Confidence in your strategy?

[LiamH]

This last two weeks I have been on paternity leave from work and have been day trading forex, some would call it scalping.

Basically, I trade range breakouts - nothing new or original, generally looking at the 1\5\15 min timeframes while being mindful of what's going on in the bigger picture. I trade generally a 10 pip stop, 10 pip target but if the trade has some momentum can let it run.

Originally I started with GBP\USD and a couple of days ago decided I would trade EUR\USD instead. I don't quite know why but I'm sticking to one market so EUR\USD it is.

I have placed 19 trades, 10 losers and 9 winners. I have made 3 mistakes in terms of profitable trades becoming losers through me not sticking to the rules properly and I have identified a range maybe 7 or 8 times which has broken out and travelled my 10 pips while I've gone off to do something else quickly.

My account at this moment is 3.45% up and I risk 3% on each trade (only $100 account on fxcm micro so no major cash on the line).

The highest drawdown to now has been 8% and the highest profit has been 7% but like I say, I have missed a good handful of profitable opportunities and have also made a couple of mistakes.

Without these mistakes and without missing the breakouts of ranges I had clearly identified I would be 20-30% up over these 19 trades.

So... How big a sample size do you need before you are comfortable that the strategy you are using works? I still get a little irritated when I have 3 losers in a row and this has happened twice but I don't want to make the mistake of ditching what could be a sound system. Also, I don't know how confident I can be when there is some discretion involved - For instance, the figures are a little skewed because originally, for say the first 5 or 6 trades I was risking 5% - Ended up down at $92 then had a trade that got some momentum behind it which I let run and that put me back over $100. This is the point at which I dropped to 3% risk which I am much more comfortable with. The thing is, the 'runners' are what have kept my head above water but they are discretionary trades, I guess they could have turned around and been scratched for b\e.

In the same breath, had I not missed some opportunities and had not made any mistakes then I would have been up with just 1:1 R:R.

So yeah, how long before you become confident in your strategy? What do you need to see and go through before you can say it works?

EDIT: Thought I should add, I've only had two 'runners' and I don't always allow my stop\target to get hit, some of the losers have been 5-8 pips and likewise some of the profits have been smaller where I've not liked what's been going and decided to close. Until now, two of these have been wrong decisions - There was one yesterday that I closed at 3 pip loss I think but had I left it my target would have been hit.

 

[SanMiguel]

This last two weeks I have been on paternity leave from work and have been day trading forex, some would call it scalping.

Basically, I trade range breakouts - nothing new or original, generally looking at the 1\5\15 min timeframes while being mindful of what's going on in the bigger picture. I trade generally a 10 pip stop, 10 pip target but if the trade has some momentum can let it run.

Originally I started with GBP\USD and a couple of days ago decided I would trade EUR\USD instead. I don't quite know why but I'm sticking to one market so EUR\USD it is.

I have placed 19 trades, 10 losers and 9 winners. I have made 3 mistakes in terms of profitable trades becoming losers through me not sticking to the rules properly and I have identified a range maybe 7 or 8 times which has broken out and travelled my 10 pips while I've gone off to do something else quickly.

My account at this moment is 3.45% up and I risk 3% on each trade (only $100 account on fxcm micro so no major cash on the line).

The highest drawdown to now has been 8% and the highest profit has been 7% but like I say, I have missed a good handful of profitable opportunities and have also made a couple of mistakes.

Without these mistakes and without missing the breakouts of ranges I had clearly identified I would be 20-30% up over these 19 trades.

So... How big a sample size do you need before you are comfortable that the strategy you are using works? I still get a little irritated when I have 3 losers in a row and this has happened twice but I don't want to make the mistake of ditching what could be a sound system. Also, I don't know how confident I can be when there is some discretion involved - For instance, the figures are a little skewed because originally, for say the first 5 or 6 trades I was risking 5% - Ended up down at $92 then had a trade that got some momentum behind it which I let run and that put me back over $100. This is the point at which I dropped to 3% risk which I am much more comfortable with. The thing is, the 'runners' are what have kept my head above water but they are discretionary trades, I guess they could have turned around and been scratched for b\e.

