daily range Strategy

Quoting JillyB

Disliked

Oops sorry forgot to mention it - thanks for pointing this out.

The data I posted in entry 397 relates to the straddle of the London open at 8.00am GMT.

Ignored

Quoting JillyB

Disliked

Oops sorry forgot to mention it - thanks for pointing this out.

The data I posted in entry 397 relates to the straddle of the London open at 8.00am GMT.

Ignored

Quoting 62a94

Disliked

HEY jillyB on your post 397 you state you backtested 90 wins and 26 losses
my question is are the 26 losses the result of the 25 sl being definitely hit and the 90 wins definitely being hit before being stopped out.

MR DETROIT

Ignored

Quoting llambert3

Disliked

permanetjuan..


Great job. I posted the other day what I found, but deleted it because it read and looked like craziness. I'll post again when I get a chance to organize it.

60 straddle - yes perfect. 35s/l is what you need. I looked at data since 3/8/06 on the 15 M timeframe (as far back as metatrader will let me go) I spent a TON of time on this. I am so confident now that I am using this live.

Now look at your data. Add days of the week to the data. Do you notice a tp on certain days have a higher success rate? Exclude Monday and Wednesday (never trade those days with 60 straddle - especially wed), that leaves Tue/Thur/Fri to trade, and hint they all 3 have different tp for the highest percent wins. My take is Thur is higher because of Thurs news and expected Fri news. Fri is less than Thur as buying/selling pressure is coming to an end for the week. And Tue is used to make a little bit of extra pips. I won't say my numbers yet. I want to see what you come up with to see if we about match, if you feel like doing that.

Ignored

Quoting 62a94

Disliked

PERMANENTJAUN is your backtesting performance the result of a definite stop loss being hit and a definite target being hit. referring to post 394

Ignored

Quoting permanentjaun

Disliked

Howdy llambert,

GFT doesn't show what day of the week a date is. So I had to manually look on my computer calender. Our success rates depend on our TP levels. I only gave examples in my previous post to show that we shouldn't be looking for multiple TP levels. There is an efficient point out there which I haven't found yet.

So I can't say exactly what the success rates where considering I haven't given a TP level yet. For sake of argument we'll assume I entered trades on up days with a 60 pip straddle and was only looking for a 5 pip profit. (This kept the losing days to a minimum so I could quickly see what days of the week they were.)

I may have counted wrong in my first post, but I only see 26 bad trades now. Meaning the up movement hit 60-68 which would not have allowed me to exit with a 5 pip profit. Here is the breakdown of what days of the week those losing days were.

Sunday - 8
Monday - 5
Tuesday - 3
Wednesday - 5
Thursday - 5
Friday - 0

I will now note that my data is formed from candles that open at 5 PM Pacific Time. I'm in Arizona. I think this means my candles form at 00:00 GMT. Correct?

Unfortunately the data set is small. Except for Friday and Sunday trades I think the numbers are pretty evenly averaged out. Meaning, it would not be worth while to count out certain days over others since they play relatively the same. I believe if a larger data set it used then friday and sunday will average out the same as well.

The only way to play each day differently is to look at the probabilities from those days discreetly. For example, see how many Tuesdays had the greater movement up. How far up did it move up from the open? How far did it move down from the open? How does this compare to the typical Wednesday movement? There are more questions to break it down, but I'm not going to complicate things right now. I don't have the capabilities to look at any weekday specifically so I can not create a system for different days. You seem to be able to and I see nothing wrong in doing so. You'd essentially be creating 5 different variations of the same system.

Something we should consider though is trader psychology. In the quick data I have looked at there was not one single up day that stalled at 61 pips up. It went from 5 days stalling at 60+ then only a few days at 62+, 63+ and then several at 65+. I think I can safely say this is because of traders using even numbers to make their trades. So I think we should alter our straddle accordingly to our TP. We could have saved 5 trades from stalling on us if we had simply entered at 1 pip more, or 61+ from open.

Oddly enough, there was only one day that stalled at 70+ and then no days at 71+ or 72+. Then there is one day at 73+ and several days at 75+. We should find a level that has few days of stalling on it, but also compliments our TP such that we're allowing enough opportunities to reach our TP. We can't put a straddle of 100 and expect to TP of 80 many times. Right now I think a 61 pip straddle could be best since it will filter out being on the wrong side of the candle much of the time and still allows a candle to move many pips on an average day. If we have a 30 TP it is not unreasonable to expect it to get hit frequently; even a 40 may not be unheard of.

