No registration required! (Why?)

Time Horizon to Target

I hear of many traders whose targets are never reached (although they should have been without threat of their stop being hit) because they sabotage their trades by mismanaging the Time Horizon.

The Time Horizon to Target is the amount of time that you would reasonably expect to elapse in order for the market to reach your target.

Let's take a look at the E-mini S&P500 contract. The average range for this contract during Regular Trade Hours (RTH) is about 12 points. We can probably say that on average the time taken between the extremes is about half to three quarters of RTH so about 3 to 4.5 hours. (These are rough numbers used for this example.) Let's use 4 hours to make the calculation easy. So, on average, we say that the ES moves at a speed of 12 points per 4 hours or 3 points per hour (PPH). Just like in a car, we can drive faster or slower and reach our destination sooner or later. The same will happen in the ES.

Remember that this is very much an average and just like travelling in traffic you will have periods at full speed and periods when you are in a jam and moving nowhere.

Now say that your target in a particular setup is 6 points. At 3 PPH it's going to take your trade about 2 hours on average to reach the target, assuming that it is not stopped out and does actually reach the target.

The problem that a lot of traders face is impatience. They have not calculated their average Time Horizon to Target and will scupper the trade prematurely only to watch it go to and exceed the target (without threatening the stop) in the subsequent time frame.

A more sophisticated approach to the Time Horizon to Target is to calculate the average time taken for each particular setup to reach its target when it was successful. This can be done instead of using the average speed at which your market moves at, in our case the E-mini S&P500.

During back testing you can add a dimension to the output of the test results showing the time taken to reach the target or get stopped out. If you then plot these values on an XY chart you will see clustering of points around the average time to target (as well as to being stopped out). This is also useful in allowing you to gauge which strategies should be given more or less time to play out and when you may be justified in cutting your trade short.

Remember, the objective here is to try and keep you away from scuppering potentially profitable trades and stop you from taking fractions of what your strategy and set-ups dictate.

By knowing that on average strategy A takes 2.5 hours (say) to reach its 6 point target then after watching the trade for 1.5 hours you should not feel frustrated that it hasn't reached the target yet. Well maybe you will feel frustrated because all of us want instant gratification. However, the knowledge in how long it usually takes to get to this target should help you from trying to kill the trade prematurely.

Of course this is just one dimension of trade management but it is one that I am sure a lot of traders skip and that's why I'm mentioning it. How many of you know the average time one of your setups takes to play out to the target?

Here is a setup example. A simple one: The Gap Fade.
Let's say that our strategy is to fade gaps at the open that are greater than 5 points and use a half gap size stop.

Now the astute trader will immediately notice that I've added in another dimension here. We do not know the target size because the target will be different each time. However, we can still run what is called a Monte Carlo simulation on historic data and see how long gaps of various sizes have taken to fill and using those averages (and assuming a linear relationship) we can set-up a table of expected times.

That last paragraph might sound a bit complicated but it isn't. Don't let words like Monte Carlo simulation put you off. This is a very simple concept. It just means that instead of using mathematical formulae to solve the problem we are going to through lots of data at it and see what the averages show us. Back testing is basically a type of Monte Carlo simulation.

Having a table of average times for each gap size to fill will help if you're trading this strategy. It will obviously take longer for a larger gap to fill than a small one. If you successfully traded a 5 point gap on Monday and then had an 8 point gap on Tuesday you should obviously adjust your Time Horizon expectation to target for Tuesday's trade in line with what your back testing has shown you.
- very good comments - i never know the average time my strategies take to complete so i admit guilty to that. - i would disagree with using the first method of calculating the average speed that the market moves at but i would only look at the time taken for each strategy because when a signal happens then you are entering a trade on anticipation that a certain pattern - ie a move in your direction - is going to happen - and so the speed will be influenced by this signal and will not be the same speed that the market moves at.
Very true George.

Another thing that I didn't mention is that the background to the market needs to be taken into account as that will also influence the speed and the Time Horizon expectation.

I didn't mention that earlier because I thought I had gone on long enough and didn't want to complicate things but a good example is Fed Day. The market has a very narrow range in the morning and even if you did get a signal then the chance of you hitting your target on a Fed Day morning (assuming that your target is not small or trivial) is much less and so that would have to be taken into account when doing the estimates.
- i have read your "gap study" analysis - do you have a table of expected times to "gap fill" for each size of gap in the es?
No I don't George. When I did that Gap Study (I am assuming that you're talking about the completed one in question format? This one: Gap Study) I used daily data and so did not have the granularity required to do those calculations. I have subsequently acquired that data and have the back testing system in a structure that I can do that but I have not had the time to set it up and run those figures. However, it is something that merits the time to do and so I hope to get it done at some point in time and will obviously publish the results.