you know what I’m going to say here don’t you… 
If you trade it using the one EG trade, rather than the synthetic, then you can set absolute TP points. It’s great going to bed and coming down the next morning and seeing that your position closed at the project target while you were sleeping. Sets you up for a great day…
Pipcompounder,
First it’s good that you have found this thread. Much success in your future trades.
To address your question… Yes, it’s true that I don’t use stop losses. It’s funny you ask because the trade that I’m currently in, (check earlier post from previous two or three pages) has had a drawdown around 200 pips ( or very close to it) at one point. The GBP side gave me a good whack. What I most often see it that one side will be down 200 pips or so and the other will be up a good deal to offset some or most of that DD while the trade is in progress. For instance the trade I closed this morning was down 130ish on one leg and up 190ish on the other. I closed both and took a profit of the difference. But again I cannot stress enough that my money management allows for a 1000 pip spread. Which on the time frame I use I have never seen. I use the hourly for the most part and check it maybe 3 times a day. (less time sitting in front of the computer) AND THAT IS THE KEY. THE PAIRS WOULD HAVE TO MOVE SO FAR APART THEY NOT ONLY BREAK CORRELATION THEY GO FROM POSITIVE CORRELATION TO NEGATIVE. The odds of that happening suddenly without any warning I place at 1 in 1000. ( LOL ) If I were in your shoes I would go back a few years and look at the biggest spread you can find on the timeframe you use and do a multiple of that (like x3 or x5) Then I would take that info and adjust my position size so that the spread would have to be like 5 times bigger than anything seen in recent years to do some real damage to me. anything less and I put more money in…
Yes, a little bit of calm, try not to get too far ahead 
. This is a very steady strategy where you often have to wait around quite a bit, however yes you do need to be careful, there can be large divergences.
If you look at this on a long term basis, they were going nicely for a good portion of 2011, and then from October/November have been steadily diverging. Currently around 880 pips!
(direct link http://i.imgur.com/PHwGF.png)
They still appear to be well correlated, but the divergence is definitely widening.
There have obviously been many opportunies to trade this pair since then, however, had you opened a trade in November and used the philosophy that you would leave it open until it would be in profit, not only could it be sitting nearly 900 pips against you, but you would have 6 months of financing accrued against you.
And I didn’t even need to go that far back for this one, I am sure if you look further into history you would see even larger divergences…
This is very true…IF YOU OPENED THE TRADE USING THE DAILY TIME FRAME… look in the top right corner of the chart posted and you will see that it’s the daily chart.
KEEP IN MIND THAT WITH EACH TIME FRAME THE MATH AND THE WHOLE SYSTEM IS DIFFERENT. that is why whe can see many trading opportunities on the smaller time frame that can and are opened and closed for a profit several times a day using the 1, 5, 10, 15, min. charts. Using the 30min or hourly and above you are starting to get into territory where you may have the trade open for a few days. but certainly not months unless you use daily charts/weekly charts…
Yes, agreed, that’s why I said that there have been many opportunities since October/November. However, that chart clearly shows that had someone opened a trade around that time, and used the philosophy that they wouldn’t close it until it was in profit, then they might currently be sitting on a 900 pip loss…
And like you said you didn’t have to look that far back into history to find such a big spread… Kinda why I don’t trade using the daily charts. I don’t have the commercial sized account…
THAT’S what I’m seeing, Timehopper…somebody, try this, since I can’t use overlays myself, limited at work, heehee, open 1min charts, zoom all the way out…note the spread…now wait 1 day…open two more other charts, but of same pairs, same 1min timeframe, same everything…i’ll bet you you get different distances for that same noted time period, now 1 day ago, because it’s all relative! just trade what you see! what I see is a rubberband pulled taunt horizontally, then time pulling the upper half up and the lower half down …entry time…then time makes the rubberband relax back until the two ===== halves are almost touching-TP!!!
theory for medisoft: (again!) if you just use your “indicator” that copies the pixel location of one pair over the other, couldn’t you just program EA to place trade when they are so many pixels apart? Even scale in 1 order every 50 pixels (or whatever, for example) and close when within pretty close to touching? On 1 min timeframe? Is this simple enough to demo just to try? I wish I could program-i don’t know squat about writing code- “banner command?” -oops I’m aging myself!
I tried that, but was soooo difficult, and it didn’t work very well. That’s the reason I switched to “the mean”.
And don’t think I’m very young heheheh, I also know the banner command.
DOES THAT indicator really work as good as overlaying? if so, would you mind posting it?
I just restarted the tests with all variables different on my EA. Including the NO balance/rebalance of trades, that is equal size for both trades. Let’s wait for at least a week to see how that moves. I think this will give us good light about the idea of balance/rebalance the trades.
Looking forward to the results!
Also, testing the reliability of potential charting app on my iPad…did any body who had a low on EU at 1.31463 that correlated with GU low on 1 min chart at 1:33 GMT? And is EU now lower than GU? At 2:20 GMT?
Just closed another winner alittle while ago on AUD/JPY and NZD/JPY and added to USD/JPY and CAD/JPY position which widened further from my intitial entry. Happy Trading!
I has been thinking about a better way to rebalance positions. I finally come to the idea that I want to give more weight to the side that is winning and remove the weight to the side that is losing, if possible without taking loss.
The basic idea is this:
One can be on two stages of a move, a trending move or a ranging move. If the trade is trending, after a while one side will be with profit while the other side will be losing. If in range, both sides could be in profit or in loss, but the move should be small.
