Risk and position management as an improvement of your strategy and edge

[raa]

I would like to post my comment on absolutely brilliant article by Stephen published here by Marc: The ‘Secret’ of Moving Averages & Forex Trading Plan



As I read the article, I noticed that I came to the very same conclusions when investigating longer-term "predictive" trading, but surely enough did not commit to that vast sort of analysis and study. Hat off to you Stephen.

I want to add here, for the benefit of community: I concentrated lately more on risk management and "odds-improving" and found out that proper application of these techniques can turn average strategy into the money plunger, and wrong one destroy accounts even with a best strategy. In application to your moving average trading technique, here is my thinking:

1. initial entry with only a fraction of position

2. as it moves into profit - add to position, so-called "pyramiding", again and again as previous chunks are protected by break even

3. leave a bigger leeway to let trade work - we have to assume our entries are far from being perfectly timed, so noncence to move to BE after 20 pips of profit if you target 2000

4. have a rule to kill your initial entry if price doesn't cooperate but leave an option of re-entry

5. incorporate some momentum/volume confirmation into the initial entry and add-ons, in my experience a larger candle on high volume (even tick volume in Mt4 can help) so-called climax in Volume Spread Analysis, can serve as a confirmation of a trend change, enabling entry after the cross, it enables easy automation of the strategy too: looking for a cross then for a large candle, enter, stop at the opposite side of the candle, voila, add on subsequent large high-volume candles, cut some of position on stalls, add again on break outs, kill position without waiting for the opposite cross though.

6. incorporate a rule of your position sizing: pay attention of your account equity curve and as it starts to slide down from the peak (high-watermark in money managers language), implement cutting of your position, even if you are trading a percentage based position, in my experience it make sense to cut it much more aggressively than in a 1:1 linear ratio of your equity, you need to devise some coefficient, it will flatten whipsaw of your equity curve producing some flat plateaus instead, however as your wins start to prevail, increase your position at the same rate and when you cross the watermark and tip into new equity highs - don't be afraid to re-invest your new gains into the risk! This type of risk management allows you to retreat and sit tight when times are bad and go in aggressively and rip rewards when market goes your way!

In regards to trading daily candles, I would say pay attention to London and NY closes, if candle does not retrace much, it means very strong signal as big traders don't close their books and don't take profits this day, as they prepare for a big big run instead, positioning themselves for a long term trade. We better do the same, playing their game: we add as they add, we cut as they cut.


To illustrate my point of pro-active position/risk management, here is a pick of my hypothetical pro-actively-optimized automated strategy. The strategy is not important, the point here is how positioning is managed: lots are increased when equity is going into news highs and aggressively cut when losses started to accumulate.

ScreenShot024.jpg

Just my opinion, hope it will contribute to improve the edge!

[SteveUKTrader]

Hi RAA, first an ‘e’ handshake, and many thanks for the comments. I could have provided a lot more technical detail but that wasn’t the message I wanted to get across.

Your experience and journey offer a fascinating insight to the mind of a determined trader. I can reminisce and think of many of the things you discuss.

I too am fascinated by trading algorithms, and I ‘know’ for sure that there are predictive techniques that could make stochs cross a few bars earlier that they do with standard settings. We all know that banks etc are employing ‘quants’ on 500k per year to gain this advantage. All that said, doing it somewhat more involved as you know. For example, I have looked at machine learning (ex Oxford) as one such rules learning and adaptive method. I know it works, I founded a company based on it.

I see you also noticed the retracement ‘rule’. It is indeed important but I could never capitalise on it. However I love the pyramiding discussion, I first learned of this in my ‘Gann Year’, for sure if you get it right then the cash flows.

For me, FMP has opened a door. The analysis is excellent, the people great, the results impressive and I place the emphasis on learning these techniques – my issue is that of trading – traders like to trade and I can’t get enough! I take my mind off trading when I go back to my ‘bookmaker’ roots and make some ‘quick bucks’ on betfair.

But, trading is where I want to be, so far since going live with FMP in November last year, I am up by about 400 pips. Not enough for sure and I keep learning. There is so much to get right, it’s just a matter of application.

I really look forward to many other interesting posts!

Steve

[raa]

Thanks for your kind comments, Steve
I agree with your conclusions. Surely, it is possible to predict the direction with some edge, although I tend to think about predicting market direction as predicting opponent's hands playing the game of poker: you never know but you don't need to know, all you need to know is to manage your game in a way that when you are right you pay for all your wrongs. Your approach to crossover is precisely that.

However, this is not enough to be just right at the end. Unfortunately we cannot magically start trading and move forward a year for example to see our "sure" strategy reaping rewards. We have to live with it every day and suffer when it suffering and celebrate when it's winning. So in my approach to algorithms be automatic to manual, I don't put too much attention on pip gain or even don't try to get to max reward. For me it's a risk to reward combination that matter, especially max draw-down and overall portfolio volatility. So, basically, your strategy would provide with an excellent example how sometimes lengthy draw-downs can be minimized while waiting for that big win. The method I described works especially well for strategies that work in cycles: longer draw-down phase, and them massive winning streak. If strategy tends to generate very short cycles, there is no need for proactive management, however I mostly seen these strategies only on a Tester page. In a real life almost everything is cycling, appealing for proactive risk management.

