Statistical Arb/Pairs trading strategy!

Trading Systems Free Forex Trading Systems
202

That’s a very good point.

When Kelton originally started this thread he spoke about using the 1M chart, so that was what I originally started looking at. But my own research/analysis shows that the trades are much longer running than using the 1M chart allows as the divergences can get quite big (depending on the instruments). So I now prefer to use a slightly longer timeframe chart.

Medisoft has also done some research that shows how big the gaps get, although I am not sure if he is continuing to monitor on a 1M chart or on a slightly longer timeframe.

203

I went short EU and long GU at 12:44. Then I closed the trade at 14:13 for a loss on the GU and a win on the EU with a net difference of plus 21 pips. After the FED news both the GU and the EU took a dive as the dollar got stronger. But because the trade is dollar neutral it didn’t matter if the dollar went up or down. what matters is how the EURO and GBP reacts to the dollar, not to each other.

204

Glad you made some pips. The way you worded the previous post, it sounded like you took the trade that was USD neutral, but you profitted out of the USD move after the Fed minutes. Had that been the case, then there would be something wrong with your trades as if you were USD neutral, then your position can’t be affected by the USD…

Nope, you’ve got that the wrong way round.

If you are long GU and short EU, then you are USD neutral and effectively have a short EURGBP position. It is therefore how sterling reacts to the euro that is important, the dollar makes no difference whatsoever. This is why you could have taken a short EURGBP position instead of the two other trades (short EU and long EU).

This is confirmed by the fact that both sterling AND euro fell against the dollar after the Fed minutes. This is directly because the dollar broadly strengthened, but the relation between the euro and sterling remained the same.

205

ya, i agree

206

@ jedster

Please do not take offense but you are not looking at the problem in the right way, which is why you don’t get it. This is a form of quant analysis. it’s big boy hedge fund trading, so i understand how people can not get it. you are right i could have taken a short EUR/GBP position and made money. But that is not the way to look at it. I could have been long the eur/gbp and lost money. then what? but i wasn’t… why you ask? this takes the need of picking long or short out of the situation. had the dollar dropped yesterday i would have still made money. do you not see that? according to your train of thought if that had been the case the i would have lost money by being short the eur/gbp. since i have no need to chose long or short i would have made money either way, up down or sideways. the only ways this system can fail is if the pairs move apart farther than my account can handle, and/or never come back. but with proper position sizing that should never be the case. ( in my own account the pair would have had to move 1000 pips for that to be the case.) but if the pairs are truly correlated then when they split they will come back together in time. the down side to this system is that it have a limited profit range, which is know before the trade is entered.

207

Does anyone know of a way or an app that will allow chart overlaying on iPad?

208

Hi,
I didn’t take offense, don’t worry. If you insult me, I’ll take offense, but you are not doing that, this is just a healthy discussion… :slight_smile:

I have to say though, I am very happy how I’m looking at it, and I totally get this. I was working on the trading floor at one of the worlds largest banks when this strategy was being traded in equities using options. So, I know how it works and I’m totally happy where I am with this…

What you have to do is just think about what you are saying. If you have a position that is USD neutral, that means you are unaffected by the USD. The USD can move up, or down and it makes no difference to your position. Given how you are USD neutral, how can any move in the USD affect your position? If you trade the SGDSEK, how does the USD affect that currency? Or if you trade the CHFAUD, does changes in the USD affect the CHFAUD? No, of course not, even though both CHF and AUD can be affected by the USD, the relation between the CHFAUD is exactly that.

If you take a position long EU short GU, it is exactly the same as long EURGBP. It cannot be anything other than that.

You said:

this takes the need of picking long or short out of the situation. had the dollar dropped yesterday i would have still made money. do you not see that?

No, that is not the case. There is no picking to be done, there are just simple mechanics. If the dollar dropped yesterday then a long EURUSD position will be up. A long GBPUSD position will also be up. My EURGBP position however, will be absolutely unchanged. It has to be that way.

What I was saying in the previous post was just based upon general knowledge of the situation. Now I just took a look at the charts from yesterday, from before and after the Fed minutes. What we actually see, in addition to the dollar broadly strengthening, is a drop in the EURGBP. This means that, at the same time as the dollar strengthened, the euro weakened slightly against sterling. Economically this makes sense (but I won’t go into that).

