to clarify the 7.34 is dollars that comes to about 45 pips so far this week. I will use this account for awhile it started out at $168.38 this week and now stands at 175.72 those that like math can do the percentage gain on that, I will update my account total as often as anyone wants, I am trading everyday I am lucky enough to have a job where I can do that, not sure how good of a system this would be for someone who can’t trade all day.
This is a great discussion, there is a lot to learn in this thread. I have been interested in learning this type of trading and the unique approaches are enlightening…
Thanks…!!
Since you have profit its good for you my friend, but don’t try to present a slightly amateur approach as a pro. You talk about correlation but you forgot the fact that correl may occur also after a change in the std of the yields’ difference. And as the math is fun for you I think you understand me.
The Board is the only judge…So, keep trade in your own way.
Cheers !!
AWESOME JOB!! now i’m off to get ready for the london session… happy hunting :59:
Cheers Danny
Yeah I opened the trade on the 1min chart and only watched the 1 min chart.
Im just saying I checked the 5min chart because people on this forum were talking about the 5min chart so I just wanted to see the difference in pip separation and noticed the pip separation on the 5 was smaller than the 1, that’s all, but I still only watched the 1min chart. Its obvious I missed the cross so im going to start fresh with a new trade on Monday.
The hourly timeframe, hmmmm interesting, less rescaling/adjusting the bigger the time frame hey
And do you set the scale like Danny does? ^
Miner, I can see you’re coming at this from an academic perspective. If so, you should understand that the proper way to create the new time series is not to difference the data sets prior to subtracting them. It’s not possible to trade price differences, only prices themselves.
Create another time serie for the daily %Δ of EUR/GBP.
(Pay attention in order to compare same day’s results from the 2 series.)Finally, find the correlation between those 2 time series The result will be 1 or perfect correlation. This means that there is no arbitrage opportunity.
As to the second point, the pairs trading strategy doesn’t care much about how far the synthetic (EURUSD and GBPUSD) differ from the cross rate. That would be a pure speed arb strategy which isn’t what is being discussed here. However, you also seem to lack experience dealing with the high frequency time series being discussed. Here is a site that might change your mind about whether any arb opportunities exist intraday: Front Run the Delta. There are more examples if you look for them on that blog of intraday arb opportunities in the currency markets. My own research bears that out.
This is happened because there is the so called “triangle arbitrage”.
Yes, but also because you’re not looking at data that is high frequency to see the pure arb opportunites.
Even in stocks statistical arbitrage is a complicate process which requires the implementation of advance maths, like cointegration.
P.S. Even if you like this way of trading you have to try to distinguish your analysis between testing and trading periods in order to use updated data.
Just saying
Cointegration is one method to synthetically create weights (eigenvectors) that effectively bind the series together and normalize the spread between two or more instruments. Cointegration works better on longer term data rather than intraday data, where correlation tends to be favored.
timehopper, I understand where you’re coming from but I think it is also important to discuss and understand the merits as well as cons of a given strategy. Some may take the strategy as is and do very well with it. Some may decide to get a better understanding of what is going on under the surface by delving into some of the underlying assumptions.
I think it is important to recognize what TalonD found, that over a long period the EURUSD-GBPUSD spread is in a downward trend, primarily due to EURUSD’s relative weakness. This strategy implicitly weights both currency pairs equally (overlaying two charts on each other). If the object of pairs trading is to reduce risk by taking a market neutral position and to capture the spread when reversion occurs, then I think it is important to understand that equal weighting the pairs may lead to a spread that also trends rather than reverts. And there are several approaches (previously discussed) to deal with a spread that trends rather than reverts.
The points about following a system are noted but with the caveat that I would never trade a system that I didn’t understand and agree with all its underlying assumptions.
I can assure you almost every trading system out there works if you adhere to the rules and never change them. That’s how it became a system in the first place . ( think the Turtle traders)
How can you assure us? Have you tried them all?
This is so often repeated on trading forums that there ought to be a snopes entry debunking it. The fact is that most systems don’t work. Most systems require something else - discretion in order to be profitable. There are very few concepts that I’ve found workable, and stat arb happens to be one of them. The turtle system was a trend following system that had been extensively tested and tuned by the developers, but it happened to be a very difficult system to trade live (due to the low win rate and high drawdown). Most systems posted on forums don’t fall into that category from my experience.
