my journal

[Yamato]

EUR-GBP-USD triangle ARBITRAGE

I woke up thinking of arbitrage. If anyone wants to discuss arbitrage or semi-arbitrage or arbitragish trading on the EUR-GBP, let me know, whether we'll do it here or in private. I got some bad ideas in my head...

Ok. I've done some research. Here's a good link on what they call "equilibrium" and "oscillator" formulas: Forex Chartist.

What do I have in mind?
Ok, so let's keep things simple, the way I always do. What is it that I want to do and that I'm calling "arbitrage" and how right am I in calling it "aribtrage"?

I noticed on charts that EUR and GBP move together. In other words the exchange rate of the dollar versus GBP and EUR moves in the same direction. Now, if, on my intraday chart, I see that the EUR has risen a lot and that the GBP has unexpectedly fallen or zigzagged, I want to bet as I always do that the EUR will bounce back down, but also that the GBP will rise. I will therefore proceed to go LONG on the GBP 2 contracts and SHORT 1 contract on the EUR (because of their different leverage).

If I am right on both bets, I will make money on both. If I am wrong, as often happens, and the EUR keeps on rising, I will lose money on that for sure. But if the EUR keeps on rising, how likely is the GBP to keep on zigzagging or even falling? Almost zero chance. The GBP will most likely rise as much as the EUR if not more, because it has to catch up its delay from not rising all day long. So the GBP will rise, and I will make money on it, because I am LONG 2 contracts. What I will lose on the EUR will be outweighed by what I'll gain on the GBP. So in this case I will make money as well. So I never lose? Right. Unless those two prices keep on going where they have been going so far, then I won't lose at least. But they are very unlikely to keep diverging for so long.

One day I should test and automate a system on this, but, before doing that, I need to understand exactly what I am doing. I am quite positive that it makes money.

How close is my idea to the concept of "arbitrage"?

Now let's see what wikipedia says about the "arbitrage" concept and see if I am right on track and if what I just explained can be called "arbitrage". I will quote the lines that I understand and that agree with my concept from the arbitrage page of wikipedia:

In economics and finance, arbitrage is the practice of taking advantage of a price differential between two or more markets...

So far so good.

Arbitrage is possible when one of three conditions is met:

1. The same asset does not trade at the same price on all markets ("the law of one price").
2. Two assets with identical cash flows do not trade at the same price.
3. An asset with a known price in the future does not today trade at its future price discounted at the risk-free interest rate (or, the asset does not have negligible costs of storage; as such, for example, this condition holds for grain but not for securities).

Not good. None of the three applies to my idea. So maybe what I have in mind cannot be properly called "arbitrage". And yet that book I talked about earlier does talk of arbitrage done on EUR and GBP, speaking of "two corresponding USD pairs" and "EUR-GBP-USD triangle".

Idea for oscillator and tradestation testing
As I was reading it, I just came up with a good first concept for a simple oscillator, also good for tradestation testing: you start at a given time of the day and measure the ticks change on EUR and GBP. How many ticks are they moving apart? Does the book say more about it? Maybe. Maybe the entire book is about the oscillator. But this little idea I got, these 3 lines I read out of the whole book, are enough to build my whole system and automation on it, so I won't go any further.

By the way, the key to testing and automating is simplifying your concepts - that's why I am so good at it. Because I spent my life explaining things to myself and others and I did so by simplifying everything, and when I need to explain things to the computer, I am already in that simplified explaining mode. Maybe because I am stupid and I need simple things, or maybe because I am intelligent and I want to understand everything. Maybe I could say that I am a hard-working dummy. That's why I like those "for dummies" books so much, and wikipedia itself is a "for dummy" encyclopedia. But I am also intelligent: to simplify and to synthesize you need to understand something first. But I am definitely slow. That's why I can write all this - otherwise my thinking would get to the answer so fast that I couldn't manage to write down my ideas and my train of thought, as they would flow too fast to be identified. These really intelligent people I met suck at explaining things.

Anyway, despite the fact that wikipedia is a "for dummies" entry, let's at least read half of it, and report whatever else is close to my idea:

Arbitrage is not simply the act of buying a product in one market and selling it in another for a higher price at some later time. The transactions must occur simultaneously to avoid exposure to market risk, or the risk that prices may change on one market before both transactions are complete.

Ok, this "simultaneously" concept is again close to my idea, because that's how I would go about it: simultaneously.

