Reducing the number of trades

[Chorlton]

Guys,

Thanks for your comments and explanations..... Things are a lot clearer now!!


Chorlton

 

[Vrothdar]

Have you tried using a condition of a maximum number of open trades, or not opening a new position if it increases the total amount of your account risked above a certain percentage?

The latter would allow you to take a larger number of signals as your capitals grows. You'd probably need to put your backtested results under a lot more scrutiny than you otherwise would though because the order of trades would have a much larger impact on the profitability of the system.

Personally I'd be doing as you are and looking for a filter as it gives opportunity for another edge but using something like I've suggested above might allow you to start trading your system while you develop a filter to gain more of an edge.

:edit: Just realised barjon made a similar suggestion. I Should have spent a bit more time reading the replies properly!

 

[Chorlton]

Hello All,

Currently, I have developed a mechanical system which is generating more trades than I have capital for.

Consequently, I am now investigating different approaches that I can take to reduce the number of trades which are being generated. In reducing the number of trades, the idea will be to increase the number of successful trades by filtering those which have the highest chance of becoming profitable.

Has anyone gone through a similar process and if so, what suggestions would you recommend?

Many Thanks in advance,

Chorlton

Hello All,

I just wanted to open this thread up again, as I am once again interested in looking at further innovative ways to reduce the number of Buy Signals which my system currently generates.

If anyone has any new ideas / suggestions, then I'd be most welcome

Thanks in advance,

Chorlton

 

[Adamus]

I would definitely try to increase the 'quality' of your signals - that means you're very likely to get fewer trades. That's essentially a filter in your mechanical system, but that was mentioned before, so what prevented you from finding a decent filter?

 

[Chorlton]

I would definitely try to increase the 'quality' of your signals - that means you're very likely to get fewer trades. That's essentially a filter in your mechanical system, but that was mentioned before, so what prevented you from finding a decent filter?

Thanks for the reply.

The different underlying ideas I have used in my system to generate the trades are relatively straight-forward and simple which in one way is why I believe the system is relatively robust over time. The downside to this, however, is that it is difficult to try to add other conditions (which relate to the original ideas) in an attempt to refine the quality of the entry signal.

Those additional conditions which I have included, to date, in an attempt to reduce # overall trades has resulted in either an unacceptable drop in Net Profit and/or increase in MaxDD.

Although, I can live with a drop in Net Profit, I was unable to accept any increase in MaxDD.

 

[Vrothdar]

You could try getting your system to paper trade after a loss until it wins again before placing a live trade again. That can work well if you tend to get losing/winning streaks and shouldn't adversely affect the robustness of your system too much.

:edit: Take a look here as well where I'm touting the same idea!

 

[Chorlton]

You could try getting your system to paper trade after a loss until it wins again before placing a live trade again. That can work well if you tend to get losing/winning streaks and shouldn't adversely affect the robustness of your system too much.

:edit: Take a look here as well where I'm touting the same idea!

Hi Vrothdar,

I'm actually following that thread already with interest. You've made some interesting comments on there to date.

Although I like the idea (as it would definately cut down the # of overall trades) I need to find a way of coding it so that I can backtest it to see how the system performs with it.

The idea is assuming that winners/losers appear in specific streaks rather than totally random. I guess I could step through my trade list to see if losers/winners are grouped together to see whether it has any merits.

Interesting..... Thanks for the suggestion.....

 

[Vrothdar]

The idea is assuming that winners/losers appear in specific streaks rather than totally random. I guess I could step through my trade list to see if losers/winners are grouped together to see whether it has any merits.

Indeed it does. If implemented you'd need to monitor the length of losing streaks carefully but I think it'd be easier to detect it failing than adding another indicator.

As far as back testing is concerned It should be straightforward enough to take a list of trades your current system would have taken, sort them by close time and filter out the losing streaks to give a side by side comparison (measured in pips or points won or lost) that you could then apply your money management to so you can compare the difference in results in £s.

I've actually just had a thought about the way I worked out the trades this idea would have eliminated for rnicoll -I think I've made some errors doing it. It's actually a bit more complicated than I said above if you have more than one position open at the same time which is what I overlooked in that thread. When you've got more than one open at a time you need to make a decision as to whether you close all open trades at the first loss and paper trade until a profit or just stop opening new trades until the losing streak is over.

 

[barjon]

chorlton

Having tried all sorts of things I came down to first come first served until I'm full up. I sympathise with the difficulty of trying to add other conditions, unless you try money management conditions. You could for example rank your potentials in terms of risk and reward, but there's no guarantee that they would trigger in the order you want. I tried something like that and there was nothing more galling than seeing my "top" choice fail to trigger whilst one that had to be sacrificed triggered and moved on to great things.