That however totally explains why a short EU long GU position would make money. It has nothing to do with the USD, but everything to do with the Euro weakening against sterling. In this particular case, it just happened to be caused by a news items that also caused the USD to strengthen, but it could have been caused by any other piece of news that simply affected the value of the Euro against the value of sterling.

209

Try the link that TopFroxx posted…

210

the EUR/GBP really does not matter. I think that is where some confusion is coming in. what matters is how the euro reacts to the dollar and how the GBP reacts to the dollar. that is where the correlation comes in. they both behave in a similar fashion against the dollar. when the EUR goes up against the dollar so does the GBP on a five min chart over the last 50 bars that is currently 84.3% as of this minute. that means 84 % of the time on a five min chart when one goes up AGAINST the dollar so does the other. Then you have to figure out what called standard deviation. since the EURO and the GBP are not exactly the same they will have a plus or minus in the amount of pips the move against the dollar. lets say 100 for ease of numbers . if one moves more than 100 against the dollar and the other does not then there is a break in the correlation for what ever reason. that is what alerts you to a possible trade. because of what’s called mean reversion one can assume that they will get back in lock step with each other and go back to the standard deviation. in terms of traders psychology the euro and the GBP are really one and the same when against the USD. that’s why the correlation happens . Again this system only works when there is a break in the correlation. For my trade yesterday if news had come out affecting the Euro then it would have moved up or down against the dollar,and because the euro is correlated to the GBP it too would have moved against the dollar in the same direction. But since the event would have been about the euro it would move at a faster pace (than the GBP) against the dollar thus breaking correlation and standard deviation. the spread in the standard deviation is where the profit opps. lies. if the euro moved 200 pips against the dollar and the standard deviation is 100 per example then there is an opp to grab 100 pips before the market is in balance again. By being beta neutral the direction of the move against the dollar doesn’t matter because the GBP will move too. That’s why ALMOST always one pair either the euro or the GBP against the dollar will be a losing trade. but the winning side will ALMOST always out preform the loser. it’s there movement as a team against the dollar where the profit lies. That’s why Kelton does not use indicators or support resistance levels or anything else. If the pairs are correlated ( teaming up against the dollar) and you are both long and short the dollar you win. i Could explain the math but it gets long and dry. hedge funds have been doing this for around 20 years now. and trust me when i say the math is solid. it’s why as private traders people often outperform the hedge funds in terms of percentage gains. but there is virtually no risk. it’s almost as if you are both long and short the EUR/USD. minus 20% or so. i hope this explains why there is no need to even look at the EUR/GBP as this system doesn’t care about them only how they team up against the dollar. just like aud and nzd against JPY.

211

jesus, you two guys mean exactly the same but you’re just talking past each other. i’m quite sure that jedstar did not mean that you have to look at eurgbp. he just meant that it’s virtually long eurgbp if you’re long eurusd and short gbpusd, nothing else. you both clearly understand the concept, but just not each other :slight_smile:

212

@topfroxx
lol. i think you are right. but my problem is why go long the EUR/GBP? and what would make you do so? what indicators? and if you are only long in something then you stand a greater chance of being on the wrong side of that trade. why not just go long and short on two things that are 80% the same?

213

Yes, ok, point conceeded. Yes that is what I’m trying to say. I always have been too wordy… :slight_smile:

214

Going forward, I’ll assume we are both right, and are both just seeing the same thing from two different angles… :slight_smile:

Rather than thinking that buying the EURGBP is one trade and you are taking one side, you have to remember that if we go long EURGBP, by the nature of trading currencies we are also going short on GBP.

FXEZ wrote a good post that included a link to a separate article. Take a look.

http://forums.babypips.com/free-forex-trading-systems/43659-statistical-arb-pairs-trading-strategy-15.html#post337536

It explains exactly why a long EU short GU position is identical to a long EG position.

215

@jedster

just a thought what time frame are you using?

216

At the moment I use 2 charts, the 5M and 30M. The 30M allows me to quickly see the last few days of correlation divergences/history, and the 5M allows me to see what is happening now.

217

oh, i get that it can be viewed as a long eg position , but by taking the long eu and short gu you are hedged, thus protecting your capital. i would make money in either direction without having to do anything. where is the protection by just being long eg? should we take this to the chat room?

