Forex trading strategy #1 (Fast moving averages crossover)

Thank you for this web site, great work!
My trading platform allows to set price bars to different time intervals. Would you set price bars based on the time frame, so for 1 day use 30 mins bars?

For one day you use daily bars, for 30 min - 30 min bars.

Thomas

I feel compelled to do as many others have done here, and thank Edward, wife, and anyone else involved in this for your time and effort.

Less than three months ago, I had no exposure whatever to Forex, and had no thought in my mind about pursuing trading in the market. Then, late one evening I ran across some robot that was supposed to be the greatest thing ever. Really, I'm not a sucker normally, but in a weak moment I decided to give it a shot. I was supposed to be able to run it without knowing anything about Forex and let the money roll in.

Of course, that hasn't happened at all, and I've lost money.

Strangely enough, I'm OK with it, because I'm smart enough to know that claims like that probably won't pan out. In addition, the major benefit to that has been that it really sparked my interest in this market. So, bad investment in the robot aside, all of this has really been interesting. I see the huge pitfalls and risks, but also the potential.

In a little over two months, I've read everything i can, from an overview of the market, to the MetaTrader manuals, to a number of sites. This site has been the one constant in my research. It is really an incredible source of information. In addition, I decided with the failure of the robot I purchased to put things into my own hands, and started studying MQL4. I have 7 kids and a job, and I won't be able to sit in front of a computer, so I'm hoping to slowly and patiently learn, test, optimize, and start trading when things look good.

OK, sorry about the bio, but that brings me to Simple Strategy 1.

I put an EA together a week ago to trade this. I have since been doing nothing but testing and optimizing against only one currency pair against one time frame: USDJPY, 1 Hour. I have many other things I can still test, but I'm about ready to go live with the EA based on the parameters I have optimized. I wanted to share what I have found, as I think there is one interesting filtering parameter I've used that I haven't seen mentioned here.

I started with 10, 25, 50 EMAs as my base strategy, and then applied filtering against that. I passed through all the parameters at least twice, and ended up with 11, 24, 49 EMA. Pretty close to the base strategy.

I tested different stop and take profit levels. I have not tested (but will at some point) a true trailing stop from time zero, an adjusted stop by following bar high/low (where bar look-back would be a parameter), or adjusted stops after certain profit thresholds.

I have only tested with the following day limitation: no new trades open on Fridays, and all open trades close on Friday afternoon. I will look more at those things as well.

Right now, my exits are either a stop, take-profit, or the open of a new trade. There are more exit strategies I can test going forward, such as an initial cross of the mid EMA, or use of Fibonacci/Pivot Points. I'm only a couple months in here, so I need a better understanding of some of those things.

Also, I can test Stochastic and MACD in the future. Haven't done that either.

All that said, here is what it boils down to:

EMAs: 11, 24, 49
ADX(14)
USDJPY only
1 Hour time frame only
Simple Strategy #1 - I only trade if the short EMA first crosses the mid EMA, and then later crosses the long EMA.

Take Profit at 189 pips
Stop loss at 80 pips. Trail stop loss once profit is at or above 80 pips.

ADX(14) must be greater than 25.
For a buy, [(+DI) - (-DI)] must be greater than minus-2. In other words, +DI should be above -DI, or can be slightly below. Flip those around for Sell orders.

However, the absolutely most critical filter in turning the results from "eh" to "ooh" was the following: On the bar that closes on the other side of 49 EMA, if the difference between close and the 49 EMA is more than 13 pips, DO NOT TRADE. I observed numerous bad trades on large leaps over the long EMA line, so I added this filter. This filter alone more than doubled the profit.

As the EA currently stands, a back-test from 1999 against just USDJPY on 1 Hr chart produces more than a doubling. I started with $3000 in the test, trading 0.1 lots. Ending balance (almost 123 months) was $6277.48.

I went through the report to see the maximum drawdown that occurred from an account peak to a trough. The largest drop was from $4,374.35 to $3,911.10, or $463.25. If I take this as the deepest "bear market" in the period, and if one is willing to accept a bear period of 20% or so, it would indicate that 0.1 lots is about right for a $2300 account, and you can increase that proportionally with account balance as a money management target. Adjust accordingly if you are willing to accept a deeper bear market.

Sorry for the long post. Just wanted to share my results. Chalk it up to "newbie excitement." Now, we'll see what happens in a live environment. In the meantime, I am going to move on to other currency pairs and optimize. I'll share my findings.

I have a follow-up question, as I consider taking my EA live.

I have read, in various places, comments about broker sniping. I do know that some robots place stops and take-profit at a level well away from the actual trigger level where a transaction is intended. In my EA, I programmed my trailing stop as an order modification, and my initial order has the actual stop and take profit numbers that I am using.

Is the concern over sniping more perceived than real? Do any experienced traders have an opinion on the true risk of the broker knowing your stop and take profit?

I could re-program the EA to have an initial "fake" stop and take profit, say 25 pips further than intended, and simply have an internal calculation close the order when the "real" take profit or stop loss is hit. This is more cumbersome, so I am just wondering if this is a real concern.