Keep up your good work, I am sure others would love to see some of your graphics and details just like me.

[SteveUKTrader]

Hi RAA, I had another look at my cross over records, bearing mind your strategy. This time I used a basket of currencies, ignored any trades that retraced more that 50% of the signal bar, timed between 1992 and 2011. The method of cross was a triple moving average and a close above or below the longest EMA.

Daily bars, there were:

936 LONG positions, 384 wins, 24,627 pips profit which is accounted for in 22 trades
986 SHORT positions, 414 wins. 53,599 pips profit which is accounted for in 46 trades

No stops, just exit when the cross over is opposite.

Now consider this, if all the trades were entered in a very small way at the start with, say, a 50 pip stop loss, the picture changes:

Longs:
416 long losers are capped at 50 pips, i.e. 20,800 pips as opposed to 62,913 pips loss, let them run. So, a ‘gain’ of 42,113 pips.

Therefore, we have 520 trades ‘active’. These produced a profit of 87,539 pips less the stop loss losers, so 66,739 pips profit. 130 profitable trades accounted for the pip total so the profitable trades really worked hard. All the other trades, about 390 of them came out level. The 130 ‘real winners’ were quite nicely spread across the years.

Bottom Line: 66,739 pips in the black, 42,112 more pips profit using tight stop loss, let the winners run.

Shorts:
437 short losers capped at 50 pips stop loss. I.e. a loss of 21850 as opposed to 67,043, let them run. So a gain of 45,193 pips.

Therefore we have 549 trades ‘active’. These produced a profit of 120,641 pips less the stop loss losers, so 98,741 pips profit. 157 profitable trades accounted for the pip total so the profitable trades again really worked hard. All the other trades, about 392 of them came out level. The 157 ‘real winners’ were quite also nicely spread across the years.

Bottom Line: 98,741 pips in the black, 45,193 more pips profit using tight stop losses, let the winners run.

Overall: Using tight stop losses, all the trades were evenly spread over the years, about 16 a year, say, one a month. Averaging it all out, that’s about 574 pips per month. (very rough)

Now, the interesting part. Use pyramiding on the really successful trades.

Let’s say, of the 287 trades that really produced the profits, we added more positions at 100 pip intervals. ALL of the 287 winners were 400 pip moves at minimum. So doubled returns on these, trebled return on 500 pip moves, quadruple on 600 pip moves.

There were also about 35 winners over 1,000 pips.

That’s another big number of well over 200,000 pips at the lower levels and over 500,000 pips for the big ones. NOW THAT IS INTERESTING. At this point I stared to rethink.

No wonder GANN, George Saros et all didn't need many winners, just the odd big one.

Steve

[raa]

Haha, glad you like it! This is how big guys are plunging markets. Boy plunger Jess Livermore pyramided big time as well!
Could you please share exactly your strategy settings and pairs you traded? I can get that implemented in EA and we can back test it on with easy on very wide type of trade management tactics.

[SteveUKTrader]

The Currency pairs were:

USD/AUD, AUD/JPY, USD/CAD, USD/CHF, CAD/JPY, USD/EUR, EUR/AUD, EUR/JPY, EUR/CAD, USD/GBP, GBP/AUD, GBP/CHF, GBP/EUR, GBP/JPY, USD/JPY, USD/NZD and USD/SGD

The results I gave you are based on:

Three EMA’s of 3,6,18 periods.

Long Signals are: when the 3 crosses the 6 up and at the same time (i.e. same bar), price closes above the 18.

The images attached are typical signals.

For a signal to be valid the next signal must pass the high of the signal bar, entry is at 1 pip past the previous bars high. Any retracement of the entry bar must not exceed 50% of the signal bar.

Exit the trade when the two shorter EMA’s cross the other way.

Notes: The ‘optimal’ percentage retracement can vary from currency to currency. Also, the entry point can be ‘optimal’ at higher than 1 pip, say 20 pips, it all depends of the pair. I also looked at post signal bar behaviour, and as you might expect, if it’s going to go, it goes quickly. Interestingly, if the fourth bar after the signal shows a positive result so far, but not too much, these offer another ‘latecomer’ entry.

Shorts are the opposite.

I also used EMA settings of 3,5,15 or 3,13,39 or 4,9,18 or 3,7,21. The results are very similar, sometimes duplicated. The longer the slow EMA, the fewer signals. I also found that some pairs reacted well, others not. I also tested dual average crosses and price crossing an SMA but these were not as successful. Personally, I think that many EMA combinations will work, I never did try a 5,9,21 combination but it looks like a good set.

I have a spreadsheet of all the signals if that would be useful. Along with hunderds of others incluidng commodities.

If you have any questions, I can skype you.

Steve

[raa]

Great job, Steve! I assume these are daily charts? What is your exit signal, is it the opposite direction signal?

[SteveUKTrader]

Great job, Steve! I assume these are daily charts? What is your exit signal, is it the opposite direction signal?

Yes these are all daily charts and Yes teh exit signal is the two shorter EMA's crossing in the opposite direction. I also have 4H results for the same pairs - about 6,000 - 10,000 bars analysed in each case.

Steve

[raa]

Thanks Steve, let's see if we can get this automated to follow through some more back testing and equity curve study/massaging.

[raa]

Looks like EURUSD is getting prepared to qualify for a trade according to Steve's system!