Buy Dips / Sell Rallies

[michaelguan326]

Hi T2W forum members,

I have been in the process of developing a system; as the title suggests, it aims to buy dips in a strong uptrend and sell rallies in a strong downtrend.

Currently, I aim to trade the the following markets with the following system...Gold, Silver, Corn, Soybean, Wheat, Crude, Heating Oil, RBOB, Nat Gas, Coffee, Cotton, Sugar, Live Cattle, Yen and Pound

Firstly I define the direction of the market with 2 SMA, short and long. Out of the top of my head, let say the short is 20 and the long is 50. When the short crosses above the long, we only look for long entries. And vice versa for shorts.

Secondly, I define a strength of the trend with a ADX above 20. This is important because we only want to catch dips in strong uptrends or else there are going to be a lot of whipsaws.

Third, I define a dip by CCI crossing below the 0 line. This is a important setup!

Fourth, the entry trigger is a 3-5 day breakout your choice.

Exits: i tested both 20 day trailing stop and a volatility stop pair each with a stop of the high/low of previous last 2 days

Above the the rough idea of the system i am developing. This is the first time ever that I looked in to these type of systems but they will definitely help in diversifying my strategies. I open up for discussion as with my little experience in these systems, I don't know how to improve it. Currently I am just fiddling around with the entry parameters which I don't want to do as its sort of curve fitting.

So my humble question to fellow members who have experience in these type of systems is....

1. are they suppose to be designed to trade a wide variety of markets?
2. i am experiencing <30% winners which is worst then typical trend following systems, is that normal or should systems have high % winners?
3. i have a great number of losers....which i am absolutely fine with, but the winners aren't bagging enough money
4. any further advice with regard to these systems would be great!


Thanks for reading,
M

 

[tomorton]

Systems like this (the classic example is the moving average crossover, but MACD, or stochastic oscillator reverals are similar) do have low win rates, but 30% sounds terrible. You will find they all give a high percentage of failed signals and the key is to take profits early. The problem is that taking profits early means the winners are small.

Where are your exit and stop levels?

 

[michaelguan326]

Systems like this (the classic example is the moving average crossover, but MACD, or stochastic oscillator reverals are similar) do have low win rates, but 30% sounds terrible. You will find they all give a high percentage of failed signals and the key is to take profits early. The problem is that taking profits early means the winners are small.

Where are your exit and stop levels?

I have a volatility trailing stop paired with a low of last two day as my initial stop.

I want to implement the system to diversify my strategies,,, but having trouble figuring this kne out. I gave it more a thought and found that actually these systems are just another type of trend following system but that they have different entries, which makes then correlated differently with ur classic TF systems.

In Experimented with MACD to define dips but then they are too slow, and get int oo late...but I wanna have a closer look. Are you a system trader too?

 

[Adamus]

hi michael

interesting post, nice to see someone start a thread like this.

first thing you need to be clear about is what your experience tells you to aim for or otherwise, whether you are still trying to find out what's possible.

do you want to ride the trends, building up your position size with the retracements or do you want to do something else?

I'd go with this approach: pick a target where the price would mean to you that your entry mechanism had worked nicely - now take this price distance and use it as the target, use the same in the opposite direction for a stop, and then run a test to see what your win:loss ratio is like.

considering you've already added a filter, you should have a reasonable hit rate.

try it on very long time frames, like weekly or monthly bars. do you get a decent hit rate there? if not, then it's probably not a decent filter or ma combination.

regarding trading a variety of markets, then look at it this way: if your system worked profitably on all markets, then it's more likely to carry on working in the future, isn't it? if the system only works on one market, considering the non-specific nature of your indicators, then you have to ask yourself why.

but all markets are not the same. you can feasibly adjust the system to each market as long as you don't curve-fit it as you are afraid of doing. you can see evidence that you have curve-fit from the results of your system across the spectrum of whatever you are varying, i.e. MA length. chart the results on the y against the parameter value on the x - if you get a smooth curve, it's hopeful - if you get what looks like an earthquake seismograph, you have curve fit it.