those 100 pip dojis do look like they could be trouble don’t they? haven’t cut 'em off yet to see what happens. let’s see, if it were an up bar, that would be going short at yesterday’s open and long at yesterday’s close. a down bar would reverse, short at the close and long at the open. may have some time this weekend and sounds like it might be amenable to a spreadsheet look. other thoughts are entering on a range breakout from today’s open or entering on swing highs or lows wherever they are.
hope to complete TRIPLE THREAT EXITS by next week also. if you can watch the screen…
eur/usd, short 1.2361, close 1.2280, +81
gbp/jpy, short 123.82, close 123.28, +54
aud/usd, short 1.0240, close 1.0218, +22
usd/jpy, short 79.53, close 79.61, -8
net, +149
As mentioned yesterday, I have a couple of EA’s that come very closer to what you are doing here. I will post them and a description for the interest of anyone who wants to look at applying them. I cannot however spend the time to test as I have other priorities but I will be most interested in anyones experience and results with them. I cannot guarantee their correct function either as I never actually used them. They are ex a FF thread of some years back.
i am going to revisit my hedging studies. something just isn’t right. it occurred to me that, on the days when we get two entries, that first one is almost certain to be a loser. so, even taking random numbers, we aren’t getting an accurate look at what happened because many of those will be winners. at the point where a second trade in the opposite direction would have been initiated, we would have been at our maximum loss, yesterday’s range. a hedge position would just halt the loss and certainly isn’t likely to make the trade any worse.
doesn’t look promising. our base on gbp/jpy is just over +24,000. when i cut the wicks and enter on the opens and closes, i get around +5,500. still looking at entering on the open +/- some spread.
sammy, if you mean the wins and losses, the spreadsheet formulas to determine pips won and lost for both long and short trades are in post 1. from there, determine the number of wins and losses and divide into the pips. something like this: =if(f11>0,1,0) where f11 would be our column for trade outcome. the little formula here says if our trade was a win, count one. next, =if(f11<0,1,0) to count the losers. add 'em up and divide.
if you meant calculating trend slope, go back four bars, not including the current bar, and compare closes. =if(e10<e13,“UP”," ") where e is our column for closes. this says if yesterday’s close was higher than the close four days ago, give me an UP, otherwise, leave it blank. then, you would want to start a statement like this: =if(r14=“UP”,… and finish with whatever you wanted that condition to look at.
the initial mistake i made, and i think bijoymj made, gave us incredible results because we included the current bar in the calculation. can’t do that. when we are entering on the open, we don’t know what that close will be. so, use the four days PRIOR TO current bar.
i will reiterate that my studies to find an effetive trend filter have been futile and, so far, no one has posted better results using one. hope i helped.
I am trying to find out a solution without any indicators, but with some mathematical definitions of spread sheet as previous data is our base. If found something, beneficial, I will let all of you know…
Well…that’s much worse than I imagined…at least we can eliminate trading at the candle body open/close. :34: Thanks, pipwoof.
BTW, how did you determine these 4 currency pairs to trade (EU, GJ, AU, UJ)? Had I known the future results of just May and June, I obviously would have traded only the EU. The EU did not hit an emergency stop, and captured the most pips. Also, with just one currency pair, I would have been able to place 4 times the number of lots on the solitary EU.
Would there be any value in evaluating, say the last 3 months at a time, and trading just the top performing pair, or two pairs? Better yet, if we want to continue trading all 4 pairs, then rank their past 3 months’ performance. We could then assign 40% of our lot total to the number 1 performer, 30% to number 2 performer, 20% to number 3 performer and 10% to number 4 performer.
it isn’t what i consider a day, but what your broker considers. for me, in central time, oanda’s day is 11 pm to 10 pm. for ibfx, 7 pm to 6 pm. the easiest way to see your broker’s day is to open a one-hour chart. right click, properties, common, show period separators. draws a white line at 00 platform time. that could be anything, your time. see also posts 21, 30, and 113 for more coverage on this matter.
not sure i could package any better spreadsheets. at this point, mine are out the door and around the block with questions asked and answered. they are confusing, even to me. lol.
