Triple threat exit strategy

[pipwoof]

reviewing, i found an error in the july results reporting. it was just one day for the gpy/jpy, but enough to turn it from a winning month to a losing for this pair. if anyone tries to use this as their basis for comparison, it might have confused them and we want to be as accurate as possible. new july:

gbp/jpy
7-2, sell 124.99, +15, +15, +15
7-5, sell 124.22, -30, -30, -30
7-6, sell 123.82, +55, +55, +55
7-9, buy 123.53, -30, -30, -30
7-10, sell 122.99, be, be, be
7-11, buy 123.56, -5, -5, -5
7-12, buy 123.98, -25, -25, -25
7-16, sell 123.15, +40, +40, +15
7-17, buy 123.36, +35, +10, +10
7-19, sell 123.07, -20, -20, -20
7-20, sell 122.89, +10, +10, +10
7-23, sell 122.25, +25, +75, +75
7-24, sell 121.11, -40, -40, -40
7-25, sell 120.94, -30, -30, -30
7-26, sell, 120.92, -35, -35, -35
7-30, sell 123.25, +25, +75, +25
7-31, sell 122.20, -30, -30, -30
closed trades thru 7-31: -45

aud/usd
7/2, sell 1.0223, -5, -5, -5
7-5, sell 1.0258, -15, -15, -15
7-6, sell 1.0240, +40, +40, +40
7-9, sell 1.0181, +5, +5, +5
7-10, buy 1.0213, -15, -15, -15
7-11, buy 1.0244, +30, +5, +5
7-12, sell 1.0177, +25, +50, +25
7-16, sell 1.0218, -10, -10, -10
7-17, buy 1.0259, +40, +15, +15
7-18, buy 1.0319, +30, +80, +80
7-19, buy 1.0373, +25, +25, +25
7-23, sell 1.0323, +45, +45, +20
7-24, sell 1.0241, +40, +15, +15
7-25, sell 1.0208, be, be, be
7-26, buy 1.0336, +35, +85, +85
7-30, buy 1.0493, +5, +5, +5
7-31, sell 1.0474, -25, -25, -25
closed trades thru 7-31: +800

eur/usd
7/2, sell 1.2638, -15, -15, -15
7-5, sell 1.2506, +30, +105, +230
7-9, buy 1.2290, -15, -15, -15
7-10, buy 1.2324, -25, -25, -25
7-11, sell 1.2233, +30, +30, +5
7-12, sell 1.2211, +35, +10, +10
7-13, sell 1.2165, -10, -10, -10
7-16, sell 1.2233, +30, +30, +30
7-17, buy 1.2289, -40, -40, -40
7-19, buy 1.2306, -30, -30, -30
7-20, sell 1.2227, +25, +25, +25
7-23, sell 1.2104, be, be, be
7-24, sell 1.2065, -10, -10, -10
7-25, buy 1.2138, -15, -15, -15
7-26, buy 1.2170, +30, +105, +80
7-30, sell 1.2285, +35, +35, +35
7-31, buy 1.2330, -30, -30, -30
closed trades thru 7-31: +400

gbp/usd
7-2, sell 1.5660, -15, -15, -15
7-5, sell 1.5571, +45, +45, +95
7-6, sell 1.5497, be, be, be
7-9, buy 1.5497, -5, -5, -5
7-10, buy 1.5534, -35, -35, -35
7-11, buy 1.5548, +25, be, be
7-12, sell 1.5483, +30, +10, +10
7-13, buy 1.5516, +30, +30, +30
7-16, sell 1.5565, +40, +15, +15
7-17, buy 1.5654, -5, -5, -5
7-19, buy 1.5668, -20, -20, -20
7-23, sell 1.5588, +35, +60, +35
7-25, sell 1.5484, -15, -15, -15
7-26, buy 1.5551, +25, +125, +150
7-30, sell 1.5712, -10, -10, -10
7-31, sell 1.5623, -30, -30, -30
closed trades thru 7-31: +445

portfolio total for july: +1,600

[rpotor]

I ran the Londonbreak EA on a backtest simulation for July 2012 for the same pairs pipwoof reported in post #331. Results are obviously different, but look promising (detailed results can be found by clicking on the links below). Backtest simulations give results in actual money, so the pip values below are estimated using the lot size (the EA by default trades 0.01 lots):

GBP/JPY
pipwoof says -45 pips
Londonbreak EA says -798 pips

AUD/USD
pipwoof says +800 pips
Londonbreak EA says +787 pips

EUR/USD
pipwoof says +400 pips
Londonbreak EA says +1051 pips

GBP/USD
pipwoof says +445 pips
Londonbreak EA says +58 pips

Total
pipwoof says +1600 pips
Londonbreak EA says +1098 pips

[pipwoof]

let us prepare ourselves for the inevitable. we will have losses. with the exception of the claims made by some spread trading and martingale methods, any trading will have losses. my understanding of the math for spread and martingale is that those may work for a period of time, but have ultimate doom as part of the program. could be wrong, open to learn, just sayin' that vegas isn't going to let some guy with a roulette system go home with any money. consumate mathematicians, those folks.

db/tt had losses during the eighteen month +july eye/hand backtest. some were serious and scary. here are some points i have come across in studying our losses:

1. as much as these markets seem to move together, past data indicates they don't as much as we might think. periods of losing do not correlate well. while g/j was busy giving up pips in early january '11, a/u was putting them back in and waited toward the end of the month to take losses. same thing with e/u losing early in june '11 while g/u made up for it until the 27th thru 30th.

2. data from the original four-pair portfolio show one losing month overall, where several pairs indeed decided to give 'em up together. that was february of this year when we saw g/j +915, a/u -620, e/u -210, and g/u -530, net -445. we do note that, had e/j been included in our portfolio, february results would have been positive.

3. looking at losses in a row can give us important information about how much pain we might have to endure and help us decide if we are up for it. in the test period +july:
g/j had 4 losses in a row 5 times and 5 losses in a row 4 times.
a/u had 4 losses in a row 2 times and 5 losses in a row 2 times.
e/u had 4 losses in a row 4 times, 5 losses in a row 3 times, and 6 losses in a row 1 time.
g/u had 4 losses in a row 2 times and 5 losses in a row 4 times.
e/j had 4 losses in a row 2 times and 5 losses in a row 4 times.

strings of losses can certainly try our patience and it is definitely disturbing when the pairs get together on those strings. but, if we have some confidence in our projection to win based on what happened in the past, then we must also have confidence that our losing will resemble that same history. if everything changes, we are sol anyway.

[rpotor]

gbp/jpy
7-11, buy 123.56, -5, -5, -5
7-19, sell 123.07, -20, -20, -20

aud/usd
7/2, sell 1.0223, -5, -5, -5
7-5, sell 1.0258, -15, -15, -15
7-10, buy 1.0213, -15, -15, -15
7-16, sell 1.0218, -10, -10, -10

eur/usd
7/2, sell 1.2638, -15, -15, -15
7-9, buy 1.2290, -15, -15, -15
7-13, sell 1.2165, -10, -10, -10
7-24, sell 1.2065, -10, -10, -10
7-25, buy 1.2138, -15, -15, -15

gbp/usd
7-2, sell 1.5660, -15, -15, -15
7-9, buy 1.5497, -5, -5, -5
7-17, buy 1.5654, -5, -5, -5
7-19, buy 1.5668, -20, -20, -20
7-25, sell 1.5484, -15, -15, -15
7-30, sell 1.5712, -10, -10, -10

pipwoof I took a look at the above results. Please take a look at the above quoted lines. My understanding is that the strategy as written calls for the initial stop loss to be placed at least -25 pips but max. -50 pips away from the entry price. So, if we have a trade showing the same three numbers of pips lost it should mean - because of the "jumping" SLs - that all three positions were stopped out at the initial stop loss. In that case however losses lower than -25 pips are impossible.

Just a random example:

7-9, buy 1.5497, -5, -5, -5

Entry price is 1.5497, therefore our initial stop loss should be 1.5450 (giving us a -47 pips stop loss; SL of 1.5475 is "not enough" because it would give us "only" -22 pips of stop loss), so this trade should have a result of -47 / -47 / -47 by strict strategy rules. So this -5 / -5 / -5? How come? Was it maybe a judgement called trade?