I concur. The need to understand the inner workings of a system before going live is undoubtedly very important. My comment was made because it seems at least to me that the Kelton way of doing this is not being discussed. I seem to recall that this system as intended had no need for any other indicators, messing around with the synthetic that’s created or anything else. I think that we as a group should focus on understanding the system as is. And make sure that we get to explain in as simple terms as possible how and why it works. Because it does and as evidenced by some on this blog every time they do it as intended they get positive results to varying degrees. There are some that don’t so me being a team player I would like us to focus on those who are not meeting success and helping them find out why. Then once EVERYONE is WINNING (zero sum game yeah I know) I’m all for discussing Stat Arb as a whole and the varying ways it can be deployed. I thought that this particular thread was about Kelton’s system and if it was good enough to be the system of the month. I must have been the one misinformed.
[/QUOTE] I think it is important to recognize what TalonD found, that over a long period the EURUSD-GBPUSD spread is in a downward trend, primarily due to EURUSD’s relative weakness. This strategy implicitly weights both currency pairs equally (overlaying two charts on each other). If the object of pairs trading is to reduce risk by taking a market neutral position and to capture the spread when reversion occurs, then I think it is important to understand that equal weighting the pairs may lead to a spread that also trends rather than reverts. And there are several approaches (previously discussed) to deal with a spread that trends rather than reverts [/QUOTE]
Again I concur. Make no mistake TalonD and a few others have posted some very interesting theories and I do take the time to research them and implement them sometimes. But even in a trending spread environment the strategy as is works fine. Mean reversion will always be in play while trading and the empirical rule states 99.7 percent of the data will sit within +/ - three standard deviations of the mean. If one wanted they could in fact measure the mean of the spread over a given time get the standard deviation and have an ideal of how the spread will behave and time the entry a little better. Even during a trend. I still fail to see though how the direction of the trend alters this system in any meaningful way. Because up dow or sideways you will still meet with success a good deal of the time. You just have to WAIT. I know nobody like to wait for the profit but still…
[/QUOTE] How can you assure us? Have you tried them all? [/QUOTE]
No, I haven’t tried them all. YET :8: but every system on this site does to varying degrees. Now prove otherwise. I made reference to the Turtle traders because it’s the classic case most of us have heard about. A proven system which made millions for some and not for others in a controlled experiment. The failure was clearly because some of the participants did not stick to the rules. Some people couldn’t deal with the “low win rate… high drawdown” and broke the rules of a PROVEN system. High drawdown is subjective sort of like beauty. Do you disagree? I don’t “Turtle Trade” because of the high drawdown. Still doesn’t mean the system as is does not work.
Maybe someone should start a thread on ADVANCED ARB strategies. Being a lowly newbie myself sadly I cannot. Lets get everyone on the same page first, and focus on helping others grasp this system first. PLEASE
Many of you have far superior math skills than mine but I wanted to share some thoughts and get some input. It appears to me that we have an absolute max profit point which is the center point between the sell and buy. The perfect trade to catch the exact spread difference would be for the 2 pairs to move to the exact center point. Although, as long as the lines meet somewhere between the trade entry lines it appears the profit is decent. However, as the pairs both move up or down out of that range the profit potential shrinks even though they will meet at some point. So with this market neutral position we have a max potential profit and a max potential loss? Or is there never a potential for loss if you wait and close the trades when they come back together no matter how far they go? Is this correct? So the only weakness to this strategy is the drawdown and making sure you do not get a margin call? Thanks:confused:
You’ve got a good grasp of this method. LEAVING THE COMPLEX MATH ASIDE… yes there is for all intents and purposes a “max” profit potential. you are pretty much spot on about how to eyeball that. there is a potential for losses however you can basic round it to zero if you are there for the perfect exit. Now keep in mind you may have a profit of a single pip or less. but that’s fine with me. the MAIN weakness is the drawdown. if you get in and they continue to spread out it can lead to big drawdowns but with proper position sizing that is nothing more than an opportunity to add more to the position. using stops will only hurt you in this method. even today with the 100 pip plus moves in the eur/usd and gbp/usd my entire position is still only around - 7 pips. one side of the trade should mostly offset the losses of the other. I took a very large position a few days ago using the 4h charts. Stat arb is a very safe way to trade, however if this is the only way you trade for an entire year you can expect about the same percentage returns has the hedge fund guys. this way of trading has a LIMITED profit range. that’s why as private traders on a percentage basis we can out preform the paltry gains of major funds. hope that helps.
