Statistical Arb/Pairs trading strategy!

In reference to my previous post, I was going to post this indi a week or so ago and provide a better explination, but I was unable to figure out how to actually upload a indi or create a hyperlink to another site. It kept saying invalid file or something.

[B]Please don’t miss my previous post. I have been ignored on this thread before lol. [/B]

I promise, using the indi will change your perception on stat arb in forex.

Yes… The Pip weights are all the same this way… Gbp is more volatile then eur… Aud is more so then nzd… So by making sure the two volatile pairs are working to oppose each, I get a more friendly behaving basket.

The basket that isn’t balanced for volatility is
(gbp + aud - eur - nzd)/4

If someone will instruct me on an acceptable file ext for an image I would gladly upload an example.

Here is an example. The bottom is the indi and the top is A/N.

In the indi the yellow line is the basket profit line, the blue line in the middle, and if you squint real hard you can tell there is a green line just above price. The top green line above the profit line (yellow line) is the +2 st dev line.

There are minor differences between them but keep in mind, it is approx 75pips from ±2 st dev to the mean. So red line to blue line in the middle is 75 pips. So when I say minor differences, I mean MINOR differences. These differences mean one or two pips.

Also, this indi does accommodate for the spread of each pairs.


So richard, are you still trading this way? How have you adapted your system to deal with the times when A/N may be trending?

Honestly I don’t.

That doesn’t mean it couldn’t be done. I have just kind of moved on. I haven’t given up on the system or A/N. I am just branching out at the moment.

If you download the indi you can see that just earlier this week there was a great opportunity to bag about 70 pips. Price, and by price I mean the basket profit line which is an indication of A/N position, was slightly below -2 st dev. As one can see it is now just above the 1st st dev. If you had held price from the point the indicator showed A/N to be below 2 st dev (or outside of the 95.4% of where price has been over the last 5000 5 min periods ie. 17 days) and held it you could have bagged a 100 pips or so. Price actually moved from -2 std to +2 std so you could have made much more if you were really aggressive. But of course, hindsight is 20/20, you wouldn’t want to make that a practice of holding price as it bounced between the 2 std’s.

Most of my trades took about a week to come to fruition. I didn’t just use A/N you could have just as easily set the two pairs in the basket pairs as G/U G/J and traded U/J if it is ranging.

The system still works, I’m just not working with it at the moment.

I’m still keeping an eye on the possibilities. I just got burned so I’m staying out and watching for a while.

The way I was trading Stat Arb was really bollinger bands on crack lol.

There are indi’s out there that are relatively good at showing trending pairs. If you use them on higher time frames like the daily.

Now this is more interesting!

Here’s the basket’s performance over the past 3 years, from 1/1/2010 to 31/7/2012 using daily data.


I should mention, the center red line is just a linear regression to the data, and the two dashed lines have been offset vertically by ±0.45 .

Let’s just say a RSI-like indicator, within the context of it’s natural ranges… works pretty darn good on this basket :wink: Nearly all of the breakouts of the ranges were the result of major economic news headlines/releases/policy changes… If my RSI type indicator is indicating a turning point, the price is on a range extreme, and there is nothing on the fundamental side that supports a breakout, it usually takes a marquee event to do so, then it’s a nice, safe swing trade with r:r of 1:4-8.

The above chart should demonstrate how much more obvious the levels of support and resistance are. Good for swinging… And on a small time frame, scalping for 3 or so pips is easy as pie on the days where no major news releases are scheduled for any of the pairs, as long as you have some way to open and close all the pairs in the basket instantaneously.

All good ideas!

Oops made a mistake when I calculated the basket and took the difference of the first two pairs with the second, not the average.

Had a quick look and taking the average works too. Will try and post a picture later.

Have you tried other baskets?

Oh yeh… You can kind of tailor the behavior of the basket by which pairs you use… Experiment , just make sure you are aware which pairs have differently weighted pip values.

Hi Richard, this is my first post but I’ve been following this thread from the beginning,

I was wondering what settings you were using for SMAPeriods/StDevPeriods on the indicator you are talking about? Tried 5000 and some more settings but I can’t get it to show like yours.

Thanks


If you need anything else don’t hesitate to ask.

Going to test it. Thanks. Hope that if this ultra-synthetic pair is limited to range trading it could be traded using simple oscillator like rsi or stoch
But let me test.

By the way, i closed all my baskets today for about 0.6 %

Did anybody took any trades today.?? UK PMI spiked GBPUSD but not much activity on EURUSD. Did anybody traded that price difference.??

No today is very slow. Chears!

I took positions and profits on EU/GU and AJ/NJ

I’m starting testing AJ/NJ with the same settings as for EU/GU

12 % and counting since august.

I started measuring audjpy/nzdjpy audusd/nzdusd audchf/nzdchf eurjpy/gbpjpy eurchf/gbpchf and cadjpy/audjpy

This using my lock at center method.

Great results Medisoft. I’m having good results too 9% total for last 4 weeks. I am trading eu/gu on the 1 minute chart. I was looking at your MYFXBOOK. What timeframe are you using, because I noticed your trades are opened for a much longer time than mine are opened. Thanks

the same as you, 1 minute charts.

I’m simply wait for more profit per trade, less trades, better risk-reward relation. That way a loss doesn’t means lots of winners, but only 2 or 3.

I want to say that this method is FOMC proof, non-farm pay roll proof and weekend proof :slight_smile: