My short trade moved back to BE after coming within less than 10 pips of the TP 
I wasnāt around to monitor the trade unfortunately or I almost certainly would have exited! I would have noticed that 1.62 is a support line with multiple bounces. But good to note that targeting half of ATR(14) pretty much picked the bottom in this case. GBP seems to be ranging. Will see what next week brings.
Orpips,
My live results from 2009 are 7 wins and 25 losses for a win rate of 22%. I have a return of 13.14R for the year.
I suspect our small differences are just the result of variations in different brokerās data feeds. I remember a couple trades where the difference between a win and a BE (or a loss and a BE) was less than 5 pips. We had an example of this a few weeks ago when Lavaman had a winner and the rest of us had losers, and this week with Reiās BE while the rest of us lost.
But overall Iād say our results are close enough to say that theyāre matching up.
I love your idea about moving the stop to a small profit instead of BE! I tested about a dozen different ways to doing this system but honestly I donāt remember if I tried that. Itās been 7 months since I did the backtesting and I only kept the results of the rules Iāve been trading.
If your data holds up through more backtesting Iām definitely switching to that method.
Iām also very interested in hearing your backtesting results from 2004. My data started on Oct 11, 2004 so I donāt know what happened before that, but from Oct '04 to Jan 31, '05 my backtesting shows 1 winner and 19 losers! At 2% risk that would have caused a 28% drawdown, but after that the system started winning and had recovered all the losses by the end of March. Iām interested to see if all of 2004 was a bad time for the system or if it was just a random losing streak. 
thanks Jason. Sorted now.
Backtest for 2008:
-17 wins (22%, or less if you include the BEs in the pool)
-59 losses
-total return of 18R (was looking very grim for the year until late October; in November a 10R winner and a couple more good winners were hit; definitely hammers home the necessity, when trading a mechanical method like this with a low hitrate, of taking absolutely every trade and not skipping any)
My idea about moving the SL to profit rather than BE wonāt work, it seems. Would have cost us 3 or 4 winners in 2008.
Note: Some of the winners, losers, and BEs were very very close, within a pip or two, so differences in spread would have changed the results. One reason I hate backtesting ā no way of knowing what the spread was at the time. I just figure 5 pips at all times to be conservative, but there are times that it saved trades and times that it cost trades.
So total return for 2008 to present would be 35R.
More slogging to comeā¦
Iām showing 21 winners and 48 losers for 2008. Thatās a bigger difference than we had in 2009, but because of the really close trades you mentioned itās still within the realm of broker price differences.
I also did not count the trades on Sept 8 and Oct 27. The Sunday candles were outrageously high on those weeks and I can honestly say I would have skipped those trades had I been trading the system live. That moves our 11 loss difference down to 7, which makes our results even closer together.
hmmmm, whatās the verdict?..
I read something on these forum before about measuring a systemās performacne using xR but I forgot what it meant. Is xR the return on risk? R=%risk? So if you are risking 2% of capital on each position then 35R over one year is a return of 70%?
Right, āRā is the return on the risk. 1R means you gained the same amount you risked, 2R means you gained twice your risk, etcā¦ So if you had risked 2% of your account per trade youād have a 70% return. Actually it would be 70% only if you didnāt compound your winnings into each trade. If you did (and you should) it would probably be closer to a 90-100% return.
For both of the years Iāve backtested so far, compounding didnāt increase the final total significantly over the R-multiple, which didnāt surprise me because the losses come in much longer strings than the wins. Just looking at 2008, the final balance would have been less than the R-multiple would suggest.
The differences in our results for 2008 make me think I must have been making mistakes somewhere. Of the two trades you mentioned not taking because of the large Sunday candles, one was a winner, so those would have been a wash as far as win/loss ratio goes.
Looking at the ones that were close, I can see that a few of the BEs could have been wins by a few pips, and one loss would have been a BE, but that should be the extent of the deviation; all others were too far away from win/BE for minor variations to make a difference. Iām thinking maybe the historical ATRs on the chart that Iām using are weird.
Here are this weeks tradesā¦
Long: 1.6530
Short: 1.6482
This weeks trades have a SL of 38 and a TP of 240.
Orpips, is your ATR 479? Just want to check and see if we have the same ATR because of our backtesting results.
Short trade just triggered for me ā¦ does it always trigger so quickly?
Well GBP seems to be at the top of the two week range so maybe its a winner this week.
Sometimes it does. Iāve even missed a few pips before because it triggered [I]while [/I]I was setting up my pending orders.
My ATR right now is showing 481. So close enough, especially since I think it has changed as the new candle has developed.
Two things about ATR as far as backtesting goes, though:
-
I think I was looking at the wrong weekās ATR for the backtesting though, i.e., on the weekly chart, I was grabbing the ATR for the week of, say, 1-06-08 to set the TP for the trades for the week starting 1-06-08, rather than using the ATR from the previous week. Not really sure which one would reflect the number I would use for TP.
-
In any case, I donāt think the ATR shown for the past weeks on the chart matches up with what the TP actually would have been no matter which week I use. For example, at the time of last Sundayās trade setup the ATR was 495 or thereabouts, which matches up with yours; but looking on the weekly chart now, as I would be for backtesting, it shows as 513 (if I used the previous week, that shows as 538).
After reading that it just clicked in my brain what the problem may be, and how I compensated for it in backtestingā¦
The historical ATRs shown on the weekly charts are the values as they are as the end of day on Friday, but weāre using the number as it is on Monday morning. The Monday morning ART will always be slightly less because of the small Sunday candle, so in backtesting I rounded the numbers down to the previous 25 levelā¦ Meaning I would have figured 513 as 500, and 539 as 525, etc.
I should have remembered I did that, especially since all the ATRs on my backtesting results are multiples of 25, but itās been so long ago and I overlooked it on my spreadsheetsā¦ Sorry!
Plus, itās not like half the weekly ATR(14) is some magical number anyway, any number around there should give you positive results over time, just not perfectly correlating results between different traders.
Hi Phil,
I am learning the Sunday Breakout strategy now.
My questions:
(1) Is the setup for ATR(14) same as ADR? The are 2 attachments given by s050399b, which one to download?
(2) My ATR 14 at GBPUSD weekly chat shows 0.0485 now, is correct?
(3) Ref to your backtesting file, what is Range + 10? is that what you ref to
"Draw a line 10 pips above the high and 10 pips below the low" in your post?
Thanks a lot
hmmm, sell stop loss triggered,ā¦
Thereās a v2 behind the filename which refers to version 2ā¦
(1) ADR is a little different from ATR, we need a weekly ATR set to 14 periods for this system.
(2) Yes, that is correct. The .0485 means 485 pips.
(3) Correct again. The 4H Sunday candle is the same as Sundayās range, so our stoploss is 10 pips above the range, or āRange + 10.ā
Same here. It flew up around London open and triggered the stop loss.
long triggered too, hmmmm,