I'm starting to wonder if we should use only 2006 data to come up with our probabilities though. Did GBP/USD trade in 2004 like it trades now?

Ignored

Quoting 62a94

Disliked

JILLYB can you tell me on your backtesting results how many consecutive losing trades you had in a row and how many winning?

MR. DETROIT

Ignored

Quoting llambert3

Disliked

ok, you went about it differently than me I suppose. I liked Mircardo's 50 straddle, so I went backtesting. At first, just September -15m to get accurate s/l data. I noticed far too often it hit his 50 then reversed. I kept increasing it until I came to the solution of 60. What I have is a spreadsheet for each day. Beside the date I put levels of 5, 10, 15, 20, 25 up to 70. If after the 60 straddle hit it went up say 40 pips, I subtraced my 5 spread and said that day produced 35 pips. I then put the number in each column (ie 5, 10, 15, 20, 25, 30, 35) If after it went up 40, it reversed and fell drastically to hit the s/l, I put the s/l (in my case -$40) in the other columns up to the 70 columns. If the trade stalled and the day closed (either positive or negative) I closed the trade at whatever the open of the new day was and the result was put in the other columns up to the 70 column.

So after September I decided to go back from March to now. A friend of mine was reviewing the data with me. He suggested looking at the days of week, and even if one Thursday failed, the chances of the next Thursday failing. He gave me the ideas, I did the work.. LOL

At first I did percentages based on whether the trade was positive or negative (regardless of whether it hit the number for that column - meaning if for column 35 pip, the trade closed at end of day with only +12, I counted it as a winner). Then we broke it down to whether the actual pp was met for a percentage. Finally, we broke each set of data out for the days of the week. I had Monday - Friday broken out into sections. I looked at these percentages and it was amazing. Wednesday alone from 3/8/06 to today for 5 pips up to 70 would all be a net negative number. So why trade Wednesdays? I choose to ignore Wednesday. Monday was positve across the board, but barely, like less than 100 pips since March, so it too was useless to me.

As far as your other questions. Yes, your 5 pm is my 8 pm which is start of new day. I also did as you regarding straddle. I found 87 straddle weeded out a ton of problem days for 15 pips profit. You had to wait sometimes 3 days to 6 days to get in. I wanted something to trade more on. Also the day I tried it, it failed on 3 lots YIKES so I thought crap that didn't work.

Finally, finally, you are right about us using 2006 data. My friend on the percentage thing is worried because we don't have Nov-Feb data. He wants to use the Month idea too.

Ignored

Quoting Natel

Disliked

Looks lika cool website. Thanks for sharing
- New kid in the block

Ignored

Quoting JillyB

Disliked

Just a thought, but you haven't related the any of the main winning days to news days have you? NFP, CPI GDP etc? I was wondering if on these days you were more/or less likely to have winning trades.

Also what about days when there are no news releases, do they have any effect on the win ratio?

Ignored

Quoting JillyB

Disliked

Most wins in a row = 14
Most losses in a row = 3 (this was during the 7 week period from middle of June to end of July when there were 14 losing trades in a total of 35 trades)
Since then the most losses in a row have been 2.

Ignored

Quoting 62a94

Disliked

JILLYB the reason im asking is i am using a similar straddle of a 25 trigger

20 TP and a 20 SL and i have had 3 losers in a row from thursday of last week to today. if your forward testing or live trading please tell me what was you results over the same 3 days. the day are nov 16th, 17th and 20th.
maybe im doing something wrong. please responed.

Ignored

Quoting AL4715

Disliked

Hi JillyB,
A newbie here and this is my first post. I am very interest in your way of trading. Especially the Winning Ratio of > 78% on the backtesting, I am very excited..

Regards to your trading method, do you know if there is EA that follows your strategy. I know Aha made an EA for Micardo but it doesn't consider the Trailing Stop after the 15 PIP movement.

Or can somebody modify the Aha's EA so that you don't have to manually backtest.

Thanks.

AL

Ignored