What am I not seeing?

Discussion in 'Strategy Development' started by dpg2020, Jul 1, 2006.

  1. dpg2020

    dpg2020

    I'm not a newbie, nor I am an advanced trader. I'm looking to expand my trading setups to include a 'pairs trading' setup. In doing some research, on the surface, it seems surprisingly simple but, as in most things in life, I know it cannot be quite so easy.

    I give you an example. Let's take, for the example, the following pairs, SPY and MDY. During every day, there is a time at which these ETF's will diverge by approximately 0.2% and then go back to even with each other. The same will occur with say VGK and FEZ. It seems that the real problems are the following:

    A. The ETF's move against your original position such that you are stopped out.
    B. Lack of liquidity in one or both of the ETF's does not allow you to exit your position.
    C. Needing to trade volume in order to allow for sufficient profit.

    Let's also say that one know that average maximum divergence for the MDY/SPY pair intraday is 0.20% occurring approximately 70% of most days.

    Would it not be profitable then to simply take this trade every day it occurs once it sets up with a 0.20% stop?

    What am I missing?:
     
  2. Transaction Costs and Bid/Ask Spread
     
  3. cashonly

    cashonly Bright Trading, LLC

    The illiquidity is what you may be missing. Would you be able to just take the offer or hit the bid and still get that same .2%? Then it may be ok... but if you have to buy on the bid or sell on the offer, to get that, then it could be problematic. And you shouldn't need a stop for this type of trade... if it goes against you, I would think that you would layer it, using historical extremes as your max loss.

    Cash
     
  4. jrlvnv

    jrlvnv

    Cash when you layer a trade to add the same amount of shares to both sides? Example... 500 long SPY 500 short MDY.... moves against you... Add 500 shares each?
     
  5. cashonly

    cashonly Bright Trading, LLC

    Quote, yes, add the same amount and when it comes back to where you put on your prior layer, take it off, taking your profit. Repeat as needed. THe real trick is determining the price difference between layers so you don't have to put on too many, cause if you do, and it never comes back, you could be really screwed... thus the importance of historical extremes.

    Cash
     
  6. jrlvnv

    jrlvnv

    how far back in time would you go to find your historical extremes?
     
  7. There is never any garantee that a correllation will exist in the future. There may come an event where this correlation is destroyed for ever.

    Add to this that the bigger players are engaged in automatic arb'ing (and in case of the banks let the retail traders cough up the spread....)

    In other words: not a game for the retail player.
     
  8. cashonly

    cashonly Bright Trading, LLC

    Yeah, that's where the art of it comes into play. No pat answer there.

    When you do it with stocks, you throw in fundamentals, so with indices, you might want to throw in some technical indicators like MoneyFlow, VWAP, MA crossovers, stochastics, MACD, etc to determine if you want to be on that side for that index. That would increase your win rate.

    But eventually, you will have something go against you and you'll have to bail with a multi-level loss. You just need to do it on enough "pairs" so that you have some diversification.

    Cash