Correlation and Leading/Lagging?

Discussion in 'Technical Analysis' started by hcour, Dec 9, 2005.

  1. hcour

    hcour Guest

    I'm curious as to what ya'll think of my reasoning here: I'm trying to understand correlations relative to how one market may lead another. If two FX pairs are very highly inversely correlated then wouldn't one look for the abnormalities, where they become much less correlated, as the points when one pair actually starts leading another? Because how much could any one lead the other during a high-correlation cycle? Wouldn't the highest correlations normally occur during a trend and the lowest during a trading-range/consolidation? Could the intial breakdown of that high correlation possibly portend the shift from trend-to-range and the subsequent reversal back into that high correlation signify a cycle from range-to-trend?

    I'll try to post a chart illustrating this soon. Just wondered if I was on the right track...

    dratsum likes this.
  2. Enginer


    For 'correlation' read 'herd mentality.'

    We're trying to anticipate where the herd is heading, or will head next. Obviously, at a turning point emotions prevail. I wonder if the consensus improves or weakens? Yes, that should show up in a change in correlation index.

    I am looking for software that does the same as grabing two (or more) T&S swatches and pasting them in a Excel XY chart with time as the X axis. Correlation is easily seen visually, what I am looking for is consistent lags - to use one as the trigger for action on the other. I can imagine easy and powerful ways to do this, and because it is so obvious it must exist somewhere. Have you bumped into anything reasonable...?
  3. correlation is always backward looking.
  4. Enginer


    Correlations come and go (improve and dissipate). The real question is can they be found in real time:

    1) with long enough existence to profit...and

    2) can they be anticipated beforehand, as in "70% of the time rrr has a correlation index of .45 (great for raw data) with sss and a lag of 3.5 seconds (tradable). I have seen correlations in the e-minis of >.65 with lags of 30 seconds or so.

    "I have seen" means from a Excel spreadsheet 'way to late to be of value. Yes, the correlation only lasted an hour or two, but that's ample time for a trade or so :)

    p.s. The next level to 2) is to find a technical indicator that anticipates the likelihood of a strong correlation developing.
  5. I have caught a few trades with the EUR/USD and the USD/CHF successfully. I have not had time to get into the research of why and the measurement of the WHEN to do.. Sunday evenings reveal a lot of opportunities in other pairs also...but I do not know why yet.

    Thanks for this thread.

    Michael B.
  6. How do you measure correlation? I want to write a computer program that analyzes correlations.
  7. hcour

    hcour Guest

    Forgive all the lines.

    As David Letterman used to say: Is This Anything? Michael mentioned EUR/USD & USD/CHF. This is a daily chart w/an 8d correlation. I think it supports some of what I'm talking about. In the first part of the chart each market is coming off a long-term trend and they climax and consolidate and reverse in pretty much classic trading-ranges. Note during this period from 12/04 to 3/05 there are several sharp relatively lower periods of correlation (red lines at -0.80ish), then in Mar when the trends get in gear following lower-highs (USD/EUR) and higher-lows (CHF/USD) and the ranges break, correlation remains steadfastly strong at or near -1 thru July and beyond.

    In July the first serious pullback/consolidation of this trend, but is this reversal or continuation? Hindsight is 20/20, but note the still strong short-term trends which continue to correlate almost perfectly. Does this mean the longer-term trends will continue? They do briefly but then in Nov correlation starts to show the first real volatility since the trends began and price subsequently is certainly ranging more, even in the short-term it's much sloppier as it tries to break significant support/resistance. Then last wk serious price breaks suggesting shakeouts and upthrusts.

    So is correlation giving clues as to reversals? Will this work on an intraday chart, hourly or 15-min? Could one look at correlation aberrations as clues to standard technical analysis of trends/consolidations/reversals of higher timeframes which could be played in lower timeframes?

    Just wondering, just exploring...

  8. Norm


    Is there a commercial software package that looks for correlations?
  9. So what do you do to make money when the 2 markets diverge? Sell the leader that makes new highs because of "non-confirmation" by the lagger? Buy the lagger for a catch-up play? Sell the lagger/buy the leader for a relative strength spread? IMO, just more excuses to put on a position and/or explain away a loss. :)
  10. Hi Hcour,

    You are so close to figuring out the most reliable price action I know that can be traded for consistent profits.

    The answer to your above questions...

    Yes, Yes and Yes.

    They can be found in realtime because I do it many times per week as a price action only trader and you don't need any indicator nor special software...

    All you need is the ability to think outside the box.

    Simply, you are going in the right direction and I will leave it at that (no further explanation).

    Good luck.

    (a.k.a. NihabaAshi) Japanese Candlestick term
    #10     Jan 10, 2006