Correlation and Leading/Lagging?

Thanks to all for your comments.

Bolter - I really don't worry too much about the specific correlation coeff of the moment. As Mark commented in another thread, it doesn't much matter to me whether it's .90 or .95 or .85. I know such-and-such markets are highly correlated because I can see it in the charts.

As for detrending, I'm not familiar w/it. After reading your post I read some explanations of detrending but I still don't really understand what it means when it says "eliminating the trend". What exactly is being eliminated? I still see trends in the charts using detrended data. Forgive my ignorance, as I said at the beginning of this thread, I ain't real smart, 'specially when it comes to math stuff. I suppose this is a method of more accurately comparing 2 data sets?

I get my daily data from CSI/UA which I import into MetaStock. CSI has a detrending option and I do see a difference between it and the regular charts. The detrended chart does show a clearer divergence in the recent Feb/Mar/Apr lows of the CHF and EUR, w/the EUR clearly showing more strength/demand in the range. I shall investigate further.

Enginer - Again, my ignorance shows as I know nothing of FFT. As a "Wyckoffian" I regard the "actors" rather simply as either "smart money" or "dumb money". Pros vs the public. The pros get in early, buying as the public is selling (accumulation), then the pros get out early, selling just when the public starts buying in earnest (distribution). This is, of course, a very generalized explanation. I look for standard price/volume behavior such as climaxes, ob/os conditions, lack of follow-thru, upthrusts and shakeouts, tests, trendline/channel breaks, support/resistance, and so on.

As for cycles, I've tried them but could not incorporate them into my strategy. This is not to say they aren't viable, everything works for somebody. For me a complete cycle is a rally and a pullback and I look at the relationship of both legs to each other and to the previous "cycles", but that's about it.

Thanks again,
Harold
 
http://tinyurl.com/zhqn6

This is the detrended chart I mentioned earlier, the top pane is the EUR/CHF spread. The mkts enter a trading-range in Nov and both upthrust in Jan and react back to the bottom of the range in Feb. We know from the long-term spread (detrended or not) that the EUR almost always leads, up or down. During the beginning of the range, as can be seen in the spread in the red box, the EUR leads up very nicely. The reaction during the blue box is very shallow on the EUR relative to the CHF and then in the latter part of the range during the green box the EUR is obviously showing significantly more strength prior to the bo. So during the whole of this relatively flat trading-range (when looking at the individual mkts), we can see a subtle but clear stronger uptrend (demand) in the leading mkt, the EUR vs the CHF.

H
 
Quote from bolter:

In order to derive more useful information from correl coeffs you really should be using stationary data - ie: detrended. This is a very common mistake.

The easiest method to introduce stationaratity to a time series is to use differences. That is, subtract the close from the previous close and measure correlation on this time series.

Good luck.
More...
You should understand what that does to data before using it is a filter. Ajacent differences is a high pass filter. Simply taking adjacent differences because it makes data stationary (it doesn't always but for market data it will about 90% of the time) is a very dangerous thing to do willy nilly.

If you are going to be doing this type of analysis, a decent book to have as reference is the Masters book.

nitro
 
Harold,
Detrending eliminates the "drift" in the time series. In essence you are making mean stationary (stable). This gives you a much truer correlation measurement by eliminating the effect of the drift.

Now this may or may not be important for what you are doing, but you should at least understand what is going on.

The easiest way to understand the importance of detrending can be demonstrated in excel. Copy two time series to columns A & B. In column C & D calculate the differences for A & B, ie: close-close(-1). Calculate the correl ceoff for cols A&B and then C&D. Compare the two.

Let us know what you discover.

All the best.
 
http://tinyurl.com/pg5qz

http://tinyurl.com/pts53

http://tinyurl.com/m9dc2

One Daily and 2 60-min charts.

First of all, on the longer-view 60-min chart, note I to J to K price once again traverses the channel, then momentum falters from K to L to I, where price stays in the lower-half of the channel, there's the deepest break of the channel at I, and price is now again hugging the demand line w/o making any progress into the upper-half of the channel.

Looking at the zoomed-in 60-min chart, there is all kinds of convergence of support below price right now - horizontal s/r, the still-rising 20d & 50d ema's, and the channel demand line. But consider the loss of momentum and all the upthrusting bars here, suggesting futher consolidation as the EUR nears significant s/r on the daily.

Whatever, the daily is certainly showing a Sign of Strength in this rally off the bottom of a long trading-range and (yes) back to the top of a previous range.

H
 
http://tinyurl.com/pkfxa

EUR/USD & USD/CHF 60-min:

After a nice intial rally from its most significant break of the channel demand line at J, the EUR creeps back to mid-channel where it struggles there and momentum fails once again at K, while the CHF plummets and becomes severely oversold. The EUR reaction off K culminates in the sharpest downswing in the channel thus far on 3 wide spreads closing on the lows, breaking all kinds of support and back to below the demand line once again. Here one might start to expect some more serious consolidation following a nice uptrend.

H
 
EUR & CHF & Spread Daily (Detrended):

http://tinyurl.com/r4op6

EUR/USD & USD/CHF 60-min:

http://tinyurl.com/f4zls

EUR/USD 60-min PnF:

http://tinyurl.com/m2t9o

Following the breakdown of the trend, the daily EUR consolidation has thus far appeared bullish, w/sideways action on very little retracement, but the CHF seems to be breaking down a bit more, suggesting still further consolidation and perhaps some deeper retracement yet to go, which the EUR/USD 60-min appears to confirm.

60-min Charts: The EUR/USD on the 17th there is minor selling-climax on the 3 wide spreads closing at their lows. This is followed by a lower-high on the rally on the 18th, then the subsequent reaction back to the previous low is a secondary test that takes about half the time and from here we might define a trading-range. On the 22nd there's a test of the lows followed by a nice strong rally back to the top of the range. From here thru the 23rd price clings to top of the range, attempting to breakout but failing. On the 24th there is a 62% retracement at mid-range followed by what appears to be a nice rally back to the top of the range. But on the 24th there is an upthrust, quickly followed by the widest-spread down bar on this chart thus far, closing at the low, a lower-low back to near the bottom of the range. The subsequent congestion the 24th-26th struggles mightily to achieve any price gains (while the USD/CHF clings to the top of its range) and then on the 26th another very wide spread closing on the low, back to the bottom of the range and wiping out that little congestion area in one fell swoop. Meanwhile CHF breaks out on a massive spread and presently is pulling back on very narrow spreads and shallow retracement, while the EUR rallies on narrow spreads and shallow retracement. So looking for the EUR to breakdown from this range and some further retracement.

Note some classic Wyckoff trading-range price behavior: At the beginning of the range volatility is high on wide spreads and wide swings, then around the 23rd (what is presently the middle section of the range) volatility starts to contract, then the 24th thru 26th there is a tight consolidation in the middle of the range, then we start to see some volatility returning as price starts to breakdown.

I've also included a PnF chart of the EUR 60-min. While this is not exactly Wyckoff-style PnF, as MS won't do it, it's pretty close. At this point it certainly appears to indicate supply in this timeframe.

H
 
EUR/USD & USD/CHF 60-min

http://tinyurl.com/ncbq6

EUR/USD 60-min PnF:

http://tinyurl.com/posbm

Ah well, I anticipated the break of the trading-range too soon, as I tend to do too often in my arrogant ignorance. This, of course, is exactly why Wyckoff tells us to wait for a Change of Character in the nature of the price behavior w/in a range, followed by a Sign of Strength or Weakness as one of the safer points to take a position early in a trend.

60-min: Following X on the EUR, where I left off previously, there is a shallow test of the 26th low, then a nice rally once again back to the top of the range. However, the reaction on the CHF is shallow relative to its range lows. The EUR upthrusts and reacts back to the bottom of the range while the CHF rallies to its highs, both sharp trends. This last swing on the EUR is one of the sharpest in the range, from top to bottom, since its beginning, suggesting a possible change-of-character w/in the range as it follows an initial stage of high volatility (17th - 23rd) and then low volatility and consolidation (23rd - 30th) and now may be in the final phase of returning, increased volatility prior to a breakout. I know, still anticipating... Is that so wrong? ;)

The EUR, at support/resistance a couple of hours ago, the next-to-last bar a massively wide spread closing at the high, breaking the short-term downtrending channel decisively, suggesting another trip back to the range highs at least, if not the beginning of a SOS. Waiting for confirmation, or failure to follow-thru, at this point...

H
 
You must log in or register to reply here.
Back
Top