218

I think it should stay here. It goes right to the crux of whether this strategy should be traded by going long/short with two instruments, or by trading the 3rd currency.

but by taking the long eu and short gu you are hedged, thus protecting your capital. i would make money in either direction without having to do anything

Lets say you are long EU short GU. You are hedged in that you are neutral to the USD, so both EU and GU can rise and fall (in a correlated fashion) and your position remains neutral. You are however, expecting the EUR to strengthen (against the USD) and the GBP to weaken (again against the USD), and at that point, you are in profit and close the trade. If however, the EUR weakens further (against the USD) and the GBP strengthens (again against the USD) then your position moves against you. The further they diverge, the bigger your potential loss as you have no natural hedge in place against this happening.

For me it is exactly the same. Lets say I am long EURGBP. If the EURUSD rises AND the GBPUSD weakens I close my EURGBP trade in profit because the EURGBP must gone up due to the triangular relation between them. If however, the EURUSD weakens and the GBPUSD strengthens then my trade is at a loss. Finally, if the EURUSD and GBPUSD rise together, because they are highly correlated, my EURGBP position will simply remain unchanged.

Check out this post 301 Moved Permanently from earlier. It has a screen shot showing the GU falling, the EU rising, and the EG more-or-less staying unchanged.

219

i understand what you are saying but my question is this. What would make you certain of the market direction in the EUR/GBP? if you are long then you stand a chance of being on the wrong side and getting stopped out. With Stat arb I dont care about direction since either way i would make money. I have already stated that the two (eu and gu) could spread out and never correlate again. very small chance of that happening but a chance none the less or eur or gbp could default debt wise. other than that there is really no chance of me losing unless my position size is too big and it drains the account. but like i said earlier they would have to move 1000 pips to crush me. not going to happen in the time frames i use. 5min. i am currently in the market as we speak, and i don’t expect to lose. it’s only a matter of when can i get out. since there is no stop i just have to sit still during the drawdown phase which i know is going to happen. one must lose for the other to win. and if my math was right at the time of entry i know about how much i can expect to make. i just think that not having to decide long or short makes a little more sense as far as risk management goes. granted by just being long you have a chance to make more pips but it’s less certain.

220

ok. I am certain of the direction of the trade because I do exactly the same analysis of the correlation between the EU and GU as you do.

At this precise moment in time, lets say the GU is strong, compared to the EU. My analysis shows they have diverged sufficiently more than they should do, given the correlation between them. I therefore expect the GU to weaken, and/or the EU to strengthen, relative to each other.

So, I could go short GU and long EU. With a short GU and long EU trade, the counter currency in both cases is USD, so my USD exposure is zero since I am both long and short USD. So, my remaining exposure is limited to the movements between the EUR and GBP. Therefore instead of having the two trades I simply place a long EG trade instead. I am not on the wrong side of anything. The GU has diverged against the EU, and they have diverged statistically more than they should do. My EG trade will be in profit when the EU and GU re-converge.

The same applies to me. It makes no difference which way the GU and EU move, given that they are so highly correlated. if they diverge further then my position will become a loss. If they converge (which is what I am expecting), then my trade will be in profit.

Yep, same here. Providing I can accept the temporary drawdown, then my trade can be left open.

Again, I know pretty much exactly what my target is, based upon how much I am expecting the GU and EU to re-converge.

You say that you are not taking sides, but you are taking exactly the same side as me. The side that you are taking is an expectation that the GU will fall and/or the EU will rise, relative to each other. My position does exactly the same thing.

I am not just long, I am long EUR, short GBP. it isn’t any less certain than if I had taken the EU GU trade, it is literally identical. There are however, advantages doing it this way.

Firstly, the trade cost is cheaper because I only have the spread in one trade. If I have two trades, then there are two sets of spreads pay for. It might only be a couple of pips, but it all adds up.

Secondly, if I am going to leave the trades open for long time (until they re-converge), then I have to consider the financing of the trades. Now this might work to my advantage, depending on what I am trading, but the financing cost is definitely a factor if the instruments diverge for a long period of time.

Thirdly, and very conveniently, I can actually set a TP on my trade since I know exactly how much my target is (directly and mathematically calculated from the divergence between the GU and EU). Thus if the GU and EU were to converge in the night while I was asleep, and then diverge again, that’s fine, my target would have been hit and my trade would have closed for the targeted profit. This couldn’t be done if I had a pair of GU/EU trades, since the movement is relative to each other, so there is no way of knowing at what price they will be at when they finally converge.

221

Jedster do you trade with a stop or just a target? I was having issues with my stops being triggered but the chart showing the price never came close.