the German 10yr Bund is going to be look like? It would be two mammoth outrights moving around with less weight on the U.S. to German correlation... and when the short-end Eurodollar or 5-year note move, the 2yr notes will be packing a punch... and the German Bund will not care.
The game is hard enough as it is so you can't afford to cut corners.
2. Rate on how it travelled
What I noticed is, if I went through by hand and actually wrote down the distance from high to low, and made notes on what it did, then I was much more aware of the product versus someone who "eyeballed it" from memory. For example, on some spreads I would note down "Once it hit point X, it was the point of no return". Once you make many mentions of this, and highlight your excel row in Red, you will now get some visual representation of what it does and can start to make a trade idea/plan off it.
3. Equity Spreads
If you do spi-mini then you might want to just do a basic "Follow the Leader" type of model. What I do is I drop the co-efficient of the Leader and I go through a few weeks just to see what works best. What you're really supposed to do is do a manual regression and grab the real number.
For example, the spi following the s&p500 in europe/usa 1 for 1, is very directional and suicide. All I did was drop the weighting to the minis by 20%. The chart is AP-2*EP (real ratio: AP-2.4*EP)
All you do is measure in distance what the spi should do if minis drops 1%. I just concluded it should only drop around 80% in these times. Maybe in the future it will be more.
Anyway with equity spreads you really need to know what the hell moves with what. I don't have access to the stat tools but if I did, I would plot the correlation between London FTSE and S&P500 (during different times too), the Nikkei and S&P500 (separate times too), the Aussie spi & S&P500, the Dax and Eurostoxx, the Eurostoxx and S&P500... every combo you can find. Then probably do your ATR or level strategy on whichever gives the best set up of the day.
I would say to just do spi vs. emini but the Aussie exchange, ASX, just let in two enormous market makers 2 months ago and they have really stacked the books in terms of bids/offers and gobbling inefficiencies. I am almost certain they spread the Spi with everything they can to form a portfolio of sorts and do something with it. So we need to be smarter and pick spots better to survive (this is why I mentioned that you need to get every combo out there if you're serious about winning... you can't be overtrading with trash every day or you'll give wins away)
Thanks for all your help. I have plenty to keep me busy for a while now. If I'm not pushing my luck, I had two more things I was wondering about.
Assume I get daily data for currencies and daily dv01's for the past year. When I'm doing research, would it be useful to change the coefficient for the spread for each day in the past year or is it not worth the effort?
How do you feel about shorter term spreads like IB v bank bill, Eurodollar v bank bill, bank bill v aussie 3 year? Are these a viable or do I need something that moves a bit more intraday?
Assume I get daily data for currencies and daily dv01's for the past year. When I'm doing research, would it be useful to change the coefficient for the spread for each day in the past year or is it not worth the effort?
How do you feel about shorter term spreads like IB v bank bill, Eurodollar v bank bill, bank bill v aussie 3 year? Are these a viable or do I need something that moves a bit more intraday?
The my thoughts are obviously incorrect, as I know you are using industry standard ways of correctly trading bonds between countries. It's just me sorting though stuff in my head about all this. I'm really going to have to read up on basis points points and comparing bond yeilds between different countries, as I know nothing yet, as illustrated by the following:
I don't really understand how you got the co-efficient as 1/10? ...so I need to read up on it (/find out where to read up on it).
Just looking at it at first glance, in all my ignorance on this subject, if we are just comparing the DV01 between Australia and Canada (?), but we want ticks down the side of the spread chart instead of bps:
HXS DV01:CB DV01 (basis points of one country compared to another)
HXS @ $100AUD :$CB @$102AUD ... so roughly 1 to 1 bps ..
>>HXS DV01 = ~$100 (Aussie Bucks) = $100AUD = 2half ticks/bps*$50AUD/half tick/HXS contract = 2 HSX contracts to make $100AUD/HSX half tick (lets call it a tick) = 1bps.
>>CB DV01 = ~$104 (Canada Bucks) = $102AUD = 10ticks/bps*$10.2AUD/tick/CB contract = 10 CB contracts to make approx $100AUD/CB tick (actually $102AUD) in a tick = 1 bps.
So charting the spread of HXS vs CB, if you wanted ticks down the side of the spread chart instead of bps:
Chart the spread: 2 contracts of HXS - 10 contracts of CB (2*HXS-10*CB), hence each tick down the spread chart would be $100AUD.
...or better...
Chart the spread: 1*HXS-5*CB, hence each tick is now $50AUD down the side of the spread chart
I've really got to stop thinking in ticks and think about bps, and my calcs are probably wrong.
>>>>>>Regarding your co-efficient, if you instead want to trade 1 contract HXS to 1 contract CB, my tiny brain understands:
1*HXC is the same value as 10*CB, if we see the half tick as a half tick, and comparing bps of one country to another,so 1*HXC-10*CB (where the numbers are coefficients and not contracts), so 1/10*HXC-CB, so the 1/10 co-efficient is in front of the HXC and not CB.
or... 2*HXC-10*CB if we see the half tick as really being a tick on the autospreader, which is:
2*HXC-10*CB=1*HXC-5*CB=1/5*HXC-CB , so use a co-efficient of 1/5 (for HXC) if we are spreading HXC-CB, 1 to 1.
...but it now seems like we are now just comparing contracts, and we are no longer comparing the bps between 2 countries.
... just my rambling thoughts. I need the name of a book on this.
I don't really understand how you got the co-efficient as 1/10? ...so I need to read up on it (/find out where to read up on it).
Just looking at it at first glance, in all my ignorance on this subject, if we are just comparing the DV01 between Australia and Canada (?), but we want ticks down the side of the spread chart instead of bps:
HXS DV01:CB DV01 (basis points of one country compared to another)
HXS @ $100AUD :$CB @$102AUD ... so roughly 1 to 1 bps ..
>>HXS DV01 = ~$100 (Aussie Bucks) = $100AUD = 2half ticks/bps*$50AUD/half tick/HXS contract = 2 HSX contracts to make $100AUD/HSX half tick (lets call it a tick) = 1bps.
>>CB DV01 = ~$104 (Canada Bucks) = $102AUD = 10ticks/bps*$10.2AUD/tick/CB contract = 10 CB contracts to make approx $100AUD/CB tick (actually $102AUD) in a tick = 1 bps.
So charting the spread of HXS vs CB, if you wanted ticks down the side of the spread chart instead of bps:
Chart the spread: 2 contracts of HXS - 10 contracts of CB (2*HXS-10*CB), hence each tick down the spread chart would be $100AUD.
...or better...
Chart the spread: 1*HXS-5*CB, hence each tick is now $50AUD down the side of the spread chart
I've really got to stop thinking in ticks and think about bps, and my calcs are probably wrong.
>>>>>>Regarding your co-efficient, if you instead want to trade 1 contract HXS to 1 contract CB, my tiny brain understands:
1*HXC is the same value as 10*CB, if we see the half tick as a half tick, and comparing bps of one country to another,so 1*HXC-10*CB (where the numbers are coefficients and not contracts), so 1/10*HXC-CB, so the 1/10 co-efficient is in front of the HXC and not CB.
or... 2*HXC-10*CB if we see the half tick as really being a tick on the autospreader, which is:
2*HXC-10*CB=1*HXC-5*CB=1/5*HXC-CB , so use a co-efficient of 1/5 (for HXC) if we are spreading HXC-CB, 1 to 1.
...but it now seems like we are now just comparing contracts, and we are no longer comparing the bps between 2 countries.
... just my rambling thoughts. I need the name of a book on this.
Last edited: