Can mechanical forex systems work?

[Chorlton]

The exact win rate is hard for me to calculate now as I've made some modifications from the previous results. I'm estimating it at about 80% based on the pre modified version and my results manually closing trades using the principles now in the system.

I expect 2 losses in a row, will monitor at 3 and panic at 4. But it basically cuts the stake by 30% at each loss, and ups it by 20% at each win. I have seen a couple of winning runs of 12 trades.

It's a bit of a gamble really - the testing isn't as thorough as i would like - but I'm finding back testing hard to do. I've been working on it for so long (about 6 months) that I just want to try the blooming thing properly.

As I think I've said elsewhere I have arrived at a system that I think identifies human behaivour patterns given certain price movements in key price areas. I have been up so many dead ends with just looking for TA signals that I came to the conclusion that they were pretty meaningless for an automated system. I thought I would look for patterns, I also decided that there won't be many patterns that could be found so the best place to look for them would be in areas of stress in the price movement and try to come up with my own indicators to identify any pattern in responses to these. I think I may have done.

It continues to show symetry and also works in all the FX markets I am tracking which is my biggest pointer that it is correctly identifying patterns and is not just random.

Time will tell. I don't expect a completely smooth ride - but I am maintaining optimism for the moment.

North,

I think your approach of getting concerned after 4 trades seems reasonable. FYI Based on a win rate of 80% I calculate that there is only a 0.16% chance of this event occuring.

I'm also in the process of developing a system although mine won't be automated. I agree that it can get frustrating trying to find suitable TA patterns. But i guess thats part of the development process.

All the best with it

Chorlton

 

[north5]

Thanks for the well wishes.

Do you have any opinion on the idea of cutting stake by 30% after a loss and increasing it by 20% after a win. TBH It's a new addition and I just plucked the idea and figures from thin air.

 

[Chorlton]

Thanks for the well wishes.

Do you have any opinion on the idea of cutting stake by 30% after a loss and increasing it by 20% after a win. TBH It's a new addition and I just plucked the idea and figures from thin air.

My personal belief is that one can greatly increase the performance of a system by focusing on the Money Management, so I think you are right to investigate different ideas. In addition, the idea to increase risk when winning and vice versa when losing makes perfect sense (Anti-Martingale approach)

Unfortunately, I cannot comment on this particular idea. However, I would highly suggest that you effectively backtest it before going live to ensure the results are as expected.

I remember playing around with the risk parameters in my system, (%risk per trade, Max Position Sizing, etc) and what surprised me was that small changes had a great impact on the overall system !! ... and sometimes not always as expected!!!

So in essence, by all means investigate new ideas but unless you can effectively test it, I would maybe suggest just adopting a simple approach......

 

[north5]

Thanks - I think I will go with this for now, and try and optimise it as I go.

Quick Question: Why would you not automate your system when it's finished?

 

[Chorlton]

Thanks - I think I will go with this for now, and try and optimise it as I go.

Quick Question: Why would you not automate your system when it's finished?

Its a longer term system which generates on average one trade per week. Seems pointless to automate this. Also, as this is my first system which I have developed from scratch I would prefer to see how it performs using Human intervention

Although I can definately see the merits with automating a system, one needs to have 100% confidence that it will perform as expected regardless on what the Mrkt throws at it. Issues such as Black Swan events, which can't be coded in advance may affect it in ways that one may not have accounted for.

Another issue could be hardware failure for example . Say you have a trade on and then something unexpected happens. What happens to that trade? How is it monitored in case the stop is hit ??

These are only examples, and there are many people who trade in this manner with probably great success but for the time being I would rather use a more traditional approach

 

[rnicoll]

OK - I've made the adjustments I wanted too - and added some auto money management. The success rate is now pretty high on this system so I'm going to give it a go live. Most of you will have a heart attack when you realise that I'm going to be risking about 10% per trade - but this drops pretty quickly if it starts losing on trades. It averages about one trade per day - but doesn't trade everyday - and some days it trades more (friday for instance it opened 8 trades - 7 were successful)

To give you some idea - a success wins 20 pips and a loss loses 30 pips (again please don't have a heart attack on my behalf) - at the moment with about 80% success.

Okay, my system makes 10 pips on a the average good trade and can lose 50 pips on a bad trade, with 65% success over long periods of time. I may have to be restrained physically once I take this thing live, lest I poke incessently at it. Yours isn't going to give me a heart attack, because at the end of the day I trust in the statistics to come through.

80% seems incredibly high though, I've only seen accuracy figures that high from overfitted runs of my strategy (and even then they tend not to go over 75%), how much data is that tested on?

 

[rnicoll]

The exact win rate is hard for me to calculate now as I've made some modifications from the previous results. I'm estimating it at about 80% based on the pre modified version and my results manually closing trades using the principles now in the system.

Interesting that you're closing out manually. Exit points are the biggest obstacle I'm currently facing, and I have wondered about doing that myself...

As I think I've said elsewhere I have arrived at a system that I think identifies human behaivour patterns given certain price movements in key price areas. I have been up so many dead ends with just looking for TA signals that I came to the conclusion that they were pretty meaningless for an automated system. I thought I would look for patterns, I also decided that there won't be many patterns that could be found so the best place to look for them would be in areas of stress in the price movement and try to come up with my own indicators to identify any pattern in responses to these. I think I may have done.

This sounds like you're working on a similar theme to what I'm doing, but with different indicators. Definitely, absolutely with you on developing your own indicators though. If it's in a book, odds on everyone and their dog are trading it, and you'd have to be superman to make significant profit from it. Also, a lot of indicators out there are of dubious statistical meaningfulness at best, which is okayish for helping a human spot patterns, but a computer will just crash and burn on them.

 

[north5]

It's only about 8 weeks of data - but across 9 forex pairs. As I say it's not ideal - but you've got to pull the trigger at some time! I am going to watch it carefully.

I've also got it looking at 5 other stratgies - none of which I am willing to go live with yet.

 

[north5]

Interesting that you're closing out manually. Exit points are the biggest obstacle I'm currently facing, and I have wondered about doing that myself...
QUOTE]


That was what I was doing and it helped me decide on what the best exit strategy would be. It's OK looking at the results - but watching your trades unfold and trading them yourself really helps get a feel for your system that stats just can't give you.

As I say that has led to the latest tweeks and the idea is now to just let them run.I will be keeping an eye on it - but hopefully over time that will become less and less frequent.

 

[north5]

One completely auto trade down and one win. I really had to restrain myself from taking +15 pips profit before going to bed rather than let it run. Overnight it came within 0.5 of a pip to stopping out - but it didn't. :cheesy:


I had it on £10 a point so +£200 so far - 100% success.

Leaving it on £10 a pip for now.

 

[rnicoll]

One completely auto trade down and one win. I really had to restrain myself from taking +15 pips profit before going to bed rather than let it run. Overnight it came within 0.5 of a pip to stopping out - but it didn't. :cheesy:


I had it on £10 a point so +£200 so far - 100% success.

Leaving it on £10 a pip for now.

Very nice!

I've got mine on batches of 100k (I know, it should be tethered in a specific currency). Just made 20CAD, but should have made more (I closed out early). Then it dipped a bit lower, tried re-opening the position, got buy/sell mixed up, and... well, there's a reason I'm trading on automated systems and not normal platforms, y'know?

 

[north5]

OK second trade and this one was never in doubt.

2 days +40 pips +£400 100% success


I know it is seriously hard not to poke it Rnicoll but we have to learn to trust what we've done! Good luck with yours now it's live(ish)

 

[rnicoll]

OK second trade and this one was never in doubt.

2 days +40 pips +£400 100% success


I know it is seriously hard not to poke it Rnicoll but we have to learn to trust what we've done! Good luck with yours now it's live(ish)

I've been playing with some alternative closing strategies, which are designed to:

1. Lose less money on bad trades (better for my nerves).
2. Be easier to trade algorithmically.
3. Involve less time in the market.

They're forming up rather well. I'll be keeping my system online for bits and pieces of time when I think the market is in a good state, but this might be a better plan. Sneak preview, on unoptimised "best guess" numbers for GBPUSD: 5.4% profit (without margin) over 6 months, max drawdown 1.14%. Now, that's half the profit, but it's also a quarter the drawdown, which should help calm my nerves.

 

[north5]

I had a good day with it today - but my nerves are still shot to bits. There is something about watching that makes you feel completely out of control even though you did design it.

To date (live hands off):

8 trades, 7 wins, 1 loss (0.875 win ratio) - 110 pips - £1100 gain (I'm still just keeping at £10 a pip)

So that's a 50% profit margin in two days - if it keeps that up then I will be happy.

 

[north5]

OK overnight there were 4 trades - all wins.

So live system to date:

2 days - 12 trades, 11 wins, 1 loss. + 190 pips.

However - I decided to drop the stake to only £4 a pip before I went to bed (nothing worse than waking up to a smaller account than you went to sleep with) and it only actually opened 1 of the trades as the web interface to the SB account froze for some reason. So if all had gone well and I had just left it alone I would have made £800 overnight - instead I made £80.


I know this is premature - but could I have found the holy grail? :whistling

 

[Rossini]

you would be the only one ever.....

 

[the blades]

I know this is premature - but could I have found the holy grail? :whistling

no doubt this comment is tongue in cheek but;

I'd again advise you look at the forex system's on C2. Lot's of holy' grail's......for a few months. Think about how to milk, then get off the ride with most of your gains. IMHO.

UTB

 

[north5]

It was tongue in cheek, but if it would work for another six months then it could make a LOT of money so would be a holy grail of sorts.

Anyways - two losing trades in a row What to do now? It's going to get nervy again.

To date:

14 trades , 11 wins, 3 losses (78% still) - +130 pips.

The good news is that there must be a bug in order execution because it didn't open them :clap:. I suppose that makes up for the winners it should have opened over night. Strange thing is opening trades has worked for months so some monitoring and digging required.

 

[rnicoll]

It was tongue in cheek, but if it would work for another six months then it could make a LOT of money so would be a holy grail of sorts.

Anyways - two losing trades in a row What to do now? It's going to get nervy again.

To date:

14 trades , 11 wins, 3 losses (78% still) - +130 pips.

The good news is that there must be a bug in order execution because it didn't open them :clap:. I suppose that makes up for the winners it should have opened over night. Strange thing is opening trades has worked for months so some monitoring and digging required.

I'd love to know what your losing trades were, BTW.

If you haven't got unit based tests into your trading system yet, and your platform supports them, probably a really good idea to add some to check things like order opening does work properly.

Advice: drink plenty of water, stay hydrated, and possibly have a friend restrain/distract you.

I've made 3 pips today, on two trades. Had the live version offline when a 15 pip trade would have gone through *sigh*

 

[rnicoll]

While I'm thinking about it... do you ever find yourself staring at your system going "Why won't you trade?" when it seems to be missing obvious money? I mean, I know I'm just seeing artefacts of the statistics (I'll miss some winning trades, but hopefully less than I miss losing trades), but it's kinda frustrating...