Examples
Suppose that the exchange rates (after taking out the fees for making the exchange) in London are £5 = $10 = ¥1000 and the exchange rates in Tokyo are ¥1000 = $12 = £6. Converting ¥1000 to $12 in Tokyo and converting that $12 into ¥1200 in London, for a profit of ¥200, would be arbitrage. In reality, this "triangle arbitrage" is so simple that it almost never occurs. But more complicated foreign exchange arbitrages, such as the spot-forward arbitrage (see interest rate parity) are much more common.

Wow, this tells me three things. First thing is that, once again, what I am planning to do on forex is not exactly arbitrage. Second thing is that everyone is very organized in this field and I won't have a chance to succeed at it even if I do it. Third thing is... there were just two things.

My idea is more like arbitrage + overstretched (oversold/overbought) concepts. Or maybe my idea is exactly arbitrage and the fact that such fast computers are already doing it doesn't keep it from being profitable for me. Yeah, because these superfast computers are doing it immediately, within a few seconds. But I am talking about trades that last a few hours. So we're both talking about arbitrage but in different timeframes. No wait... we're not talking about the same thing at all, and my trading idea is NOT arbitrage, according to this wikipedia entry so far. I won't buy and sell and realize an immediate profit. Arbitrage is 100% guaranteed profit or similar: my thing is just a trading system based on divergences between EUR and GBP being cancelled. But my idea is totally arbitrage according to the book. This means that, just like for "automated trading" and "algorithmic trading" (which could either mean the trading I do or the high-frequency trading banks do), also "arbitrage" has two different meanings.

But let's keep on reading, to see if wikipedia mentions at all anything close to my idea (which is, once again, a high % of wins with trades that last a few hours, rather than 100% of wins with trades that last a few seconds, as in this example: "Converting ¥1000 to $12 in Tokyo and converting that $12 into ¥1200 in London, for a profit of ¥200, would be arbitrage"). Ok... I'm gonna keep reading.

One example of arbitrage involves the New York Stock Exchange and the Chicago Mercantile Exchange. When the price of a stock on the NYSE and its corresponding futures contract on the CME are out of sync, one can buy the less expensive one and sell it to the more expensive market. Because the differences between the prices are likely to be small (and not to last very long), this can only be done profitably with computers examining a large number of prices and automatically exercising a trade when the prices are far enough out of balance. The activity of other arbitrageurs can make this risky. Those with the fastest computers and the most expertise take advantage of series of small differentials that would not be profitable if taken individually.

Ok, here once again we're far from my idea, and close to high-frequency trading, which has nothing to do with what I wanted to do and nothing to do with what is within my power to do.

Sports arbitrage – numerous internet bookmakers offer odds on the outcome of the same event. Any given bookmaker will weight their odds so that no one customer can cover all outcomes at a profit against their books. However, in order to remain competitive their margins are usually quite low. Different bookmakers may offer different odds on the same outcome of a given event; by taking the best odds offered by each bookmaker, a customer can under some circumstances cover all possible outcomes of the event and lock a small risk-free profit, known as a Dutch book.

Yet another example of how far I am from what they intend by "arbitrage", that is "risk-free profit". That's exactly the point actually: I am not talking about risk-free profit. I am talking about a regular system, like all my other systems, looking for a trade that maybe will make money, or that most likely will make money, or, to be even more precise, a trade that probably will make money, but not certainly, as this wikipedia entry has said of arbitrage so far.

Ok, I've finally found it. This is it. This is what I meant by arbitrage and "arbitragish": they call it a "modified form of arbitrage". Here it goes:

Some types of hedge funds make use of a modified form of arbitrage to profit. Rather than exploiting price differences between identical assets, they will purchase and sell securities, assets and derivatives with similar characteristics, and hedge any significant differences between the two assets. Any difference between the hedged positions represents any remaining risk (such as basis risk) plus profit; the belief is that there remains some difference which, even after hedging most risk, represents pure profit. For example, a fund may see that there is a substantial difference between U.S. dollar debt and local currency debt of a foreign country, and enter into a series of matching trades (including currency swaps) to arbitrage the difference, while simultaneously entering into credit default swaps to protect against country risk and other types of specific risk.

Potential Disaster in my type of arbitrage

Oh, look, there's more about my type of arbitrage:

Another risk occurs if the items being bought and sold are not identical and the arbitrage is conducted under the assumption that the prices of the items are correlated or predictable. In the extreme case this is risk arbitrage, described below. In comparison to the classical quick arbitrage transaction, such an operation can produce disastrous losses.

Ok, I read later, by their examples that after all risk arbitrage is not used for what I am planning to do, and that maybe the closest concept is Statistical Arbitrage. Now let's go and read that whole entry, and talk about it in my next post.

 

[Yamato]

Here is my next homework:
Statistical arbitrage
Pairs trade

My next search will be:
http://www.google.com/search?q=Pairs+trade+statistical+arbitrage+EUR+GBP+USD

There's money to be made in this field. I am going to do this last feat seriously and then I'll retire from system development. After this I will just focus on self-defense, and on exercise. I gotta get in shape. Too much sitting has ruined my body. Too much abuse has gone on for too long. From now on there will be 50 pushups each morning, 50 pullups. There will be no more pills, no more bad food, no more destroyers of my body. From now on will be total organization. Every muscle must be tight.

 

[Yamato]

Wow, I started my homework, by reading Statistical arbitrage, and I am quite happy with what I am reading, because I am understanding it and because this wikipedia entry is giving me a name for the idea I came up with, yesterday, while looking at the chart. An idea of a trading that is close to "ideal arbitrage" but it is not quite the same as "ideal arbitrage":

In the field of investing, statistical arbitrage refers to attempting to profit from pricing inefficiencies identified with mathematical models. Statistical arbitrage attempts to profit from the probability that prices will move toward a historical average. Unlike ideal arbitrage, statistical arbitrage has risk.

Let's read more:

Academics versus hedge fund industry
In the world of finance and investments "statistical arbitrage" is used in two related but distinct ways:

1) In academic literature, "statistical arbitrage" is opposed to (deterministic) arbitrage. In deterministic arbitrage a sure profit can be obtained from being long some securities and short others. In statistical arbitrage there is a statistical mispricing of one or more assets based on the expected value of these assets. In other words, statistical arbitrage conjectures statistical mispricings or price relationships that are true in expectation, in the long run when repeating a trading strategy.

2) Among those who follow the hedge fund industry "statistical arbitrage" refers to a particular category of hedge funds[3] (other categories include global macro, convertible arbitrage, and so on). In this narrower sense Statistical arbitrage is often abbreviated as StatArb. According to Prof. Andrew Lo, StatArb "refers to highly technical short-term mean-reversion strategies involving large numbers of securities (hundreds to thousands, depending on the amount of risk capital), very short holding periods (measured in days to seconds), and substantial computational, trading, and IT infrastructure".

Ok, for sure the idea I had is not related to StatArb (#2), but only to #1. I'll go home now, and read more later or even tomorrow. For now, let's take a break. I can't wait to get done with my 43rd system on Statistical Arbitrage and finally call it a wrap and focus on self-defense.

 

[Yamato]

Ok, good. I came home and thought I had seen a good opportunity to test my new statistical arbitrage system, and immediately went long on the EUR and short on two GBP, and after 30 minutes or so I am losing 200 dollars. It's very very interesting. I thought the GBP had risen too much and the EUR had fallen too much, but instead the GBP just kept on going up and the EUR started rising but wasn't rising as fast, and as a consequence I am losing 200 dollars now. But the experience was worth trying. It is just as they say it in wikipedia:

Unlike ideal arbitrage, statistical arbitrage has risk.

I am losing 200 dollars still, but it was a great experience: a new exciting feeling, to be able to lose money even with arbitrage. What a great trader I am!

 

[Yamato]

Ok! Now it's looking better. Only losing 100. For today I'd be happy if I could make 100 out of this trade. If it ever gets to +100, I'll close both positions. Now the really interesting thing is that whatever system I think of never works in my own hands, but then I backtest it, automate it, and it works. However, I am hoping that this will be the first discretionary method that works in my hands.

 

[Yamato]

Something else I am noticing as my position evolves is that 2 GBP is not equivalent to 1 EUR. Two GBPs are more powerful in terms of leverage, because the GBP does move more than the EUR in terms of ticks. So what happens is that if my choice of 2 GBP SHORT was right, then I'll make even more. But if it was wrong, and GBP goes up as EUR goes up, even if they rise equally (in terms of % of each future's daily range), I will not break even but will lose slightly. On the other hand I do make more money, if I happen to be right, so I should start such statistical arbitrage EUR-GBP trades at a reversal time, like at 8 PM CET, in order to increase my chances of being right. But then I don't know (without backtesting) how much that will interact with other (maybe more) important factors such as the divergence in ticks between in GBP and EUR. Most likely this last one will be the most important factor and I will leave out all other parameters, also in order to avoid over-optimization.

