Sunday Breakout Strategy

[pirateboy]

hello s050,

have been wondering a long time - what is the meaning of OSP?

also, just to confirm, the indicator doesn't work unless you have MT4 on before the sunday candle. for example, if i turn it on now (saturday), no lines are drawn.

also, how did you calculate ATR?

// double ATR = (iATR(Symbol(),tf,period,0))* MathPow(10,Digits);

// if (timeframe=="H4")tf=240;

so is the result same if i use the ATR indicator built into mt4, turn on a tf of 4h, and use ATR(240)? (the result is different, don't know why, I am missing something).



EDIT: I know the answer now: I shld use the 1wk tf, and put the MT4 ATR(14).

You need a flag to check if the BE point has already been changed, and to make sure it's only changed once, otherwise the EA is going to change it again everytime the price moves.

It's a bug that wouldn't hurt the EA since it's being changed to the same value over and over, but it's still a bug.

Other than that it looks great!

[phil838]

so is the result same if i use the ATR indicator built into mt4, turn on a tf of 4h, and use ATR(240)? (the result is different, don't know why, I am missing something).

It's different simply because you are using a 4H chart. No matter what amount of time you're using for the period an ATR on a 4H chart will be the average range during a 4 hour period. We need the average range of the week...

[phil838]

I did some backtesting on EUR/JPY and the results are amazing!!!!

I ran through a few different sets of rules since the Sunday candles are so big on that pair and came up with these new rules, which are slightly different from what works best on GBP/USD.

Entries - Same as GBP/USD - 10 pips above/below the Sunday candle.

Stoploss - The opposite entry point. This is NOT the same as GBP/USD! The normal rules are the other side of the Sunday candle, but this is 10 pips past the other side of the candle. This extra buffer is very important for EUR/JPY because of the larger price swings.

Takeprofit - Same as GBP/USD - 1/2 the weekly ATR(14)

Breakeven - 1/4 the weekly ATR(14). 2x the stoploss didn't work here. That was too high a number because of the large candles.

Here are the results. Keep in mind these are only for this year, and may not hold up over the long term. I haven't gone back any further than a year yet, and probably won't for a while... It takes way too long to backtest moving from the mouse to the keyboard every 10 seconds with one hand...

Compounded return rate - 87.6%!!!
R multiple- 34.18R
Win rate - 14 winners, 24 losers = 37% (BE's not counted)

I'm going to trade both GBP/USD and EUR/JPY tomorrow, risking 1% per trade on each pair.

EDIT: If anyone saw these numbers earlier I had a math error and corrected it. That's why the numbers dropped a bit..

[Azy]

I did some backtesting on EUR/JPY and the results are amazing!!!!

I ran through a few different sets of rules since the Sunday candles are so big on that pair and came up with these new rules, which are slightly different from what works best on GBP/USD.

Entries - Same as GBP/USD - 10 pips above/below the Sunday candle.

Stoploss - The opposite entry point. This is NOT the same as GBP/USD! The normal rules are the other side of the Sunday candle, but this is 10 pips past the other side of the candle. This extra buffer is very important for EUR/JPY because of the larger price swings.

Takeprofit - Same as GBP/USD - 1/2 the weekly ATR(14)

Breakeven - 1/4 the weekly ATR(14). 2x the stoploss didn't work here. That was too high a number because of the large candles.

Here are the results. Keep in mind these are only for this year, and may not hold up over the long term. I haven't gone back any further than a year yet, and probably won't for a while... It takes way too long to backtest moving from the mouse to the keyboard every 10 seconds with one hand...

Compounded return rate - 159.41%!!!
R multiple- 51.09R
Win rate - 18 winners, 24 losers = 43% (BE's not counted)

I'm going to trade both GBP/USD and EUR/JPY tomorrow, risking 1% per trade on each pair.

Hi Phil,

nice to c u back. I find ur system very interesting.... but want to share few of my concerns... I was laid off few months back so back tested some systems... they work amazingly for few years.. but then go down when u go back 6-7 years... so i wish v could check ur system... lets say up to 2000 or so... I trade with fxcm and my cnadles are a bit diff then urs... plus I have EST in canada... and fxcm starts trading at 17:15. but d advantage which FXCM offers is dat even with a micro account they offers u the charts which go back up to 1993!

