Momentum Traders' Construction of Price

Discussion in 'Strategy Development' started by clearinghouse, Sep 29, 2011.

  1. Sorry if I'm posting simple questions lately, but I keep seeing the support/resistance/stops debate on ET.

    So, for example, say we have a boxed range between support and resistance. And let's say we're at the bottom of the range, and we get a momentum signal (like we see a large buyer supporting a level). The ordinary day trader who read a few books on technical analysis might say, "Well, hey, this is probably going to go all the way up to resistance, I'm going to put a limit order just under that resistance and hold my price there because I need to let winners run."

    So the day trader says the "true" market price (M) must be somewhere around resistance, and throws a limit order just under that price.

    But, let's say, in this fake hypothetical universe, on a "bad day", there's some slim chance that there's an algo trader who wants to game the day trader, so he doesn't let the price go to resistance. He decides that at some price, Resistance-Offset, where offset is where the algo trader wants to trade against the buyers, that he's going to just unload into the buyer. Day traders call this a pattern failure.

    So then the day trader has to counter-game the algo trader; he may not get his price and he sees the algo trader has created a kind of new resistance. So the day trader cuts the price sooner, and so you go towards the pattern failure. He may even step ahead of the algo trader and punch down the price noticing the algo trader is looking for day traders like him to take the price up, and so on.

    So in a strict mathematical sense, the day-trader/momentum-trader can never really have a fixed price forecast, even if the price forecast is a mean value. He always has to recondition his forecast assumption based on new information. So, is it safe to say momentum traders can never have a fixed price projection? Say the trader normalized the support/resistance range and did a linear regression on how far into the range the price got and set a fixed forecast. This would just find the mean movement into a range and never really be correct.

    So, for a momentum trader, the price always has to be calculated based on new events. So the process would go like this, from the day trader's perspective:

    Forecast(t0 | Information before t0)= Target price is Resistance-Offset, where offset is just some price just under resistance that is more likely to be hit.
    Forecast(t1 | Information to t0 ):
    if( up to t0, no contradictory price info ) Forecast at T0
    else( if new contradictory information ) Forecast( t1 | contradictory info ).

    So these debates on ET about where to set your stops really can't work on fixed approaches. A fixed hard-coded project can't necessarily work either. No support/resistance algorithm can really work continuously in the presence of smart traders on the other side without constantly reacting to the smart traders on the other side.

  2. A few comments, but first, back to market basics:
    1) Current price reflects all known information; there is no systemic mis-pricing.
    2) no one trader can significantly affect market prices in the long run.

    That being said, everyone thinks the algos have proven the capital market theory as wrong. Algos simply take advantage of price movement in a mechanical fashion.

    Trading against algos can be hazardous since you are effectively reducing your average time-in-trade to theirs. In other words, if your target average time-in-trade is 10 minutes, why in the world would you be concerned about an algo whose time-in-trade is around 15 seconds ?

    It is my belief that average time-in-trade is a really important metric behind any strategy-based system.
  3. Good points.

    Re #1: This is true if you assume the markets are efficient. Even if they are efficient, it's possible to pin trajectories on underlying factors affecting a stock price. They might be accurate, they might not be.

    Re #2: I'm inclined to believe this, but feel as if the situation for large traders is more comparable to no-limit hold'em big-stack situations. For example, a big-stack player in the final table of a hold'em tournament can really really alter the psychology of small players in that he can force a shakeout of marginal hands much easier.

    I didn't want to make any assumptions about the player, but for all intents of purposes we can assume the day trader wrote his own bot. I used a human in my example because this is the sort of stuff humans are talking about on the forums.

    I like your concept that time-in-trade is a useful metric. But, time-in-trade in terms of real time trading has to be adjusted for new information too. As an extreme example, if your average time in trade is 10 minutes, but some guy just hit the book against you because he found out the company behind the stock is going bankrupt, do you still want to adhere to the idea you are usually in the trade for 10 minutes? I think the obvious answer is no.