Another important remark is that my constant urge to pick tops and bottoms may actually prove useful in this type of trading, whereas it's always brought me losses and blowing out accounts in all other types of trading. Yep, because if I am right, I make money, and if I am wrong, I have an insurance, a safety net, which will certainly keep me from blowing out my account.

 

[Yamato]

Ok, now I've read both entries:
Statistical arbitrage
Pairs trade

IN the meanwhile I am losing 600 dollars (I've opened a bunch more contracts, always paired), and I am understanding that I can never outperform my systems, as I simply lose every time I try, no matter what strategy I use. Also, the GBP is now making money, and the EUR losing it. They've both started falling.

I will now move on to:
http://www.google.com/search?q=Pairs+trade+statistical+arbitrage+EUR+GBP+USD

You know what? Most of the links on this search page suck. I have learned enough and will figure out the rest myself. My next step will be backtesting. Not next weekend but the one after it.

The position has been evolving in the meanwhile. For a while the EUR and GBP were both going against me at once, and then they reversed and they were going both in my favor at the same time, to the point that I am now only losing 45 dollars. This has taught me a lot, but mainly this: they do not move together at all times. They could even go in opposite directions.

Ok, I've closed both positions (EUR and GBP) at an overall profit of 150 dollars. This would have never happened with just one trade. This system has a long way to go. I will definitely get a profitable system out of this.

Ok, I am half way between 11k and 12k, which is really really good for me, and Wednesday is up next, which is the most profitable day of the week. I wish I could end the week with the much desired 15k, but I'd be ok with just 13k.

I've also avoided an approaching compulsive trading binge by opening a bunch of trades in opposite directions, thanks to this new statistical arbitrage method. It was great because I've felt the excitement of trading real money without actually risking very much, and I ended up actually making money. It felt really strange and a new feeling altogether: fun with trading, but no guilt nor losses.

 

[Yamato]

Damn! Damn! That sick maniac, my dad.

Deo-dorant! Deo... dorant! Not using it for a few days because it's irritating my armpits. I don't care if I smell.

Damn! Damn! That stupid smart cleaning lady. She keeps on putting away my slippers...

She. Let's hope she calls me again. I miss her.

Anyway, here's the deal:

Back-testing on TradeStation the "EUR GBP pair statistical arbitrage" system
(to be tested in the second half of November 2009)

timeframe: 15 minutes
Data1: EUR
Data2: GBP

What the hell is going on?! Is that an airplane? I don't care: with money, I'll fix all my problems.

"How far apart from each other" parameter:
Now... you can't use absolute ticks because that won't be good across 10 years. So we'll do, for each symbol, price of now minus yesterday's close (remember to exclude those hours for they are continuous futures) and we'll get the distance in ticks from the close. And then we...

RELATIVIZE the distances by dividing them by each symbol's close
Say the EUR closed at 1.5000 and now we're at 1.5150. That's a difference of 150 ticks, but it's not good enough. We can't just say EUR +150 and GBP +100, then let's go long 2 on GBP and short 1 on EUR. That won't work in 1999 when their prices were entirely different. And forget about using range at all: that will be too complicated and probably useless and therefore harmful. But we do need to (webster online sucks) "relativize" those ticks. So we divide them by yesterday's close. That's 0.0150/1.5 = 1% move up. The GBP only... NOT 0.66% up, but rather 0.0100/1.7 = 0.59%. Then we...

GET the distance from one another by subtracting them
1% minus 0.59% = +0.41% (optimize everything here)...

With a positive difference (if optimization says it's big enough) we'll go long 2 GBP and short 1 EUR, and viceversa.

On top of this, we must absolutely test a "minimum range" requirement, on both symbols, to see if it makes sense to trade on days with low range, which I would expect not, but it could be the opposite.

DONE. See you in 2 weeks for the testing of this. Oh, and most likely, the exit will be a time exit. Enter whenever, exit at the end of the day.

 

[Yamato]

Holy cow, if I had stayed in my two discretionary statistical arbitrage positions I would have made a fortune by now: the EUR rose a lot and the GBP didn't move at all. But man (travis) doesn't have the capability of sitting tight. He only has the capability of being right.