If I do that, what is the minimum "safe" distance from stop and take-profit recommended? I still want to place an initial stop and TP in the event that kids accidentally shut down the computer or there's a power outage or whatever, since that would stop the EA and create additional risk without an ordered stop. Not necessarily likely, but prudent.

Thoughts?

I am currently testing this system with EUR/USD. Very interesting preliminary results. Will share more when I pass through all parameters a couple times.

Hi,

I started learning about trading just 2 months ago as i bought an expert advisor without reading about them being scams or bogus...I thought I would let it run on the demo account while i learn about trading and observe the robot's movement.
So far it has made a few small successful trades and 2 small losses. It doesnt trade much with a total of about 6 - 10 trades in 2 months.
I would like to test it on a live account just to see if it does keep up the current trend or it changes its behaviour.
On the other hand in the 2 months i've spent searching the net for something useful that could lead me to trading I couldn't find anything and the knowledge I learned is minimal and general. So I applause the creator of this website and appreciate him taking the time and effort to keep it up.

I do hope that everyone could bear with my questions that may sound silly at times but one learns from the silliest things. As I am brand new in the forex business here's my first question:

I have read about the Exponential Moving Averages but would like to know what do the numbers in the setting mean? As explained here to set them at e.g. 10,25,50 or 14,24,49, etc, etc.

What is there impact? how important is getting the right numbers other than getting an indication of when to enter or exit the market?

Regards,
Mo

Mo, I'm not a Forex expert, but I'm a math guy. The Exponential Moving Average is just a way of weighting a moving average. The weights produce a line on your chart that provides insight into what has happened in the market over the last number of periods you are calculating. Here, I have calculated the averages based on the CLOSE price of each bar (previous period of observation - in this case, 1 hour charts).

An average based on a smaller number of periods is more responsive to recent changes than an average based on longer periods. The purpose of the three different periods is to examine a movement in a short-term trend that then crosses the longer-term trend as a signal that price movement is afoot.

As for the math behind the averages, a simple moving average can be shown as follows: Suppose you want to calculate SMA, and you have ten values: 10, 8, 9, 7, 5, 6, 4, 5, 7, 8. SMA 10 is just an average of all ten values: 6.9. But there is only one data point, since it uses all the values. With SAM 5, you wouold have six data points, first averaging values 1 - 5, then 2 - 6, ..., 6 - 10: 7.8, 7.0, 6.2, 5.4, 5.4, 6.0.

Exponential Moving Average is different. It gives a certain percentage weight to the current CLOSE price, and then the rest of the weight is given to the prior EMA value. The usual formula for the weight given the most recent observation is p = 2/(n+1), where n = the number of observations you want to calculate your EMA for. So, for example, if EMA 15 has a value of 1.200, upon the close of the next bar with a CLOSE value of 1.300, the new EMA 15 will be [2/(1+15)]x1.300 + [14/16]x1.200 = 1.2125. This will, in all likelihood, be a higher value than the SMA 15, though it depends on what the value of the CLOSE value that dropped off for the new SMA calculation was.

As for my own numbers above, all I did was optimize my back-test. While the 24, 49 numbers didn't have a huge impact, the switch from 10 to 14 proved significant. I tooke the 10, 25, 50 as a general guide, and surmised that different pairs will behave differently and have different numbers that seem to apply.

Of course, if Edward has different ideas, I'd listen to him...

Hi Joe,

Welcome to Forex!
Your story was very exiting to read. Thanks a lot!

Despite being new to Forex, you seem to have a very good and quick grip on the strategic planning in the currency market. Again, I was truly exited to read about the new strategy that you've been modifying and testing recently. Particularly, an evaluation of how far away the price closes in the relation to a moving average has caught my attention (somehow I never thought of this factor before).
You've already shaped a completely new strategy. I personally like USD/JPY pair a lot. The fact that you are focusing on one currency pair at the time in order to learn its unique behaviour, daily trading ranges, trending ability and volatility factor, makes your research 10 times more as valuable and productive. Keep it up!

All-in-all, thank you for the interesting insight into your Forex journey. If there is anything I can help you with, let me know.

Regarding the question of whether a broker can see your stop loss order: I know that a broker cannot see/read your EA and the rules it trades by. However, when your EA sends orders, the data about stops, take profits etc become available. So, at that particular stage we can assume that if a broker runs a dealing desk, he can manipulate prices. A broker won't do it just for you, but if your stop is happened to be where all other stops are, then you're at higher risk. It would be an unacceptable practice for a broker by any standards, but it can happen.

Best regards,
Edward

Since yesterday's read through the website I have been able to slightly focus the knowledge I gained into action.
As a start I have decided to use the pivot points and the EMA. In doing so I have been finally able to decide when I want to enter and exit a trade rather than what I used to do....guess work.
For the EMA I have started as described in this example with EMA 10,25,50 with 15 min chart. I am testing using EUR/USD and random pair GBP/CHF.

Question: can the numbers 10,25,50 be used for any time frame if one wants? i.e. for example 1 min charts or 1 hour charts or do they need to change as the time frame changes? In any case why would they need or need not change?

Regards,
MO