Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

Trading the Sixths with Snowballs 68 replies

Gann High-Low System with Laguerre/Fisher/VSA/Sixths and others 124 replies

My Sixths trading journal with TOB EA 50 replies

sixths indicator 1 reply

Steve Trading Journal 3 replies

Attachments: Sixths trading - an EA by macman, Bob and Steve
Exit Attachments

Sixths trading - an EA by macman, Bob and Steve

Quoting duffypratt
I have a few questions. 1) Is anyone planning to trade this at 50:1 leverage? What lot sizes do you think are appropriate per $10,000 in the account? What's the smallest
account you think you could start using that little leverage?
I'm running 15 pairs on a $10K demo, with dimes. Float has been fluctuating between 1-4%. I'm thinking about putting TB on one of my 50:1 live accounts, but I will probably
only trade 6-8 pairs. I will still use .3/.1 at 10-cents/pip, and see how it works out. On a small account, I would be careful with lot-sizing and numbers of pairs being traded. Best to test it out, first, with Monopoly money.

Rob
Quoting duffypratt
I'm trading 10 pairs....
If I go live with this, I plan to trade it like I am on demo now. I'll be limited to 50:1 with IBFX. 39K using .05 lots with a L2 trade when down 200 pips and maybe another level at -400 pips , .05 lots. at each level. , primary trades only . Won't be able to hedge unless I set up a second account with IBFX. I'll use the 1,1,3 recovery if necessary but would hope to get out on a retrace( with as small a loss as possible) long before that would kick in. So far I've taken some smallish losses on the GBP/JPY on retraces but I have DD building with that pair again. I'm thinking it might be safer to periodically shut down all trades taking a small loss. I've done that once in 2 weeks. Results after 2 weeks on demo- over $800 banked with 1.5% DD at present.
Quoting duffypratt
I'm trading 10 pairs. Using defaults on a version that is almost two weeks old. I will load the latest/greatest version this weekend. Trade size is mini lot for primary, and .3 mini for secondary. The account is a 10K demo.

For the week I banked 432 dollars, and I'm currently have a drawdown of 162 dollars. That's 2.7% for the week.

I have a few questions. 1) Is anyone planning to trade this at 50:1 leverage? What lot sizes do you think are appropriate per $10,000 in the account? What's the smallest account you think you could start using...
I am live now with a 2500 dollar account , with 25 pairs but with the lotsizes as I mentioned previously on a post ---- 0.02 P , 0.01 S
I also have another EA on a blank chart screen which has a "close all trades" just in case my Equity gets below a certain level , and I am not there to catch it. Presently set at 2000,,,,
I agree with what Macman says that it is good to have lots of currency pairs being traded with a lower lotsize and to hopefully work on the assumption that the winners outweigh the losers and you get into credit
I know that 2500 is not too high, but things are working , considering I started at 2000 two weeks ago
When I get to 3000 , I am contemplating 0.03 P 0.01 S , at 4000 -- 0.04 P 0.02 S (see previous post)
I also have SL set at about 600 but this is just to protect me in case one currency pair goes "doo lally" quickly
It works with recovery as long as the SL is far away enough
I have set alarms and email sending for when the blue BE line starts to wonder away from the white line (if it has appeared that is ), and I am not there to catch it
are you planning on leaving the MPTM settings at default?
Goodmorning,

below is my statement. 28 pairs.

the settings.

0.1 lots , secondary trades.
0.3 lots , primary trades.

using macman recovery system on.

option 1.1.2.4

the rest all standard settings.

as you can see, it closed 4700 in profit . however a drawdown ( open positions) of 3400

the biggest problems are EURCHF and EURCAD and CHFJPY.

my question would be. what if i would have added the option hedge on. would this have made a difference on these pairs?

i added the whole report in zip file as well. with fake account name and number for study purposes. i really like to know what other people woul dhjave chosen for options to counter this huge drawdown. ( open positions)
Attached Image (click to enlarge)
Click to Enlarge Name: beastweekoct21th.jpg Size: 223 KB
Attached File(s)
File Type: zip beast28pairs.zip   27 KB | 215 downloads
Friends,

For the first time in more than 10 years, I have taken the (for me) most unusual step of loading up "someone else's system". My own system has served me well for very many years and I work on the principle that if it ain't broke, don't fix it.

But given the pedigree of TB, I felt compelled to take a look, and I must say I am intrigued by the possibilities.

In order to get a real feel for the behaviour of the system, I have traded this manually for a few weeks, and laterly, have let TB loose on a demo account.

Let me say straight away, that I am cautiously optimistic. If that sounds unenthusiastic, then forgive me. But after 25-odd years in trading and all the snake-oil I have seen, I rarely get too excited before I have "smelled the sausages". So "cautiously optimistic" is a ringing endorsement!!

My interest is in how we can squeeze the absolute maximum from the system. My thinking is as follows...

I have undertaken a detailed statistical analysis of the method on the EU pair. The outer limits of the Sixths (ie the 0-sixth and the 6-sixth) are the High and Low of the last N bars. Macman currently recommends 170 bars for EU on the H1.

I am going to call these +/-3 standard deviations. (They are not, of course, but the error is less than 5%). This conveniently means that each sixth line is 1 SD apart, with the MiddleWhiteLine (the 3-sixth) being, by definition, the mean.

What were are interested in for success (ie trades closing in profit) is a retrace from the outer lines to the middle. I have done a detailed retrace analysis over recent history and plotted the retrace probabilty density function (see pic). As expected, it is approximately Rayleigh distributed.

Parameter fitting to the higher order moments of these distributions is notoriously difficult, but a reasonable estimate is 1SD = 75 pips.

This implies that ideally, each sixth line should be 75 pips apart, making a total top-to-bottom distance of 450 pips.

Currently, as of close on Friday, my chart is showing a top-to-bottom distance of 463 pips, which is very close. This implies that Macmans current estimate of 170 bars is very reasonable.

However, the retrace is obviously a function of the stochastic volatility, which is always changing, albeit quite slowly. Therefore, I believe that there might be some value is reassessing the optimum number of bars to use on a regular basis, based on this kind of analysis. If this can be coded in MQ4, then we can add it to the robot and it would be self-optimising.

Best regards

Old Dog
Attached Image
Quoting Old_Dog
Friends,

For the first time in more than 10 years, I have taken the (for me) most unusual step of loading up "someone else's system"....
Excellent post - you just made my day

When Steve coded this strategy for me, he asked if I was willing to publish here. Over the years I have gained a lot of knowledge from various inspired posters, and felt I could give something back, hence this thread - thanks are entirely due here to Steve for his untiring efforts.

Since the first posts I have been waiting to see if anyone actually realised that TB was published unfinished. I had expected the first posts to be about how the look back bars were calculated - which is the whole crux of the bot. It is obvious that for optimum results each pair needs to use a different look back history - currently the vast majority of pairs are using my default setting which may be a long way from being the best setting.

Quite honestly I have been a little disappointed up until now with the direction the thread has taken, so this post is a ray of sunshine.

Yes Old Dog, the bot does need some sort of auto optimisation routine and you seem to be the only poster so far to realise that - hats off to you

So, if there is anyone out there with ideas along these lines, please speak up - between us all I am sure we can come up with a robust (mostly) profitable bot.

Good weekend to all
Old_dog:

What exactly are you plotting and proposing to optimize? I think I don't quite understand. But I'm definitely interested in the statistical approach.

What we need to optimize is not the probability of a profit but the probability of a profit * its profit - probability of loss * its loss.

All in all this is a "short volatility" strategy which will return many smaller winners with some small probability of a big loss. It takes a longer time series to see if the actual return is positive.

The issue I have is that TB doesn't always fully take losses completely automatically. (I got a backtestable version going, and it works fine until it gets into a steady trend and then trades stay open for months with ever increasing losses, wiping out all profits and more).
Guys

One thing I didn't mention in my last post, as I'm rather worried about over-complicating things, is that when I have been trading this manually, I have not just set buy and sell pending orders when the market passes the Gold lines.

I am taking into account the asymmetry of the price distribution (see chart). As you can see, over the 170 bars, the market has spent a lot of time around the MiddleWhiteLine, which is exactly what we want to see.

But due to the technical behaviour of non-stationary time series (please Google "ARCH GARCH" if you want more info), the heteroskedasticity of the distribution leads to considerable asymmetry.

It is clear from the chart that the market has spent more time near the TopGoldLine than near the BottomGoldLine.

This gives a clear steer as to which pending trades are likely to be more profitable.

Best regards

Old Dog

PS. In response to mbkennel who says "...It takes a longer time series to see if the actual return is positive...", I would agree completely. The long-term equity curve of a short volatility strategy like this is governed by a two-tailed chi-squared distribution. So, for government work, we can say by rule of thumb that we need at least 62 trades (2 x 31 for each tail) for each degree of freedom, at the 95% level of significance. This system has quite a few variables, so we will need a bit more trading time under our belts before we can draw firmer conculsions.
Attached Image
First of all thanks Steve because you add me to your buddy list
and sorry for my language
1-what is ability for adding filter like RSI I knew you add it ) or any other Indicator to the parameter (as a list ) and the order stop when meeting this filter
2-also the EA doesn't deal with any symbol outside this filter
3-also the EA deal with currency only in one direction or at least 75% in direction
i.e ( EUR buy or sell ) not in specific symbol as buy and the other as sell ( not hedging) at least 3 buy for one sell or vice verse

4-also we can avoid the symbol which come to outside the golden line with high speed ( high slope ) because this symbol forming in this case flag formation which is continuation pattern

Thanks
I'm not a programmer and i don't like ! , but only I try to catch my view !
After running the Beast for eight days, I decided to parse the history with MTreport4 and get to the nitty gritty. Based on the stats, I found that eight pairs had stellar performance, while seven pairs were subpar. My feeling, now, is why add pairs just for the sake of adding pairs, if they don't all contribute to the profitability of the strategy? The weak pairs add additional "drag" to the account, via increased float, DD, etc., not to mention added margin requirements.