And right here let me say one thing: After spending many years in Wall Street and after making and losing millions of dollars I want to tell you this: It never was my thinking that made the big money for me. It always was my sitting. Got that? My sitting tight! It is no trick at all to be right on the market. You always find lots of early bulls in bull markets and early bears in bear markets. I’ve known many men who were right at exactly the right time, and began buying and selling stocks when prices were at the very level which should show the greatest profit. And their experience invariably matched mine – that is, they made no real money out of it. Men who can both be right and sit tight are uncommon. I found it one of the hardest things to learn. But it is only after a stock operator has firmly grasped this that he can make big money.

Or after his system has grasped it.

 

[Weighbridge]

I noticed on charts that EUR and GBP move together. In other words the exchange rate of the dollar versus GBP and EUR moves in the same direction. Now, if, on my intraday chart, I see that the EUR has risen a lot and that the GBP has unexpectedly fallen or zigzagged, I want to bet as I always do that the EUR will bounce back down, but also that the GBP will rise. I will therefore proceed to go LONG on the GBP 2 contracts and SHORT 1 contract on the EUR (because of their different leverage).

You made the point elsewhere that the EUR-USD-GBP triangle is already being kept efficient by well-equipped arbitrageurs.

A consequence of this is that if you buy GBP/USD and sell EUR/USD in matching proportions (say, $100,000 of each) then you are creating a synthetic short position in EUR/GBP. You might as well save on commisions by simply shorting EUR/GBP instead.

When GBP/USD and EUR/USD move in tandem it shows that the move is dollar driven. Sometimes it will be and sometimes it won't.

If GBP/USD & EUR/USD made a good pairs trade then the EUR/GBP chart would be completely rangebound, with any wobbles up or down reverting back to a central line. However the chart isn't like that, it's like any other currency - sometimes trending and sometimes rangebound.

So unfortunately I think this is a bit of a dead end, or at least no different to playing EUR/GBP directly.

 

[Yamato]

Thanks for the feedback. I read your curriculum and I was expecting your opinion. I'll let you know if anything good comes out of it. I will be done with it within the next 3 weeks. From Nobel prize to Nobel prize (both co-winners for the drawdown formula), I think you're underestimating me. After having built 42 profitable systems, if I spot something that seems to be profitable, most likely I will find a way to make it work. And this time it seems even better than usual. But I am glad to be underestimated. I'd be alarmed if it were otherwise, I've been underestimated my whole life. And that's what kept me working hard. I think I will start playing dumb in a while. I want to focus on acting and have fun to see people's reactions when they think I am totally dumb. That's where I'm going soon, the picture here below. I'm going there. Give me another few months, and I am going there. Maybe it will be reckless but I am going there: my island.

 

[Yamato]

 

[Weighbridge]

Thanks for the feedback. I read your curriculum and I was expecting your opinion. I'll let you know if anything good comes out of it. I will be done with it within the next 3 weeks. From Nobel prize to Nobel prize (both co-winners for the drawdown formula), I think you're underestimating me. After having built 42 profitable systems, if I spot something that seems to be profitable, most likely I will find a way to make it work. And this time it seems even better than usual. But I am glad to be underestimated. I'd be alarmed if it were otherwise, I've been underestimated my whole life. And that's what kept me working hard. I think I will start playing dumb in a while. I want to focus on acting and have fun to see people's reactions when they think I am totally dumb. That's where I'm going soon:

I don't doubt your ability to create good systems, you've got years of positive results to back them up. I bet you've created a few bad ones as well though, so it's not like every idea is guaranteed to be a winner. It works because you filter out the good from the bad.

I thought my post might give you a real lightbulb moment, but it hasn't. Either I haven't grasped what you're trying to do, or you haven't grasped what I was trying to say.

I'll be happy if you make a winner out of this of course.

 

[Yamato]

Hey, maybe we could talk on msn about this. What do you say...?

 

[Weighbridge]

Sure, I'll PM you my address mate.

 

[Yamato]

I've added you but you don't seem to be online. I wish to introduce you to other people I've met on my journal. It's ok, I most likely would have kept you up all night talking about trading systems, and we would have sent each other to hell tomorrow morning, dead tired, at work. Let's meet tomorrow afternoon. Much better.

 

[Yamato]

You made the point elsewhere that the EUR-USD-GBP triangle is already being kept efficient by well-equipped arbitrageurs.

A consequence of this is that if you buy GBP/USD and sell EUR/USD in matching proportions (say, $100,000 of each) then you are creating a synthetic short position in EUR/GBP. You might as well save on commisions by simply shorting EUR/GBP instead.

