Sunday, August 25, 2013

Week in Review: 19th August to 25th August 2013

I finished this week in the black, so I'm quite happy. :)

I only opened two trades this week, one on the GBPUSD and the other on the GBPJPY. There were quite a few more signals during the week that I could've taken, but I chose to control my exposure and ignored them while my other trades were open. I also have another position on the NZDUSD that is still open from last week. 

TRADES

GBPUSD - Daily timeframe - Hermes low volatility system

This was a straightforward setup. Friday's candle from the previous week had extremely low volatility. On Monday, I went long on the break of Friday's high. Reward was 0.5 * risk.


GBPJPY - Weekly timeframe - Set two bar reversal system

The GBPJPY offered a suitable two bar reversal setup on the weekly timeframe. I used my system 'Set' and made some profit. Price moved beyond my profit target by only 0.8 pips before it began reversing, so I was quite fortunate. Reward was 0.25 * risk.



SYSTEM DEVELOPMENT

There's some good news on the system development front. I've been testing a new system based on the 200 SMA and 21 EMA, after discovering this nugget of wisdom from legendary billionaire trader Paul Tudor Jones (essentially obey the 200 day moving average. Go long if price is above the 200SMA, or go short if below).

Suppose the 200 SMA and 21 EMA are both bullish, so your long-term and short-term direction are in sync. Look for a bearish candle. The candle's range must be between 0.75 and 1 ATR(14). The candle must NOT touch the 21 EMA. If it does touch, it may suggest that the market may be consolidating or even turning. 

The next day, enter on the break of the bearish candle's high. Place stop loss at the 61.8% retracement level between the candle's high and low. According to my backtest, a reward-to-risk of 0.4 to 1 looks the most robust, although all R:R ratios appear profitable.  Below is an example setup. 


I've tested five majors pairs from 2001 to early 2013 (EURUSD, USDJPY, AUDUSD, USDCAD, USDCHF). I hope to test the GBPUSD tomorrow. Current sample size is 325 trades. Here's how the equity curve is looking so far, assuming $10,000 and 1% risk per trade.


There's some significant drawdown during 2007, but overall the system looks promising enough to warrant further investigation. 

Tuesday, August 20, 2013

The mental discipline of not checking your trades

I'm not sure if other traders experience this problem, but I find the more frequently I check my open trades, the more anxious I become. This is especially true for those trades that take days or weeks to close. It's like an itch from a mosquito bite - you want to scratch but you know you shouldn't. 

I'm not sure about the source of this anxiety. What I do know is if I keep away from my broker or any live charts, that anxiety disappears. But the moment I check my trades or the market, I breath that anxiety back to life. 

Perhaps that anxiety is created because I create an expectation whenever I check my trades. What other reason is there to check your trades except to see how you're going? If your expectation is neutral, why the need to update yourself? There is no need. 

At the moment I have three trades open, so my total risk is at 6%. I was also thinking of keeping a pending order for a fourth trade this week, but realised my exposure may be too high, especially since three of those four trades will involve the USD. Perhaps my exposure of 6% is having a psychological effect. 

Sunday, August 18, 2013

Week in Review: 12th August to 18th August 2013

Ouch. Last week's trading was quite painful. I opened six trades, of which three lost. Since I'm using inverse reward:risk ratios for most of these trades, I finished in the red.

I still have one trade open on the NZDUSD, which is currently at breakeven. Rather than go through each trade, I'll post my losses and provide some comments.

TRADES

USDJPY - 1D - Hermes low volatility system

This was brutal. I spotted a low volatility candle on the USDJPY, 15th August, and put in a pending long on the break of USDJPY's high, and a pending short on the USDJPY's low. 




Both orders were triggered, but the market chose to be choppy that day. My short was first triggered, and price moved south until it was 2 pips short of my profit target. Then it reversed, took out my short stop loss, triggered my long, and reversed 4 pips short of my long profit target! And of course it took out my long stop loss. 

GPBUSD - 1D  - Var Daily Pinbar System

This trade was based on my 'Var' daily pinbar system. I don't use this system much because of it's historically steep drawdown. It uses a reward:risk of 2.5:1, so it's a low-probabiliy, high-reward system. I use a position size of 0.5%, which is very conservative.  

Anyway, this trade resulted in a loss. I spotted a small pinbar on the GBPUSD after the 13th August. I went short on the break of the pinbar's low, but it never went anywhere and took me out.


Because of the small position size, this trade didn't bother me much.

Upcoming week

There's a glut of signals that I've spotted over the weekend. Setups that interest me include the GBPUSD on the daily chart, and USDCHF and GBPJPY on the weekly charts.

