The Indicator Store

How To Find a Winner Trade Setup

In this note we will analyze the statistical approach to Trading, and thus answer several of the most common questions.

If you ever asked yourself some of these questions (or asked a strategy vendor), I suggest you read this note and watch the video attached at the end.

What is the effectiveness in this or that system?
How much does the X system earn?
What is the best system ?
How much can I earn with an X size account?
Why sell a system if it produces profit?
and several usual questions when starting to learn trading...

First some definitions:

A Trade Setup is a pattern, for example, a crossing of lines, a change of candle color, etc.

Even a robot, what it does is looking for a pattern, even if it is not specified which is.

Once a pattern is defined, for example a fast average crossing a slow average (this example is included in NinjaTrader) we need to find a combination of parameters (in this case the periods of both means), which produce the best results in the last N days.

NinjaTrader includes a module called Strategy Analyzer very useful for this purpose. See the video at the end of this blog to see an example.

It is important to keep in mind that we are looking for a combination of parameters that have produced the best performance in the last N days, or weeks. I mean ... in the past.

It is not so difficult to obtain good results using the analyzer. It is even possible to do it through a visual analysis, when the setup is not so complicated.

What must be taken into account is that the statistic is valid for that time interval in the past, and the objective of a trader is to speculate that in the future there will be some correlation making that combination of parameters continue to function for a certain time..

Surely you will have seen in many trading sites the mention of the CFTC Rule 4.41 which makes just mention that the results in the past do not guarantee that they will be repeated in the future, precisely because of what we have just mentioned.

The problem is that many unscrupulous sellers promote systems that claim to make profits effortlessly. Thing that makes no sense, since if so ... why sell it instead of use it ?

That is why the CFTC obliges to mention said rule when mentioning any system that works on futures.

As we have no way of knowing how long this configuration will work, we must monitor the performance of our setup when we put it to work live to detect any deviation from our report.

With respect to the range of dates to optimize, the ideal is to have one or more full weeks. This way there will be at least one day of the week or more. This is because every day of the week is different from the rest.

A common mistake is to make this date range very large; I have seen that several traders ask if there are performance reports of months or years of any strategy.

The answer is that it is possible to obtain a report of the time interval that we want, and even optimize it in any way, to get medium to low earnings with high to low risk, with high to low win%, etc ... as we see in the video.

The main thing to keep in mind is that having a report is just a photo, it can serve for a few days until we have to optimize again.

Optimizing a very long date range is not advisable, since when analyzing moments with different volatility what we are going to achieve is to average results down.

What we must take into account is the sample size, since obtaining a statistic of 10 trades is not as significant as one of 100 trades.

The ideal is to optimize the shortest possible time, in multiples of weeks from what we saw earlier, but that produces more than 100 operations.

Having a good amount of operations we make sure we have a reliable statistic, whose results we can use to monitor the setup in real time.

For example, the max drawdown, the maximum number of consecutive losers and the win%.

If any of these values is exceeded, (we can put a limit of 10 to 20% of them) we should stop operating the rest of the day.

As it may be an atypical day, we should try again the next day.

If we have a day out of bounds again then it is time to stop using this configuration and wait for the weekend to repeat the optimization process.

From the above we can then deduce that the performance we have at the end of the month is not as good as the one shown in a report, since we have to take into account the days breaking the statistics, that is, they produce a drawdown greater than calculated, and also the days in which the system is not operated, because some parameter was exceeded and we must wait for a new week to optimize.

Knowing this gives us a more realistic idea of what can be achieved by trading, and not being disappointed by false expectations after trying a system that seemed to work well and at some point I stopped doing it.

This is a constant work of trial and error, and it must be clear that nothing works forever.