Reverse engineering why indicators don't work

Discussion in 'Technical Analysis' started by IronFist, Nov 27, 2017.

  1. People say buy when price makes a new high. Examine it logically. Every time price makes a new high it has to make a recent high. Figure its previous high was 70. For it to go to 80, it has to go past 70, so getting some at 71 allows you to follow it up. But it might not go that far.

    People say buy when this indicator does whatever. When price creates a crazy trend that indicator will give that signal, but just because that indicator shows a signal does not mean price is going to keep going.

    Very important points.
    Oliver11 and lcranston like this.
  2. jharmon


    Simple: Backtest it.

    Everything else about a theory of how the market works, or how an indicator works, is pure speculation.
    d08 and Handle123 like this.
  3. lcranston


    This view was popularized thirty years ago at the beginning of the bull market, largely by Wm O'Neil, though Nicolas Darvas also benefited from this approach due to his having the wind at his back during the 50s. But people forget how and why this stuff originates and it all becomes "common wisdom". Fact is that important money doesn't buy at new highs; it buys at lows. In fact it is the important money that creates the lows in the first place. 2009 is an excellent example.

    As far as "when to buy" is concerned, one is better off understanding price behavior -- since that illustrates trader behavior -- than relying on indicators.
    johnnyrock likes this.
  4. tomorton


    People do say buy a new high - but they don't like to call it that. They like to say buy the break-out.

    Then these don't buy high buy the break-out merchants criticise me for buying in an uptrend and that its obviously too late to get in so I'm going to be buying high when I really should buy low - but which is not what they did anyway these knuckleheads. Grrrr!
    Xela, ET180 and lcranston like this.
  5. lcranston


    Apparently they don't understand that the important money doesn't buy breakouts; the important money engineered the breakout in the first place. The breakout is their opportunity to sell at least some of what they have in order to profit from what they'd bought. Which is why Wyckoff discourages the beginner from buying breakouts.
    Xela and tomorton like this.
  6. If one were to backtest a bunch of indicators over 5 years of data here is what they would find. I know this because I did it. Every indicator has some degree of lag. This is because the indicator needs two or more samples to return an indication. This is not a problem if markets continue in the indicated direction. This is also not a problem if an entry indication is not triggered.

    Were this is a huge pronlem is the number of times an entry is triggered and it turns out to be a weak move. By the time the indicator gives the entry signal most of the move is gone and you take a loss. These add up and usually create negative returns. If we look at each of these failed trades and base the entry on price action we would have got in sooner. These trades would have then been a mixture of small profit, wash trades and occasionally small losses. The result would have been a positive PNL long term.

    I am not anti indicator. Its just that I did the backtesting on this for another project and I know the outcome.
    ET180, digitalnomad and lcranston like this.
  7. Handle123


    First mistake is you are listening instead of doing, you take what people say works and test it, when it shows it don't work, find patterns of when it does work as most do have certain patterns or times of positive outcome.

    2nd you spend way too much time on the forum and little time backtesting, you think knowledge s going to be had by reading in these pages? You have to get down to business if you truly want make this a business.

    A broken clock is right twice a day, based on only looking at it every twelve hours at exact time, it is 100% right, whereas by the minute or second, horrible wrong.
  8. When you are testing anything remember to keep net profitability in mind. For some people testing an indicator means just testing its directional accuracy. Let me save you the trouble, they are all a coin flip (50% directional accuracy over the long term). This is why the term "high probability trade" is misused by people who cant even tell you the historic probabilities of a "set-up". Testing for returns is a better test than directional accuracy alone.

    Longer term trend following (1 month and beyond) is done with things like Simple Moving Averages. If you try these same simple strategies out on a shorter intraday time frame then it boils down to bid/ask spread and tick ratios to intraday ATR movement. You can get chopped up badly. Strategies that show a Gross Profit will go to being a Net Loser once you factor in bid/ask spread for market orders. If you can conservatively assume a Limit Order Fill then the erosion wont be quite as bad. Expect about a 15% to 20% erosion if you use Market Orders even half of the time. Corn and 10 Year Bonds go to an absurd 40% and beyond. I am just speaking about my entry signals. There is no way to avoid these costs intraday even if you trade just once or twice a day in a given product. Imagine backtesting the Emini S&Ps intraday with just 1 trade a day (1 entry & 1 exit) market order for 200 days. You should factor in (2 ticks * 200)=-$5000 per contract. Just an example.

    It really depends on what you are trading and what time frame. You must factor these things in because it is easy to find fools gold when back testing.
    Last edited: Nov 27, 2017
    trendo and digitalnomad like this.
  9. Overnight


    Thus the need for an atomic clock, like the one in Colorado. If it breaks, it is right twice per day down to the millisecond. It's the nanosecond stuff that goes awry. *hides*
  10. eurusdzn


    I dont know what big professional money does from my recliner but I would think a big volume
    Breakout (as advertised ) is the last thing a pro would want to operate in.
    The big volume on a day bar is probably thin liquidity at all the prices (volume at price indicator?) for the large volume day bar. From there it is probably not at all the market structure that pros want to use as price is marked up to the new value level.
    Might as well venture a guess that a trend may continue and buy the base of the previous area
    where great liquidity is provided by those that fear an uncertain future.

    Sometimes clean , low volatility, channel behavioir appears well below a previous high and lasts for week(s). Breakouts from these channels in direction of trend sometimes can be seen and are well off the recent high. Breakouts dont have to be to new highs. They may be a changein a recent pattern.

    Does Whckoff draw lines at the single high and low prints.? As a generalist, amateur I view these ranges,boxes,S/R not as single price/line levels but as areas with a centerline and approximate tops and bottoms.
    All my attempts are to operate with a methods in these areas with the intention to position for the next markup.
    #10     Nov 27, 2017