Tuesday, August 13, 2013

Week in Review: 5th August 2013 to 11th August 2013

Last week was pretty quiet. I only opened one trade, which closed in profit. 

Trades

USDCAD - 1W - Set two-bar Reversal system

I was actually quite fortunate with this trade. A two-bar reversal pattern appeared on the USDCAD weekly chart. I took profit using a 0.25:1 reward-risk before price re(?)-reversed 4 pips later.





Presenting new trading system - "Al-Kutbay"


I finally finished designing this system last week after formalising its rules. I've named it "Al-Kutbay" after the Nabataean god of commerce. 

It's based on the neutral doji pattern, where the body (open and close) of the candle occurs within the 38.2% - 61.8% levels of the candle's range. A graphical example is below.




This candle is unremarkable on the lower time frames. However, it does seem to possess some potency on the weekly timeframe. I tested it over 17 currency pairs from 2001 to early 2013, generating 431 trades. Entry was the break of the high or low of the doji, with my stop loss positioned at the 61.8% retracement level of the doji. I then tried to optimise my profit target. The results of different profit targets, or R:R ratios, are below:



A reward-to-risk ratio of 0.25 seemed to provide the greatest profit factor. The win rate per trade was 86.54%.



The following is an equity curve from 2001 to early 2013, assuming a $10,000 starting balance, and 2% risk per trade.


0.25:1 reward:risk


R^2 is 0.92, which is very good. 

One thing to bear in mind with small reward:risk ratios is increased sensitivity to losses. However, you exchange this in return for greater robustness in choppy markets.

All that's left is forward testing to see how it goes. 

Saturday, August 3, 2013

Week in Review: 29th July 2013 to 4th August 2013

I finished this week in profit. I opened three trades, two using my Hermes low volatility system, and one using my Set weekly two-bar reversal system.

Trades

GBPUSD - 1D - Hermes low volatility system

I spotted this setup on Monday morning before the market opened. The GDPUSD closed on Friday with low volatility, which made a good entry signal. I went short on the break of Friday's low and used a 0.5 reward-to-risk ratio. Looking at the chart below, I know I could've bagged alot more profit (4 * risk?), but those are the breaks with mechanical trading.



USDCAD  - 1D - Hermes low volatility system

It's a similar story with the USDCAD. There wasn't much movement on the USDCAD during Monday. I went long on the break of Monday's high and took profit using a 0.5 reward-to-risk ratio.



GBPJPY - 1W - Set two-bar Reversal system

A two bar reversal pattern appeared on the GBPJPY;s weekly chart. I used a 0.25 reward-to-risk ratio. Price reversed shortly after hitting my profit target.


System development - weekly dojis and a few lessons on R:R ratios

I'm 90% complete on developing my weekly doji system. There's only one more pair I want to backtest, and then I'll see what I can achieve with optimisation. Below are two equity curves of my un-optimised system. The backtest results are from 12 years of data across 16 currency pairs, using 2% risk per trade on a starting balance of $10,000.


The first equity curve is the result of using a 1:1 reward:risk ratio. My profit factor is 1.31. Max drawdown was 24%.

# of Trades

The second equity curve is from using  a 0.25:1 reward:risk ratio. Profit factor is 1.36. Max drawdown was 6%.


What can we learn from this? It got me thinking about using small reward-to-risk ratios, and why they may be better than large ratios. 

A small reward-to-risk ratio usually mean small, consistent wins, occasionally punctuated with a massive loss. That's the price we pay. But what else can we glean about the nature of small R:R ratios?

We can also say that they are less susceptible to choppy markets. This seems to bear out in the equity curves above, where the 0.25 R:R equity curve seems more smoother than 1:1 R:R. The dips and spikes in the 1:1 equity curve seem to indicate greater sensitivity to changing market conditions. 

From a risk-management perspective, small R:R ratios also provide an earlier "system failure" warning than large R:R ratios.

For example, suppose we have a breakeven system using a 0.25 reward : risk ratio. Since this system is breakeven, we expect 4 wins for every loss. If we risk 2% per trade and experience 5 consecutive losses, we know that something is wrong and we can halt trading. In this case, our drawdown would be 10% (2%  * 5 losses).

However, suppose we use another breakeven system, this time with a 4:1 reward : risk ratio. We expect one win for every 4 losses. If we experience 5 consecutive losses, we are still within our expectations. It may take 10-15 consecutive losses before we halt trading. By this stage, our drawdown is now between 20%-30%. 

We have two systems with the same profit factor and expectancy. However, the system with the smaller R:R ratio is better able to alert us to system failure, and protect our equity.

These are just a few ideas that's been bouncing around in my head.