As traders, we all have personalities that drive what we like and what we should avoid!  This article is meant to illustrate how to find indicators that are useful for finding market conditions with certain trade characteristics.   It is assumed that the reader has some familiarity with indicators and typical conditions of interest.  This will NOT be a comprehensive article covering all indicators and their uses.

I'm not a details type of person, so I partner with technology to help me take care all the fussy details that go with trading.    Analyzing my own trades is sometimes painful and frustrating.     I've attempted to classify the reasons a trade doesn't work out.    It's really frustrating when I've got the direction correct, but didn't take enough risk and got stopped out.    Sometimes I ask too much and a trade that would have helped the bottom line ends up as a scratch.   Sometimes I get the direction wrong only to observe some obvious signs after the fact.   To that end, I've keep track of four primary metrics and one additional metric that is useful when attempting  indicator combinations.

​I find five metrics to be of use, in no particular order.

For this evaluation of the mentioned five metrics, we'll contain the problem/solution space by putting constrains on the data.

​Here is the list of indicators used and the test condition for this article:
The same settings were used for both markets under evaluation.

​On each chart, the 'baseline' will be listed so we can compare against each indicator.  The baseline  calculation is performed using all market data and no conditions.

Lots of traders would love to know what direction to trade in.  What can these indicators tell us about direction?

These charts show us that  Bollinger Band, Ultimate Oscillator and William %R are among the best of this set at detecting market direction while indicators like ADX(average direction index) and Hilbert transform have very little effect on being able to predict direction.   However, direction by itself is nowhere near the complete story.   We must consider some of the other factors to understand what an indicator may tell us.

Risk, Risk Risk!    The previous charts showed us that some indicators are much better than others at detecting market direction.   What about risk?   Do any of the indicators affect the risk taken when entering a trade?    When we consider the effect an indicator has, we will consider the mean, the standard deviation, and the mode.  The mode will not be shown, because most indicator don't effect the mode, but we'll discuss the one indicator that does.

These charts tell us that most indicators increase the amount of risk.   The Average Directional Index is one of the view that both decrease the risk AND the standard deviation.   The Hilbert transform to detect trending or not also decreases both the risk and standard deviation, although not by a large amount.    Something very interesting is shown here.     Check out the Bollinger Bands both 7 and 13.   It's shown that they decrease the risk, but they also are at the top of the chart increasing the standard deviation.  The mode (most commonly occurring) (not shown) also increases for the Bollinger Bands.    So what's going on here and what's another way to look at this?    If the mean decreased but the standard deviation increases, this indicates that we should take a look at the distribution for the risk data.  To save space, I'll only look at the CL market, but the same effect is present in NQ also.

Note the difference in shape of the histograms.   The top chart is the baseline, capped at .50 cents risk.   All risks are assumed to be at lease .02 cents (thus the spike starting at .02).    The larger standard deviation of the Bollinger Bands gives a fatter/flatter distribution,  which is enough to move the mode several pennies (ticks for CL) larger.    Notice the ADX histogram can reduce risk but the shape of the histogram is very similar.   Risk by itself is somewhat boring without looking at the reward parts too.

In the same way that we can look at risk, we can also examine reward.   This means we'll be seeing both the mean and standard deviation.  Two of the indicators moved the mode and I'll note those at the end of these charts.

These charts indicate that many of the indicators decrease the average reward but increase the standard deviation.   It's worth noting that the average directional index both decreases the risk and increases the reward while decreasing the standard deviation of the reward.   The mode was increased by both the Chande Momentum Osc. 13  [(indData.cmo < -50.0)] and relative strength index  [(indData.rsi < 25.0)].
We can look at risk and reward individually, but we can also look at them as a ratio.

The previous charts showed the effects that these indicators have on risk and reward, but what about density?    Do these indicators have an effect on how likely it is that *something* will happen after you enter the market?    Shown in the charts is a 'No Odds' situation.    This describes a condition where the trader enters the market and both directions carry more risk than reward.    In other words, probably in the middle of volatile congestion.  This condition happens more on a shorter time frame since the market generally moves one way or the other given enough time.

Most indicators increase the possibility of no odds (bad) and decrease the odds density of have a ratio of 1.5 to 1 or higher for the tested conditions by varying amounts.   Again, this isn't the complete story.   We need to consider how often these
indicators happen, both on a daily basis and number of raw minutes. 

The following charts show how often the conditions indicate happen.   Two charts are used, both in percentage.   Once shows the percentage of minutes while the other shows the number of unique days.     If the number of days is low, we might consider either changing the indicator parameters OR investigate to see if the indicator is telling us about a certain market condition or type.