[pipwoof]

rpotor, your observation is correct. any time you see the same loss for all three positions across a pair and it is less than -25, it would have been a judgment call. you will see more of these on mondays and fridays than other days, but they do occur on other days. i have confessed to being especially jumpy on those days that begin and end the week and am willing to lose the possibility of a satisfying move to run quickly for the exit.

i have assumed that most experienced traders could see the logic in at least most of my judgment calls and end up in agreement. but, let's take a look at a few of these and see what you think.

gbp/jpy, july 11, i jumped out with a -5 each position. this was not a monday or friday, but, in previous discussions of discretion, i have also mentioned the situation where there is a strong runup into positive territory followed by a fairly immediate failure. in this case, we entered long at 123.56 on the first of what turned out to be two strong up bars reaching a high of 123.97. so, here we are +40 in profits, but 3 pips short of being able to claim profit with one and protect the other two. and we find ourselves in this situation merely because of the quarter structure of the system. there are just times when judgment has to override that. we simply don't want +40 to turn into -30 by letting it slide all the way back to 123.25, which would be the logical conclusion of a strict application of the stop loss rules. after the high, we see an immediate and strong down bar reaching below 123.75 to a low of 123.66. at that low, we have now watched our lovely +40 turn into a +10. it manages to rise back above 123.75, but only for a moment and starts to fall again. at this, i've had enough of july 11 and pull the plug -25 early at 123.50, -5.

gpy/jpy, july 19, jumped at -20. look at our short entry in the midst of 4 or 5 strong down bars followed immediately by strong up bar, doji, strong up bar. run! out at the next available quarter, 123.25, licking a -20 wound instead of the -45 that would have been required by strict application of the rules.

7-2, several pairs exited early. monday. be jumpy, be schizy, be trigger-happy. from my observations, mondays and fridays will either move steadily in a direction, making us pips, or trampoline around all over the place to drive every kind of trader bonkers. get out early and wait for another day.

most of the others will also be monday, friday, or runup followed by failure to reach first target days. regarding these judgment calls, i know they are a programming nightmare and i'm not even sure they can or should be quantified. it might work to move stops to be after a profit level is achieved, say, +20. in any case, i think it is fair to offer to both you and robopip that if you want to program for testing, just stick with the fixed rules of the method. i suppose it should still be profitable and not rely entirely on the judgment calls.

i said earlier that i believe most experienced traders would see the logic in my discretionary decisions, but not necessarily. that's why we call them judgment calls. my judgment may not be your judgment. we have a lot of knoweldgeable and experienced traders following this thread. i would like to call on some of you to comment on this matter. when you see those early exits against the rules, would you have done the same or similar? i will appreciate your thoughts.

i do believe an important separation occurs between the gambler, i.e., "i'm gonna give it room, i sure would like for it to come back," and the probability trader who cuts losses quickly. sometimes more quickly than one of his/her favorite methods calls for.

rpotor, you are doing great work and making a terrific contribution to the thread. not only in your efforts to program, but in your eye for detail and sensible challenges to the method. it is a pleasure to respond, so keep coming. thanks.

[rpotor]

Okay, I hear you pipwoof. :-) That explains the numbers I've flagged. :-) Unfortunately you're also right in that these judgement calls simply cannot be programmed into any EA, because they just require a human sitting in front of the screen. :-) A human trader might say "okay, I'm jumping out now, at 5 pips profit" for one trade and "I'm jumping out at 18 pips on this trade". You can't program this into the EA, because an EA is by its very nature kind of robotic :-) You can't say in code something like "look EA, this is the rule, but sometimes it isn't". :-)

Anyway, your explanation was exactly the one I thought, when I wrote my post, however it's good to see it stated in your words, so if anyone comes here and reads the topic, he/she will be aware that this is a limitation in the EA, which cannot be "solved".

