Last year, I did a review of my superstock trades on this blog and that review shaped my trading strategy. I had a good year so far trading superstocks using the strategy that I develop during the review. Unfortunately, superstock is the only strategy out of the four strategies that is profitable this year. Forex was my bread and butter last year but it has gone downhill since the last quarter of 2016. I am determined to reverse the trend and decide to have a review of my 2017 trades so far to fine-tune my strategy.

Like the superstock review, this is going to be a multi-part post. I am going to bring you through the process of developing my strategy through these posts and as of now, I have no idea where the review will bring me. At the very worse, it may tell my that I have no edge at all and should stop trading Forex all together. If that is what the data tell me, I will gladly accept it.

Research Questions

Though this review, these are the questions that I hope to find out

1) Do I have an edge trading my current strategy?

2) What factors determines the probability if a trade if profitable?

3) Am I cutting losses / shifting my stop-loss fast enough?

4) Is my stop-loss set correctly or should I shift it closer or further?

5) Am I leaving too much money on the table with my exit strategy?

In part 1 of the review, we will be mostly focusing on question 1 and 2.

Overview

I log my trades using a program called Edgewonk. The good thing about it is the 'Trade Analytics' function that all you to generate various statistics on your performances. It is something like the 'pivot table' function in Excel, with a nicer interface.

I made a total of 42 trades from the start of 2017 to 8th June 2017. Of these, 5 trades are wrong entries, i.e. trade by mistake or wrong order. Therefore the review is based on the remaining 37 trades only. The sample size is a bit on the low side but I will have to make do with what I have. I will have to do some out-of-sample testing and refresh the review with a larger sample size in the future for a more robust analysis.

 My per trade risk is SGD 1000 (+-S$50), and the total loss from these 37 trades is S$6.2k, more than 6 times risk. My win rate is just above even at 56.8%. My average loss is $1013.54 while my average win is just $479.68. Intuitively there is a big issue with my risk-reward ratio.

Setup

I trade 3 setups for Forex - pinbar, double inside bar and bullish/bearish outside bar (BUOB/BEOB). Outside bar trades were the most common, contributing 19 out of 37 trades, followed by 14 pinbar trades and just 4 double inside bar trades. 

The double inside bar strategy worked the best with 75% win rate but I will not bank on it due to a small sample size of 4. The outside bar strategy has a decent win rate of 68% but the expectancy is too low at $18 per trade when my risk per trade is $1000. The pinbar strategy is what I need to work on with a low win rate and very negative expectancy. It is the sole contributor for my losses incurred in first quarter of the year though things can be improved for the other strategies.

Determinants of a Winning Trade

When I started logging my trades in my trading journals, I suspected four possible factors that I think may affect the outcome of a trade. They are:
1) Size of bar
2) Time till profit
3) Short-term trend (10MA)
4) Short-mid term trend (30MA)
4) Location

We shall look at each factor separately below.

Size of Bar

Intuitively, I would want a outside bar and a pinbar to be as big as possible. Since they are reversal bars, a large size will signify strength needed to reverse a trend. Whereas of inside bar, by definition they should be small.

I measure the size of the bar by its size relative to its average true range (ATR), in increment of 0.5 ATR.



Among the 14 pinbar trades, 6 trades have pinbar of size 1-1.5 ATR and 8 of size 0.5-1 ATR. The win rate is low regardless of size. The pinbar setup is probably not profitable to begin with and the size of bar does not affect the win rate or expectancy.

Among the 19 outside bar trades, there are 5 trades with size 0.5-1 ATR, 11 with size 1-1.5 ATR, 2 with size 1.5-2 ATR and 1 with more than 2 ATR. The effect of the size of the bar is not obvious from the statistics.

As expected, all 4 double inside bar trades have size of bar less than 1 ATR.

I tried to look at the size of bar through another measure - the size relative to the last 5 bars.

Once again, the size of bar using this measure does not show if bigger bars are more or less profitable. Especially for the outside bar, the 2nd and 4th largest bar performs better than the largest and the 3rd largest bar.

Time Till Profit

The time till profit measures two things - which is the first day that the trade is in profit at the close of the day, and whether the trade is profitable at the close of each day thereafter till the trade is exited. The table below shows the statistics of the trades in each scenario.


One important finding from the table is that, if a trade is not profitable on the first day, it only has 4 out of 18 chance of being profitable on the second day. This means there may be a chance to cut loss earlier.

The trades with subsequent loss may appear to have high win rate, but that is due to shifting of stop to near breakeven after the trade is profitable on the first or second day. That explains the low expectancy for these trades.

Trend (10/30MA)

"The Trend is your Friend" is the number one rule of many traders. I used 2 measures to determine the trend, the 10 day moving average and the 30 day moving average. Do not ask me how I chose this number as the choice is mainly arbitrary. The 10 MA does look very important and I will explain more about it in the next post.




Looking at my short term trend (10MA), I am not following "The Trend is your Friend" rule since 27 trades are counter trend trades and only 4 trade that is in the direction of the trend. The remaining 6 trades are whipsawing, i.e. the 10MA is flat or at a turning point.

I further broke down the trade by the location of the last price, whether it is on the side of the trend or is it on the other side. For example, if the 10MA is trending down and the last price is above the 10MA, I will classify the trade as "Counter Trend and Price Not on Trend Side".

Only the with trend trades have a positive expectancy, so the saying above is true after all.

If we breakdown the statistics further down by setup, we can clearly see that the pinbar and outside bar counter-trend trades are those that drive down my trading performance. In particular, counter trend pinbar trades does not appear to stand a chance.

With further testing, there may be opportunity to reverse my counter-trend trades or at least drop them from my trading all together.

I did a similar analysis, this time with the 30MA. The results were not too conclusive with the 30MA as both the with trend and counter trend trade performed badly and turns out that the whipsaw trades performed best.


Location

By design, I will only be trading at the swing highs or swing lows. By tracking the location of the trade, I am looking at the additional effect of the trade happening at a big round number and/or a price pivot zone (where prior support and resistance overlap).

Since almost half of the trades are at swing highs/lows only, the sample size for the other locations are too small for me to come to any conclusion, even without controlling for other factors.

From this first part of the review, I see that the setup and the trend, defined by the 10MA, are the two most important determinants of a winning trade. In the next part of the review, I will make use of these findings to come out with alternative setups and compare the performance of these alternative setups with the current setups that I was using.