Reg,
Az

[phil838]

Hi Phil,

nice to c u back. I find ur system very interesting.... but want to share few of my concerns... I was laid off few months back so back tested some systems... they work amazingly for few years.. but then go down when u go back 6-7 years... so i wish v could check ur system... lets say up to 2000 or so... I trade with fxcm and my cnadles are a bit diff then urs... plus I have EST in canada... and fxcm starts trading at 17:15. but d advantage which FXCM offers is dat even with a micro account they offers u the charts which go back up to 1993!

Reg,
Az

While it is true that I haven't tested it back any further than 2005, I feel that the 5 years worth of data that we have is enough.

I may be wrong, but I just can't see how a system like this one could work so well for such a long time and then suddenly stop. This system is 100% adaptable to any market condition. If the markets become more volatile, then the ATR and candle sizes increase, which in turn increases our TP and SL. If the markets start ranging the opposite happens and our TP and SL levels come down to match.

There will be an adjustment period of a month of two before the system adapts, of course, but if you're using proper money management that shouldn't be an issue.

In fact, anytime you're using a system with proper money management you shouldn't be concerned with your system failing. This was proven recently by the failure of the NickB scalp line system that a lot of us on this forum trade. I traded it profitably for almost 2 years before it started failing. Once I saw there was a problem with the system I first reduced my lot sizes, then when it didn't turn around I stopped trading it altogether. The money I lost when the system failed was a drop in the bucket compared to what I made during the previous 2 years.

When in comes to Sunday breakout trades I would need well over 100 losses in a row to wipe my account out, and I would need over 25 losses in a row to make me lose all the profits I've made with this system since I started trading it. If I got 10 in a row I would stop trading this live, so there's really no way I can lose. Even if I lose money on the next 10 trades and never see another winner from this system I'd still have made a lot of money from it.

It all comes down to money management. If you're overleveraging yourself then you need to be afraid of losing streaks, but if you risk a conservative amount per trade then you don't have to worry about it.

[pirateboy]

Hi Phil, this is awesome and great work

I've been working on my own backtest (that's why I've been asking about ATR in last post). But I never got down to doing this properly, other than the back of envelope on E/J for 09(which I mentioned in passing), and looked like a hitter even on old rules, but you've just comfirmed that!

Could you share with me/us, how you went about your back testing, and how you optimised it?

You must have some faster way than me, because just to get ATR, I need to open the 1 wk candle, check the ATR14, go back to 4hr TF (hmmm...maybe you opened two MT4 platforms and have 2 monitors?) and ALL WITH ONE HAND!!!

I looked E/J for May - Jul 08, and I think it wasn't that great, so I dropped this idea, but I'm relooking at this cross definitely.

Could you also share what the results would be if we used the current G/U rules, without optimisation?

Thanks Phil

[EternalFxStudent]

While it is true that I haven't tested it back any further than 2005, I feel that the 5 years worth of data that we have is enough.

I may be wrong, but I just can't see how a system like this one could work so well for such a long time and then suddenly stop. This system is 100% adaptable to any market condition. If the markets become more volatile, then the ATR and candle sizes increase, which in turn increases our TP and SL. If the markets start ranging the opposite happens and our TP and SL levels come down to match.

There will be an adjustment period of a month of two before the system adapts, of course, but if you're using proper money management that shouldn't be an issue.

In fact, anytime you're using a system with proper money management you shouldn't be concerned with your system failing. This was proven recently by the failure of the NickB scalp line system that a lot of us on this forum trade. I traded it profitably for almost 2 years before it started failing. Once I saw there was a problem with the system I first reduced my lot sizes, then when it didn't turn around I stopped trading it altogether. The money I lost when the system failed was a drop in the bucket compared to what I made during the previous 2 years.