These charts show that Ultimate Oscillator as well as RSI occurs the least, both in terms of minutes and day.   Most of the other indicators happen on greater than 95 percent of the days

Trading is a multi-dimensional problem.   You can combine indicators to create
trades with a certain entry condition.    You can also study the sequences.   Let's say you enter a trade based on have a low ADX to minimize your risk as shown by the graphs.    The trade then moves into one of the Williams Percent R conditions.   You now have some edge on the likely direction (for the next 1/2 hour), but knowing that your risk has gone up, you can keep your stop a little looser.   Bollinger Bands are great at detecting direction, but that knowledge comes at a cost in terms of risk and reward.    You can use use the chart data to find a lower risk entry by using either a low ADX or not being too far to either side on a William Percent R.    The combinations are endless!   It's nice to see that these indicators do roughly the same thing on both markets, but one can use parameter tuning to be the most effective for a given market.     Thoughts, comments, questions??? 

  Trading is easy!   I'll admit the act of trading is easy, but persevering through the ups and downs of the market isn't always so easy.  In this article, I'd like to discuss some of the factors that characterize trading.   I'll be specifically focusing on risk.   No one likes to lose money, but losing in inevitable in trading, therefore the amount of *risk* we take is quite important.

  If we could predict the market behavior, we could reduce our risk!    Is it possible to predict the market?   Maybe.   If you had information about all the market participants, including the size of their accounts, and all their current positions, and knowledge of their decision making processes, and how consistently they do execute those process (which could be random), and a computer to take all that information into account in real time, with no software bugs.   In other words, possible, but not probable.    Given that we'll be working with woefully incomplete knowledge, what can we do?   First, we can look at history.    History will contain a sample of market events.   It's important to note that this is only a sample, not the complete set of actions that any market is capable of.   Just because a market hasn't done a certain thing, doesn't mean it can't.   I'm sure you've seen the statement "past performance is no guarantee of future performance".

  Given the information that we won't have enough facts to predict the market, it's virtually guaranteed that some executed trades will be losers in the financial sense. However, the amount you lose is up to you and is affected by two factors.   Those factors are size and risk.   If you buy 10 shares of a 50$ stock, you have a total risk of 500$ assuming you would watch the company go bankrupt and the stock goes to zero, like Enron did in 2001.   If you decide to stop losing money when or if the stock reaches 40$, then you have reduced your risk to 100$ (without considering slippage and transaction fees).   In this case, size is represented by how many shares you purchased (10) while risk is represented by how much loss per share you will accept before exiting the trade.  Risk is defined on a per share basis, where as total risk is size multiplied by per share risk.

  Let's work through an example.  The figure below is a 1 minute chart of Crude futures on 2016-06-13 with times shown in Central Standard time.

 Let's further assume you decide to 'go long', that is profit from rising prices, when you start to see successive higher lows shown at points 1 and 2.    It actually takes a few minutes past the point to detect it is a low.   Point 2 occurs at 8:35 am and let's say you enter the market by 8:42 am  at a price of 48.47.    How much risk should you take
before you decide the market is not going where you'd like (remember long is up for profit)?   Crude oil futures are traded inf 1 penny increments.  Each penny represents a 10$ change to your account if you are in the trade.   That means, if  you are long, a movement of 10 cents up will change your by 100$ or a movement of 10 cents down, will subtract 100$ from your account.   For this example, lets assume you are comfortable with a 200$ loss, therefore, you can set your stop (trade exit) at 48.27 (ignoring transaction fees and possible slippage).  What is the minimum stop you could have had, and still have the trade be profitable for you?   In this case, setting the stop any closer than 48.40 (the bar at 8:46 am has a low of 48.41) would result in the getting 'knocked out', that is your stop was activated and you exited the trade for a small loss before the market went your desired direction.   Since you were comfortable with a 20 cent risk, how about the other side, reward?   To give an answer, we need to introduce one more concept, expectancy.

Expectancy is the average win divided by the average loss, multiplied by the win rate in percentage, that is, how often do you make money vs lose money.

expectancy = (Win % x Average Win Size) –            (Loss % x Average Loss Size)
       
This concept expresses is how much money you will be expected to make over a set of trades.    Win rates vary per system.   Some of the systems I've seen win about 33% of the time.    If you only win 1/3 of the time, how do you make money?   That means
the winners have to be much *bigger* than the losers.  If your winners are 3 times as large as your losers, you'll be breaking even, which means you probably need a 4 to 1 ratio or higher to consistently make money.   Applying this to the current trade, setting a stop at 20 cents, you will a reward 80 cents, or 49.27.   As you can see on the chart, the market barely achieves this, and you would be so lucky to have gotten out at the top price of this chart!  What if you set the stop (not shown) at 15  cents instead?  Then a 4 to 1 ratio would be achieved with 60 cents or 49.07.   Which do you think is more likely to happen, a 60 cent move or an 80 cent move?

Setting a stop too close will make your win to loss ratio worse, but the winners can
be smaller (and therefore more probable) and still achieve a high rate of return.   A smaller winner will probably take less time, further reducing risk (less time in market) and possibly increase opportunity since you are out of the market quicker and ready to trade again.    It's all about risk!!    Questions, comments, thoughts?