Alright thanks. Sorry to make you repeat yourself!
I will try to come up with a semi-independent model in excel. (I actually did yesterday but wondered away from my computer and didn’t save…you know what happened next-I lost everything!!!)
I will be using hour data in my replica. Also I will be sorting out Sundays.
I think using hour data will help sort out situations where both trades are triggered. (of course I’ll have to figure out how to “program” or write a formula that can make a decision based on the first hour to go past high/low by 1 pip-this will be the hard part!!!)
If I’m successful at making an accurate excel model next I will do some statistical analysis (I’ll try to identify the type of distribution and then make inference on the data for optimizing limits and stops.)
All in all, that is a lot of work and a lot of “ifs”.
the shortest answer is, “i don’t know,” and if i were smart enough, i’d probably leave it right there, but…
i chose the four because each seemed to have a decent daily range and were not eaten up with dramatic spikes. in retrospect, i might have chosen gbp/usd over usd/jpy and i am doing some more work on that consideration. it may be a good fifth wheel to add.
regarding your other thought, there are any number of trade sizing algorhithms based on various performance indexes. i have a fundamental resistance to anything predictive and harbor a secret suspicion that the quest for this tool may be among the root causes of failure. i am not so bull-headed that i couldn’t be swayed on that with the right evidence. here is something i wrote at another time and place:
Let’s say the Yen and the Aussie are performing best with this method. What would be the point in trading another pair that doesn’t do as well, like the Euro? Only because markets change and past performance is no guarantee of future results. I cut my teeth on futures trading and was always looking into (buying) systems designed by Larry Williams, Bruce Bab****, Bob Buran, etc. Most of them were optimized on a diverse portfolio and it was recommended that you trade the whole portfolio rather than cherry-pick what looked best. You may have made your money last year in silver, but this year the system tanked on it and you managed to get out alive thanks to copper or soybeans.
though I am not proposing that the currencies work like futures, still we can’t know when the Yen might have a long trading range that chews us up and maybe the Euro would be going somewhere and save us.
with my new system, i am usng the gbp/usd and it trots along making fair gains. then, in may, 2011, its takes only -480 losses and racks +1,265 wins. what could have seen that coming? would there have been any predictor for the gbp/jpy on 10-28-08 to range 1,317 pips, of which daybreak captured +813?
one thing we absolutely must do is monitor the equity charts for each pair. this i do believe in. when a pair starts drawing down, determine when to pull it offline until it starts performing again. with daybreak, the equity chart from 1990 for the aud/usd looks like a huge “v”, going down like a jumper from 1990 thru 2003, then, up like we approve from 2003 to now.
opinions are like, you know, everybody has one. i have many, but opinions is all they are. i have observed that you can’t deposit an opinion or make a sandwich out of one.
sammy, np with repeating myself. i do that as a matter of course. my patient wife has laughed at the same jokes for many years. i will help with your spreadsheet if i can, but my knoweldge is limited. i did have an offer from nix to help. find his post and he may be one to give you a hand.
when it comes to canceling sunday trades, here is what i had to do. select the column with date. you see it has period separators. on your spreadsheet tools, choose find and replace. replace all the . with /. then, format the column as a date. next, in a different column, convert the date to a text, like this, =text(a10,“ddd”). this says to convert a date column to a three-word text, as in, Sun. Then, it is easy enough to take your pip results column and say =if(r10=“sun”,0,r10).
In my data set for GBP/JPY I have found that double trade day occur about 18%-19% of the time (2001-2012). This is fairly significant next I will be working on determining which trade triggers first and then using that direction.
Unless of course someone can tell me why to do something different.
my data for gbp/jpy, 2007 thru june 2012 had 273 double trades. 170 of those were mondays. i have rewritten post 1 to offer options to handle the double trades. in your research, note that the gbp/jpy data is different for several years prior to 2006. starting in 1990, using daybreak, it hit an equity high near the end of 99, then drew down around -12,000 pips between 99 and mid-06. started working again around june 2006.