In the same breath, had I not missed some opportunities and had not made any mistakes then I would have been up with just 1:1 R:R.

So yeah, how long before you become confident in your strategy? What do you need to see and go through before you can say it works?

Backtest?
Try RR bigger than 1:1 at just 10pips, the spread is going to kill you in the long run.
I'd want 100 trades before considering the ein ratio and RR of any strat.

 

[LiamH]

Backtest?
Try RR bigger than 1:1 at just 10pips, the spread is going to kill you in the long run.
I'd want 100 trades before considering the ein ratio and RR of any strat.

I can't really backtest because it's difficult to define a range within a computer program - I have looked back over all of the data I can and it seems profitable but looking back in hindsight is completely different to trading live.

The 10pip stop\target are just a general rule (looking for tight ranges), my stop is the size of the range and my initial target always 1:1 RR but am comfortable letting it run if there is some momentum.

The spread doesn't worry me, if I hit my target often enough to be profitable then fxcm are quite welcome to take their spread. I understand the profit would be much more without it but this I guess is an issue with trading this way that cannot be avoided.

100 trades you say... Would you think this is enough of a sample to judge the discretionary aspect on or would you just suggest 100 trades for a completely mechanical system?

 

[Shakone]

Without these mistakes and without missing the breakouts of ranges I had clearly identified I would be 20-30% up over these 19 trades.

Sounds like you're being a bit selective here, rather than realistic. It is irrelevant how much you think you could have been up over the 19 trades. I used to do something similar. I would look at an entry my system gave me, and see it went for 80 pips, and look and notice I had only taken 15 pips, and think wow, if I had let it run, I could be this % up etc. But in reality I was noticing ones that went for me superbly, and forgetting all those that went 15-20 pips, and then turned and would have stopped me out if I had let them run. I ignored them, because I was already out of them with my profits. Also because there is a natural tendency to look for info that supports your idea (very common and dangerous). Also you mention you missed several trades, which you then assume all would have won. But you're probably missing some that in reality you also would have taken that would have been losses.

You can try to improve and reduce mistakes, but mistakes are part of it and you will continue to make them, and you will continue to miss good trades (or bad ones too). So all that matters is how much you are making relative to the amount you are risking.

Number of trades at least 50 to be slightly confident. More than 100 to be reasonably confident, etc.

 

[LiamH]

Sounds like you're being a bit selective here, rather than realistic. It is irrelevant how much you think you could have been up over the 19 trades. I used to do something similar. I would look at an entry my system gave me, and see it went for 80 pips, and look and notice I had only taken 15 pips, and think wow, if I had let it run, I could be this % up etc. But in reality I was noticing ones that went for me superbly, and forgetting all those that went 15-20 pips, and then turned and would have stopped me out if I had let them run. I ignored them, because I was already out of them with my profits. Also because there is a natural tendency to look for info that supports your idea (very common and dangerous). Also you mention you missed several trades, which you then assume all would have won. But you're probably missing some that in reality you also would have taken that would have been losses.

You can try to improve and reduce mistakes, but mistakes are part of it and you will continue to make them, and you will continue to miss good trades (or bad ones too). So all that matters is how much you are making relative to the amount you are risking.

Number of trades at least 50 to be slightly confident. More than 100 to be reasonably confident, etc.

Yeah, maybe I am being a little selective but I don't mean to be. I try to be as realistic as possible... The trades that have broken a range while I have been away, and when I say this I mean that I had already drawn the supp\res lines and was waiting for a break before I left the computer - These really were all profitable in the sense that they hit my 10 pip target. Maybe my stupidity could have made them losers but on the face of it, had I seen them break I would have made a profit.