[pipwoof]

I ran the Londonbreak EA on a backtest simulation for July 2012 for the same pairs pipwoof reported in post #331. Results are obviously different, but look promising (detailed results can be found by clicking on the links below). Backtest simulations give results in actual money, so the pip values below are estimated using the lot size (the EA by default trades 0.01 lots):

GBP/JPY
pipwoof says -45 pips
Londonbreak EA says -798 pips

AUD/USD
pipwoof says +800 pips
Londonbreak EA says +787 pips

EUR/USD
pipwoof says +400 pips
Londonbreak EA says +1051 pips

GBP/USD
pipwoof says +445 pips
Londonbreak EA says +58 pips

Total
pipwoof says +1600 pips
Londonbreak EA says +1098 pips

also appreciate your good work here. i now have a limited study on gbp/jpy only, jan '12 thru july '12, comparing original placement of entry stops at gmt 00 with placement at the london open, 7:00. results obtained by hand, so the usual caveats apply. to get these numbers for the london open alternative, i defined "day" as the entire previous day plus the hours 00 thru 6:00. if the low at 6:00 was lower than any previous bar of the past 30, it became the low of the day. since we had additional data for monday trades, i did not require the +/- 11, but entered at the +/- 1. i also gave the london the same discretionary calls i gave the original. when there was a marked disparity, like in march, i double or triple checked. if you have an interest in this decision, you should check also.

list below shows month, results of london open entries, and results of gmt 00 entries:
1, +880, +150
2, +590, +460
3, +330, +1410
4, +830, +1105
5, +515, +690
6, -190, +365
7, +5, -45
totals, +2960, +4135

a close look can tell us a lot about what's going on here. the london often helps us stay out of an unwanted trade. especially when prices for the day are closing close to the extreme high or low for the day, we sometimes see that frustrating business where the move continues for two or three bars into the next day, triggering our entry, then gives it up and turns the other way. waiting for the london open steers us out of those first six hour bars and keeps us out of those particular unpleasant trades. an alternative would be to postpone entries, especially additional entries, until the continuation has proved itself, at least to some degree.

that seems to be the major strength of the london open variation. the biggest weakness is missing a move that occurs between 00 and 6:00 and that seems to happen often enough to outweigh the benefits of the bad trades not taken. i might conclude that there is some advantage to entering the market at sleepy time and being poised to catch a move quickly when it starts.

looking at more detail on some of the march trades:

3-2, gmt 00 enters at first bar of open and manages to get in early enough to tp at 130.00, +40, +15, +15. london gets a later start and gets caught entering just below a quarter, resulting in -25, -25, -25.

3-5, gmt 00 sells at 129.04 at 4:00 and catches the move, +25, +50, be. london is late, misses the 50 pip move between 00 and 06, and ends up stopped out, -10, -10, -10.

3-15, gmt enters long on opening bar and loses -45 X 3 = -135. by london open, prices have moved back below gmt 00 entry point and london does not enter at all this day. advantage london.

without doubt, this is a limited study on one pair and is far from convincing, but i am not encouraged enough to continue. unless we get something more promising, i prefer to leave the gmt 00 as the standard, at least for reporting results. we can leave other entry times in the discretion column and will appreciate hearing results obtained.

[pipwoof]

following five pairs, we currently have three entries:
g/j sell 122.62, currently printing just below 122.50. it is monday, stops to 122.50, risking -15 X 3 = -45.
a/u, no entry
e/u, no entry
g/u, sell 1.5617, stop loss at 1.5650, risking -35 X 3 = -105.
e/j, sell 97.12, stop loss at 97.50, risking -40 X 3 = -120.

[PipWright]

Our data must be far apart. I had a buy @ 123.164 on g/j that stopped out at 122.750 already. Also, a buy on e/u @ 1.24108 that went b.e. on a judgment call.

Did you base solely on Sunday day for todays highs and lows?

My low for Sunday shows 122.473 g/j

[Bijoymj]

Today's entry rates based on 21.00 GMT day close

Pair/Buy Stop/Sell stop

1. GJ/123.05/121.06
2. EJ/97.41/95.01
3. AU/1.0572/1.0445
4. GU/1.5658/1.5502
5. EU/1.2393/1.2164

Orders triggered so far,

1. GJ Long (Hit SL, -30,-30,-30)
2. EJ Long (TP hit, +34, +9,+9)
3. AU Long (Open)
4. GU Long (SL hit -33,-33,-33)
5. EU LOng (TP hit +32,+7,+7)

Open Air position of EJ from 3rd August stopped out @ 97.50 giving +51 pips. Retrace is still open and locked +1 pips

Total will be updated in the evening