The use of rebalancing is specially important when trending, because one side can be moving faster than the other. Let suppose that both EU/GU are moving in a downtrend, but GU is moving faster and it is on loss, while EU is on profit but a lower speed. If I never rebalance and the downtrend extends for a long time, the divergence (and loss) between both pairs will increase. If I had done good money management, my account should support that drawdown and end with profit when the pairs revert the trend, but I could be waiting a long time. But if instead of that, I add to the winner side (EU in this example) and try to reduce the losing side (if it has various trades and one of them suddenly becomes profitable, I can take that part to reduce the risk on that side). This way, EU will continue moving a lower pace, but the profit on EU side will be growing faster than the loss on GU side, if the move in pips is GU=2EU, but I set the size of the trades to be EU=3GU that will make that the profit generated from EU side will grow at a speed that it will reach and exceed the loss on GU faster, and will close the trade in less time for profit.
I’m testing this strategy and adding to my pool of MT4 tests hehehehe.
By the way, this is the current status of the tests.
Alpha 0 is balacing with ATR only first entry (no rebalance)
Alpha 1 is balacing with ATR with rebalancing
Alpha 2 is balacing with Price1 and Price2 only first entry (no rebalance)
Alpha 3 is balacing with Price1 and Price2 with rebalancing
Alpha 4 is balacing with covariance/variance using pair1 as an index only first entry (no rebalance)
Alpha 5 is balacing with covariance/variance using pair1 as an index with rebalancing
Alpha-NOBETA is not balancing, but placing same lot size on both sides, without rebalancing
MEDISOFT, IS THE AJ\CJ PAIR STILL better this way? But, if using smaller lot size to scale in as divergence widens in EU/GU pair, like Kelton did, (20,40,60 MAX of 6 trades), with total risk of all 6 trades adding up to the $100USD total in, shouldn’t this be livable? It’s profitable, no question…but about the AJ\CJ pair, if it still shows only up to 60 pip max divergence with more trade opportunities, I would need to figure out the extreme limits of risking an account to figure out what’s reasonable to trade live.
Using the technique of SMA for the MEAN this are the max divergences in pips
For EU/GU, 186
For EJ/GJ, 170
For AJ/NJ, 140
For ACHF/NCHF, 104
For UJ/CADJPY, 158
This are all the pairs I’m trading, until I find how to trade AUD/CAD and AUD/USD, and other pairs where the common currency is the first one.
Ignore the 60 pip divergence I wrote earlier, it was based on the pixel by pixel technique that was not so successful. The SMA as the MEAN is what I’m doing and I need to say that it is very successful. Like timehopper says, no losing trade since I started trading with this technique, and I have about 40 trades done in two weeks.
Where the common currency is the first one and you’re attempting to compare you might simply invert the prices
Price1 = 1/AUDCAD
Price2 = 1/AUDUSD
Invert = true
This will make them both have an AUD denominator, then you can apply your system’s regular analysis. Just remember that when it comes time to place the trades, since the prices are inverted you may need to reverse the trades (long for short and short for long) to get it to work out right.
if (Invert) {
// opposite trade sequence
} else {
// regular trade sequence
}
ARE you, MEDISOFT, also going to try demoing a strategy where the trade closes 2.5-3 pips or so close to the mean? for more trades?
I do not believe this is going to help, and I really don;t think you shoudl be trying any form of re-balancing as nothing has really got out of balance. If you are trading this with a synthetic, then you know that generally there will be one position that wins and one that loses. Of course, it might be that they both win, obviously that can occur, but generally they will finish with one win and one loss. So, that has to be taken for granted. If you start altering position sizes midway through a trade, or, start hedging one of the sides of the synthetic, based upon other factors that are affecting the price so that you can try to cut the amount of the losing trade, I think you are just heading for problems. You are forgetting why you have the two positions in the first place (so that you are market neutral), and altering that means that you are no longer market neutral.
Additionally, if you have a synthetic trade and they are diverging further (albeit with one in profit and one not), you don’t know which instrument is taking the exceptional move. In a down trend, where both are falling, but the weaker instrument is falling quicker than the stronger instrument, all you know is that the divergence is widening. You don’t know if it is because of the weaker instrument or the stronger one. By changing the position sizes you are trying to second guess which one is weakening and which is strengthening.
As you probably know, I believe that the synthetic should have equal values for both sides, so you shouldn’t be taking different position sizes in the first place. Trying to tinker with them further in the middle of the trade is, well, definitely something I wouldn’t be considering So, it’ll be interesting to see how your tests appear in the long term.
has anybody checked out the forexticket site lately and put in EU\GU? the sample period size gives very different results for every variable, and this pair looks only correlated above 0.80 on weekly…how do we use applicable period sizes on this site? Also is it real time? then you could just trade off this site…there must be a way to program or replicate the chart they have that shows correlation…it would be perfect, just open trades when far away, close when pretty close, back and forth…also there is a certain site out there I found while googling around and it offers signals for stat arb for a fee, they say 12-15 trades a month and trades hold for average of 12 days each, and they claim their last trade netted 1366 pips! but I’ll bet it’s a big time frame with relatively high DD, so I’d rather have 2 to 3 trades for 2-3% a day, compounding.
I’m not sure I understand where the problem is? If you have two correlated instruments, you just buy the weaker and sell the stronger.
For EURUSD/GBPUSD, lets say the EURUSD is weaker and the GBPUSD is stronger. You buy EU sell GU, (or just buy EG).
For the AUDCAD/AUDUSD it is exactly the same thing. Lets say the AUDCAD is weaker and the AUDUSD is stronger, you buy the AUDCAD and sell AUDUSD. You are long the AUD from the first trade and short the AUD from the second, so you have zero AUD exposure. Your net position is thus a synthetic long USDCAD position (or a synthetic short CADUSD if you want to think of it that way).
