When developing past systems I have always done so using Tradestation and have used the feature they offer to optimize entries/stops/exits. Now I just started keeping track of a new strategy on an excel worksheet. After 100 foward tested trades I now have enough data to come to some type of conclusion. How does one correctly optimize? for stops do you use the median or the mean of the total # of points trades went against you? what about for exits? thanks!!
I use two types of stops in my strategies, the first is a dollar based stop, this allows me to define the maximum loss. I will also you a stop based upon the current VOL in a particlular instrument. The stop that causes me to lose the smallest amount of money is the one I use. One other comment, specific stocks or markets may require an adjustment be made to the stop strategy for that stock or market, I conduct additional research to verify this.
Hello: If you have data that indicates a maximum adverse excursion, you should ask yourself if you can afford to place your stop at that point. This relates to your minimum profit requirement per trade. If for instance you are looking for a minimum of 3 points, you may or may not want to expose yourself to a loss larger than say 1.5. To make this decision you need to factor in win ratio and account size. As your win ratio goes up, your stop loss can also be adjusted upwards. If on the other hand you use either mean or median, you will by definition "give up" some portion of your possible profits, and depending on the optimization procedure you may still be exposed to a significant "per trade" loss. One problem with opmitization is that it usually does not take local volatility into account. I would try to add a factor for changes in volatility otherwise local and seasonal changes in volatility may cause you problems. Good luck Steve
To optimize an ATR based stop. X =Optimize; Ue a range 1 to 5 * your average range for a period, intraday or end of day. Y = Optimize; Optimize for number of bars intraday or # of days end of day. Stop = X * ATR(Y);
Why not scroll through these 100 trades and see the nature and the extent of the price action against your trades upon entry? It may take longer, but don't you think that you will get a better feel for the size of stop you will require for the good trades in relation to the surrounding price action than, say, the summary sheet of an optimization run? The stop loss size that makes the most sense for a method need not be the optimized one. Further, your reference to means and medians* in this context suggests that you are trying to extract more specificity from your analysis than you are likely to get insofar as actual, real-time trading is concerned. Finessing works far better on historical data than it does on "forward" data (i.e., actual trading). You may have tested your entries going forward, but this data is now historical insofar as your stop analysis is concerned. * Using the median of ordered values will lessen the effect of the outliers on your analysis. If you are going to use one, then you should use the other as well, since if their values are significantly different then this information has implications regarding the shape of the distribution of the observations. Personally, I would use neither in the context of this discussion.
I did alot of data mining and asked myself the question, what is the $ stop that works best? What keeps me in the trade so that I have a good chance of capturing the move, but stops me out when I'm wrong? There are many ways to do this, I did mine by going through each trade and studying the details. No substitute for hard work, IMO. It certainly pays off in terms of the improved results and the increased confidence you will have in the strategy. Hope this helps!
Personally, I think that dollar stops miss the point. I think that stops should be solely based on the volatility of the price action surrounding your entry. How you choose to measure that volatility is up to you. There is no single way that is necessarily superior to others. It depends on, and is related to, your overall method. I think that dollars only come into play in assigning how much financial risk or exposure you wish to incur once you know the point size of your chosen stop. Then it is simply a matter of determining how many contracts or shares that dollar amount enables you to trade based on your stop size.
If you have the data in Excel, just use their solver. But first you have to decide what to optimize (for example, sharpe ratio).