'course it's also galling to see a potential trade do the same when I've had to leave it because I'm full up, particularly if one that I did take does not do so well. Nonetheless, having tried all sorts, I can't find better than first come first served - if you find anything let me know :cheesy:

good trading

jon

 

[Chorlton]

chorlton

Having tried all sorts of things I came down to first come first served until I'm full up. I sympathise with the difficulty of trying to add other conditions, unless you try money management conditions. You could for example rank your potentials in terms of risk and reward, but there's no guarantee that they would trigger in the order you want. I tried something like that and there was nothing more galling than seeing my "top" choice fail to trigger whilst one that had to be sacrificed triggered and moved on to great things.

'course it's also galling to see a potential trade do the same when I've had to leave it because I'm full up, particularly if one that I did take does not do so well. Nonetheless, having tried all sorts, I can't find better than first come first served - if you find anything let me know :cheesy:

good trading

jon

Jon,

Currently I share your views!!! Its for this reason that I re-opened this thread. However, I don't believe I have explored enough ideas yet to fully put this issue to bed

But I'm not overly confident of finding something either....

 

[Vrothdar]

Having tried all sorts of things I came down to first come first served until I'm full up. I sympathise with the difficulty of trying to add other conditions, unless you try money management conditions.

That's got me thinking. If each trade you place has a fixed stop and target at the time of entry you could do a first come first served upto whichever number of trades you feel comfortable with. When a new signal is generated while you have your max number of open positions you'd close the one in the most profit as it'd be the one with the worst risk:reward.

Going to keep that one in the back of my mind - Might prove useful with certain systems although I suspect it might end up reducing profits if you regularly reach your limit for open trades.

 

[Adamus]

Have you considered buying call options instead of the underlying?

Admittedly you'd have a lot of extra work programming in the Black-Scholes formula if you have to program it. Plus you could get killed on the slippage.

 

[Chorlton]

That's got me thinking. If each trade you place has a fixed stop and target at the time of entry you could do a first come first served upto whichever number of trades you feel comfortable with. When a new signal is generated while you have your max number of open positions you'd close the one in the most profit as it'd be the one with the worst risk:reward.

Going to keep that one in the back of my mind - Might prove useful with certain systems although I suspect it might end up reducing profits if you regularly reach your limit for open trades.

Unless I'm too tired to comprehend this, this doesn't seem that logical to me :|

If I understand correctly, you are suggesting closing a winning position prematurely and replacing it with one which potentially has a 50/50 chance of making money.

... but then again if you are using Targets anyway, then you would be potentially closing it off (potentially prematurely) anyway.

In my case, I do not use Target stops as I prefer the Mrkt to tell me when I should take profits but this is because I am developing a longer term system. In constrast I can see the importance of these type of stops when trading short-term systems.

 

[Vrothdar]

You've pretty much got that right. The only way it could possibly work is, as you suggested, with target stops.

The thinking is that if you've got an open trade entered at 50, stop at 0 and target at 100 which is currently at 75 you risk is greater than reward at that time so by replacing it with a new trade you're reducing risk.

To be honest it'd probably never work but was one of those "wow here's an idea!" moments so thought I'd throw it out there. If/when I find a system of my own that fits the criteria I'll give it a test and, if I remember, post back to let you know how it went.

 

[Chorlton]

You've pretty much got that right. The only way it could possibly work is, as you suggested, with target stops.

The thinking is that if you've got an open trade entered at 50, stop at 0 and target at 100 which is currently at 75 you risk is greater than reward at that time so by replacing it with a new trade you're reducing risk.

To be honest it'd probably never work but was one of those "wow here's an idea!" moments so thought I'd throw it out there. If/when I find a system of my own that fits the criteria I'll give it a test and, if I remember, post back to let you know how it went.

I'm more than happy for you to keep throwing in those ideas... :cheesy:

 

[thierry911]

hi,

I'm more than happy for you to keep throwing in those ideas... :cheesy:

Be carefull by using too mnay filters.. not too overfit the curve.
Rather than to find filters.. maybe you should work on correlation with the other
asset you are trading... and define a maximum amount to invest...if the underlyings are high/low correlated... you can also... evaluate which asset is the weakest or the strongest at the time if you have multi signals to sell or buy....

Cutting the position which has the higher profit, to me is not a good idea because you might be in a very healthy trend.. so I would not cut it..

What you can do is.. reducing the position... if for example the volatility double etc....