When in comes to Sunday breakout trades I would need well over 100 losses in a row to wipe my account out, and I would need over 25 losses in a row to make me lose all the profits I've made with this system since I started trading it. If I got 10 in a row I would stop trading this live, so there's really no way I can lose. Even if I lose money on the next 10 trades and never see another winner from this system I'd still have made a lot of money from it.

It all comes down to money management. If you're overleveraging yourself then you need to be afraid of losing streaks, but if you risk a conservative amount per trade then you don't have to worry about it.

Hi Phil838, missed you out here.

So you are not going to trade the NickB Scalp line method any longer? Have you seen the backtesting that MetalHawk just completed on the different G/J TP and SL levels in his blog this week. The work he did was very informative. Metalhawk is planning to take 2 trades for each scalp line, each trade with 2 different TP / SL configurations...so he will basically be trading 2 different systems, I guess to try and smooth the equity curve a little bit. Both Nick and Metalhawk seem to be just as excited as ever even after the recent string of losses, and I know that Nick is circling back to his old ways and has started to trade the E/J again as well (but is not releasing any info just yet).

Anyway, glad that you are back and as usual, I totally agree with your insights shared in these most recent posts. As a matter of fact, I trade every system and method that I trade with the belief that all of them will eventually "fail", and recognizing that failure is really the biggest discretionary part of my trading outside of recognized what methods to start trading in the first place. As for me, I have not given up on the Scalplines just yet.

[pirateboy]

I agree and think it is useful to backtest maybe 5 yrs, but not really more than that. the markets were different 5 yrs back. i think as long as there is sufficient liquidity, lots of people trading technicals, and there is no perfect information (as adjusted by news), the tide tends to die down and follow a trend.

While it is true that I haven't tested it back any further than 2005, I feel that the 5 years worth of data that we have is enough.

[pirateboy]

Hi Phil, on the stop loss, I agree I hate these 'skin of tooth' shaves. but I didn't see this extra 10 pips benefiting the system in 2009. Do you mind to direct me to which Sundays this helped?

Thanks dude.

Stoploss - The opposite entry point. This is NOT the same as GBP/USD! The normal rules are the other side of the Sunday candle, but this is 10 pips past the other side of the candle. This extra buffer is very important for EUR/JPY because of the larger price swings.

[phil838]

Hi Phil, this is awesome and great work

I've been working on my own backtest (that's why I've been asking about ATR in last post). But I never got down to doing this properly, other than the back of envelope on E/J for 09(which I mentioned in passing), and looked like a hitter even on old rules, but you've just comfirmed that!

Could you share with me/us, how you went about your back testing, and how you optimised it?

You must have some faster way than me, because just to get ATR, I need to open the 1 wk candle, check the ATR14, go back to 4hr TF (hmmm...maybe you opened two MT4 platforms and have 2 monitors?) and ALL WITH ONE HAND!!!

I looked E/J for May - Jul 08, and I think it wasn't that great, so I dropped this idea, but I'm relooking at this cross definitely.

Could you also share what the results would be if we used the current G/U rules, without optimisation?

Thanks Phil

It's important to test a large sample size with a 1-2 trade per week system like this. Even if May - Jul 08 sucks it might just be a bad run. For example, if you started with $100 in the first week of May 09 and traded the next three months you'd have a balance of $96. That's a 4% loss over those 3 months, but the whole year is extremely profitable.

For this round of backtesting I put together a custom indicator to make things faster. I found a daily breakout indicator on Forex Factory, then modified the code and mashed it together with S05's Sunday breakout indicator. The result looks like this... Orange lines are entries, blue is BE, and Purple is TP. Using this speeds things up tremendiously.






For the trades where I need to move down to a lower TF I do indeed have two monitors with two different MT4's running.

I did get a little lazy this time and on a few of the iffy trades I just marked it down as a loss, even though it may have been a BE or winner. I like for the overall results to be a little conservative anyway, since real trading never goes as smoothly as backtesting.

Here's my backtesting indicator. Be careful with it though. I threw this together and it works great on my charts, but if you're using a different timezone or 5 decimals or something else different it might not work right. I didn't intend it for the general public.