Thanks timehopper
This is not right. first Assuming pip value on each pair is equal, and you can make them that way by correct position sizing, then it doesn’t matter if they meet somewhere going up or somewhere going down or dead center. Look at it this way. the pairs are a certain distance apart when you enter your two trades. If they start moving apart farther than that distance then you are moving into negative (drawdown) anytime they are closer together than that distance then you are moving into positive territory. It’s the relative distance between them that matters.
Also the touch point is not maximum profit. If they touch and continue moving in that direction then you have even more profit. So for example suppose you go short on EU and long on GU. EU moves down and GU moves up and they touch so you are in profit. But suppose you don’t close your trades and EU continues to move down further and GU continues to move up further and they continue to diverge that way then you have even more profit
The problem with this thinking is what if they don’t continue in that direction but instead, they bounce and start going the wrong way then you profit starts reducing again.
So the touch point is a good way of having a profit target that is something achievable but not too greedy.
I still think it’s better to trade eur/gbp though. that way you only have one spread cost to pay and with the hedge trade you really aren’t hedged anyway, you just think you are. But that’s beatin my head on a wall so never mind.
don’t beat your head on a wall. We need that brain of yours. You are correct the pairs can come back into correlation then continue to pass each other. I would abandon ship at that point because something rare is going on and that’s not an event I would want to be there for. it is more likely the pairs will come into correlation again and move in the same direction. That is why the touch is the profit target.
so you are 100 percent correct. it is possible.
The only thing I disagree with there is that it is rare. I think it is just as likely, roughly 50 50 chance it will cross and go that way as touch and go the other way. If you look at the charts you will see the two pairs crossing back and forth fairly often. Although with EurGbp being in a downtrend recently it may be more likely to find EU line below the GU line than the other way around so not quite 50:50. But I think that trend would affect higher time frames and not be so much of a factor on shorter time frames where price seems more random.
But still it’s good to have a well defined exit point. That’s a problem with some trading strategies, they may have well defined entry conditions but so many of them don’t seem to have well defined exit conditions.
Well, I found a bar with free wifi so, beer in hand, I’ve been looking at he charts again… 
It strikes me that there are two timescales that this can be traded on. The 1M shows several trades a day, in fact I just picked out 8 trades during a period of about 5-6 hours, with each trade giving 5-15 pips. Alternatively, it can be traded on the hourly timeframe where the separation widens to 50-150 pips, so you get one trade every few days, for anywhere between 50-100 pips.
My own preference is definitely moving to trade the the higher (1H) timeframe as clearly everytime we trade on the 1M, there is a chance the pair will just keep on widening and the trade would have to be closed out at a loss…
That would be true on any time frame. I tend to favor the 15M. It seems to fit my schedule best.
What is the best way that you have found to line up your charts? This was my actual question before, but do to my poor delivery I believe medisoft answered another question (By the way, thank you for your response medisoft).
I have seen that :
[B]The original way was to just open up the chart and start trading[/B]- This presents issues with pips per pixel and also seems to fall short when there are dynamic moves on one currency. (One currency chart will be shifted what may seem like plenty of pips to open a trade, but is later reviled to only be several pips wide, not even enough to cover the spread.)
[B]Using the ATR of each currency pair to fix the scale[/B]- which results in uneven pips per pixel.
[B]Setting a fixed scale with even pip values[/B]- this seems to be the best way so far. However, it still presents issues. It would seem that there would be a problem with this method if the pairs are not currently “together”. One has to decide a reference point to base the, lets say 100 pips (50 each side), or at least a reference area. This would place each currency pair in the center of the window, but if they are not currently together it would appear as if they were. (Please correct me if I am wrong)
I understand that most of this is relative because on the larger time frames they may have deviated some time ago and have yet to make the move back to “equilibrium”, which doesn’t really matter for other time frames. All that really matters is the time frame you are trading. But, one would still like to be on the right side of a trade and see when the pairs are actually deviating.
I understand that , again, this is all relative because even if they appear separated, when they are in fact even, they will eventually separate in the desired direction again, but this could cause trades to be ran for longer periods of time.
Yes it can. However, if you’re trading the M1 time frame and the spread widens out to the normal M60 amount, there might be a problem with the risk/reward somewhere.
Thanks for your input Talon. I have been racking my brain to try and figure out why the profit seems to disappear. I have 2 charts from same broker overlayed on 1m time frame. I should have made more profit when I go back and calulate the pip moves on each individual chart. So what I am seeing with the charts overlayed is not the real correlation picture. Maybe adjusting one of the charts up or down or throwing up an indicator such as ATR on each chart and see if that gives some signal to help view the real picture? I will continue to dig deeper.