I know everyone has their pet pairs, but these are the ones that did really well for me. As the saying goes, "The bullsh*t stops when the green flag drops."

eurusd - usdjpy - gbpusd - audusd - eurjpy - usdcad - nzdusd - audjpy

Rob
Quoting ChicagoRob
After running the Beast for eight days, I decided to parse the history with MTreport4 and get to the nitty gritty. Based on the stats, I found that eight pairs had stellar performance, while seven pairs were subpar. My feeling, now, is why add pairs just for the sake of adding pairs, if they don't all contribute to the profitability of the strategy? The weak pairs add additional "drag" to the account, via increased float, DD, etc., not to mention added margin requirements.

I know everyone has their pet pairs, but these are the ones that did...
Here's my every pair's performance in two weeks.
Attached Image
Quoting luiboy
Here's my every pair's performance in two weeks.
Looks like we have some common winners, but others appear to be opposites.
Differences can be attributed to start time, and/or different sets, etc. Are you utilizing MPTM?

Rob
Quoting ChicagoRob
Looks like we have some common winners, but others appear to be opposites.
Differences can be attributed to start time, and/or different sets, etc. Are you utilizing MPTM?

Rob
Yes. But sometimes when I want to change my settings I would just close all open trades regardless of the floating P/L. So those losers are actually are not as I just purposely closed them.
Quoting Old_Dog
Parameter fitting to the higher order moments of these distributions is notoriously difficult, but a reasonable estimate is 1SD = 75 pips.

This implies that ideally, each sixth line should be 75 pips apart, making a total top-to-bottom distance of 450 pips.

Currently, as of close on Friday, my chart is showing a top-to-bottom distance of 463 pips, which is very close. This implies that Macmans current estimate of 170 bars is very reasonable.
Taking one step at a time: are you saying that we should be looking at the number of bars back which contain a high and a low that are as near to 475 pips apart as possible?

Quoting ChicagoRob
After running the Beast for eight days, I decided to parse the history with MTreport4 and get to the nitty gritty. Based on the stats, I found that eight pairs had stellar performance, while seven pairs were subpar. My feeling, now, is why add pairs just for the sake of adding pairs, if they don't all contribute to the profitability of the strategy? The weak pairs add additional "drag" to the account, via increased float, DD, etc., not to mention added margin requirements.

I know everyone has their pet pairs, but these are the ones that did...
Cheers Rob.

Quoting ChicagoRob
After running the Beast for eight days, I decided to parse the history with MTreport4 and get to the nitty gritty. Based on the stats, I found that eight pairs had stellar performance, while seven pairs were subpar. My feeling, now, is why add pairs just for the sake of adding pairs, if they don't all contribute to the profitability of the strategy? The weak pairs add additional "drag" to the account, via increased float, DD, etc., not to mention added margin requirements.

I know everyone has their pet pairs, but these are the ones that did...
Hi Rob,
How long have you been using MTreport4? I picked up a copy of it about 3 weeks ago and I'm just getting the hang of it now....Looks like it could be a help....
Very flexible with the exception of the one computer only rule....I'd like to have it on all my demo units but that doesn't look to be possible.
Be interested to hear how you like it.......

whip
Quoting SteveHopwood
Taking one step at a time: are you saying that we should be looking at the number of bars back which contain a high and a low that are as near to 475 pips apart as possible?

Hi Steve,

With current volatility - yes. But the numbers have been quite different in the past and will no doubt change again in the future. It is just a matter of keeping track of the volatility and adjusting the lookback accordingly.

If we want to be even more sophisticated, we can track the asymmetry as well. In my manual trading my "primary" and "secondary" trades are based on retrace probability, so longs and shorts do not have equal weight based purely on penetration of the gold lines. Sometimes longs are more likely to retrace and sometimes shorts might be a better bet.

(IMVHO).

Very best regards,

Old Dog
Quoting Old_Dog
Hi Steve,

With current volatility - yes. But the numbers have been quite different in the past and will no doubt change again in the future. It is just a matter of keeping track of the volatility and adjusting the lookback accordingly.
Thanks. So,

 

  1. do we have a way of measuring current volatility?
  2. if so, is there a formula for calculating the standard deviation we need here?

Cheers

Quoting ChicagoRob
After running the Beast for eight days, I decided to parse the history with MTreport4 and get to the nitty gritty. Based on the stats, I found that eight pairs had stellar performance, while seven pairs were subpar. My feeling, now, is why add pairs just for the sake of adding pairs, if they don't all contribute to the profitability of the strategy? The weak pairs add additional "drag" to the account, via increased float, DD, etc., not to mention added margin requirements.

I know everyone has their pet pairs, but these are the ones that did...
I wouldn't draw that sort of conclusion after eight days. Your pairs all have either USD or JPY. Limit your trading to these pairs, and you may have serious trouble when one of those currencies goes strong or weak for a sustained period.

As I understand it, one of the main points of this system is to do some automatic hedging of risk by diversifying the currencies over which it trades. So, the "drag" you noticed results from the design of the system, and over time it should be a good thing. (The same thing that is dragging your winners now, will hopefully drag your losers later.)
0 traders viewing now