When GBP/USD and EUR/USD move in tandem it shows that the move is dollar driven. Sometimes it will be and sometimes it won't.

If GBP/USD & EUR/USD made a good pairs trade then the EUR/GBP chart would be completely rangebound, with any wobbles up or down reverting back to a central line. However the chart isn't like that, it's like any other currency - sometimes trending and sometimes rangebound.

So unfortunately I think this is a bit of a dead end, or at least no different to playing EUR/GBP directly.

Look, Weighbridge, I have to say it. It seems that you were right. I didn't read carefully yesterday. Once again, like for the drawdown formula, I am glad to find out that I am surrounded by people who are smarter than me. However, just like for the drawdown formula, I now need to think about this new discover for a few days before I can fully grasp it.

So basically this is the deal. If the EUR goes up and the GBP goes down (I am talking about the CME forex futures here), and you want to trigger trades based on that discrepancy, then you might as well observe, test and automate everything on the "RP", which is the EUR/GBP CME future. That way I would save myself spread and commissions. It sucks that I am not smart enough to understand all the implications of this and that I'll have to think about it for a few days. I didn't even understand when he first told me. Well, apologies.

However, one question does remain. If this idea is so useless and dumb as you have convinced me to think, why is that guy spending a chapter to talk about it in his book, the book I mentioned, "Forex Chartist Companion"? I will definitely need to spend some time on looking into this. If you have time, take a look at it, too, and tell me if that's also a useless chapter. Thanks.

 

[Yamato]

No no no no... hey, I got it! How could you miss this part of my system? Aside from the fact that the RP is more like a R.I.P. future, in the sense that it seems to trade very rarely, and this is a major problem, the whole thing we'd be missing out if we only traded only the RP, and looked only at EUR/GBP relationship is that we'd missing out on where they both stand with regards to the USD. I need to know that as well, because I am entering when one of the two is oversold/overbought with regards to the USD. So I am going to go ahead with all my planned testing, but I could definitely trade it on just the RP, if I can figure out and solve the volume problem. But I need to test it on the USD relationship as well. Probably you were already discounting this and were only talking about trading it. Anyway my next homework is reading the whole chapter of Forex Chartist Companion on this triangle, because I may indeed be a bit dumb when it comes to forex relationships.

 

[Yamato]

OK. I've read a bit of that book, and it says nothing of statistical arbitrage: it's talking in fact of ideal arbitrage, where you have risk-free profit. So now I have Weighbridge against me, the book against me... but still I think that there's something my original strategy that makes it valid and worth testing. However, I will now look on the internet and see if anyone else is talking about this thing, because otherwise it may really be nonsense. Someone, on elitetrader or here or somewhere on the internet must have talked and thought of this thing before me, or it will probably mean that it's useless. So here's my next homework:

http://www.google.com/search?q=Pairs+trade+statistical+arbitrage+EUR+GBP+USD

 

[Yamato]

Ok, Weighbridge, let me get closer and closer to my point as I see it appear clearer and clearer. What you said was: if you wait for a big distance between EUR and GBP and then go long on one and short on the other you are creating a synthetic position on the RP (EUR/GBP future), so you might as well open a position on the corresponding future and save yourself spread and commissions. Correct.

But the problem is that, even though I did not say this, I not only had in mind a big distance between EUR and GBP, but also a state of being oversold/overbought with regards to the US dollar. Maybe I can't explain it correctly and that's maybe because I do not understand it correctly, but I sense I have a point somewhere. I sense this system has some sense. What I mean to say is this: I want to go LONG on the EUR future if it's oversold and if it's also much more oversold than the GBP. It needs to be overstretched both towards the dollar and the pound: and then and only then I am betting on a reversal. So I do need both, don't I? I can't cut the dollar out of the picture, can I?

Now I ask myself: why do I need, at that point, to also go short on the GBP? Why don't I just go LONG on the EUR? To protect myself. But wouldn't I make more money simply by opening one position, in the long run? I'll have to test this. Certainly it wasn't my original idea.

I can't make sense out of everything, but the fact is that I think this strategy makes money. It made money yesterday. If this strategy makes money in back-testing (and it might be difficult to do so), then I should use it, regardless of the fact that I don't even fully understand what it's doing. I wish I could find someone on the internet who explains it to me.

My next homework is this:
forex statistical arbitrage - Google Search
FX statistical arbitrage @ Forex Factory

I think on the second link there's another nut who came up with my idea: actually I felt like I was reading myself. That's coming up next, once I get done with some work here at my office.