This does not mean I am confident in the strategy though - This is why I started the thread, I know this sample size is tiny and was just wondering at what point you guys would think I can say it's a profitable way to trade long term. I mean, I know that the more trades the better and nothing works forever but we must have a general rule? So far it seems to be 100 trades...

 

[Pi3141]

100 trades seems way to small for me. I'm sure there are some traders who will do 10-20 trades a day if they are a day trader or scalper. So if that were the case then that trader's confidence in their system is only based on about 2 weeks worth of data.

I'm demo trading at the moment and currently testing out a strategy. What I'm doing is using the smallest stake possible (£1 per point) to try and reach £500. The point about using the smallest stake possible is that it makes testing more valid. Its easy to win 500quid if your stake per point is £10 because that only means +50 points and you've achieved your target. But if the stake is only £1 then it means you need to accumulate 500pt to reach your target, which is much harder but much more credible.

 

[LiamH]

100 trades seems way to small for me. I'm sure there are some traders who will do 10-20 trades a day if they are a day trader or scalper. So if that were the case then that trader's confidence in their system is only based on about 2 weeks worth of data.

I agree... In poker, a profitable player can have break even or losing stretches which span more than 100,000 hands due to variance. Now, I know that poker has far more variables than trading but it does worry me that I could trade this way for the next three months and then suddenly lose it all because it turns out the strategy was just in an upswing at the time I started trading it!

 

[meanreversion]

Are you absolutely sure you can't backtest this strategy, or is it just beyond the limitations of your coding ability (this is not meant to be an insult, it's a genuine question).

If it's really not backtestable, then you'll just have to keep going and after a couple of years you'll know. That's the only answer I'm afraid. I've backtested many strategies which might lose money over 5 years but have periods of 4-6 months when they're making money (or vice versa in fact), so any sample of under a year isn't going to tell you very much.

What I would say is that you should try to reduce your trade frequency, maybe have a 20 pip stop (in half the bet size) and look for 40 pip profits. If you trade too much you will a) give too much money to your broker and b) have no time for your family ....

 

[LiamH]

Are you absolutely sure you can't backtest this strategy, or is it just beyond the limitations of your coding ability (this is not meant to be an insult, it's a genuine question).

If it's really not backtestable, then you'll just have to keep going and after a couple of years you'll know. That's the only answer I'm afraid. I've backtested many strategies which might lose money over 5 years but have periods of 4-6 months when they're making money (or vice versa in fact), so any sample of under a year isn't going to tell you very much.

What I would say is that you should try to reduce your trade frequency, maybe have a 20 pip stop (in half the bet size) and look for 40 pip profits. If you trade too much you will a) give too much money to your broker and b) have no time for your family ....

Probably beyond my backtesting ability - I wouldn't know how to define what I see on the chart in the backtesting software - Basically just tight ranges and wedges looking for breakouts. I don't have a formula to say exactly what constitutes a tradeable range... I just see it.

This particular strategy wouldn't work with larger stops\targets or at least I don't think it would, I haven't actually tried\tested it.

As you know, I am looking to trade longer TF's but this is more of a little project - Something to trade when I'm on late shifts or have days off in the week.

 

[meanreversion]

You can try to improve and reduce mistakes, but mistakes are part of it and you will continue to make them, and you will continue to miss good trades (or bad ones too). So all that matters is how much you are making relative to the amount you are risking.

Number of trades at least 50 to be slightly confident. More than 100 to be reasonably confident, etc.

Herein are a couple of good reasons for automated trading (and if that cnut ODT pops up on this thread I'm leaving).

Firstly, your trading system is completely rules based and removes cognitive biases which hinder consistent and profitable trading.

Secondly, the automation removes scope for error (provided the coding is done correctly, the software/data feed/broker don't malfunction in some way).

 

[Shakone]

Agree with meanreversion and pi, but at the end of the day, you NEVER will know if it will work forever, what exactly the edge is, or whether that edge will alter over time. There isn't going to be a guarantee, so you have to make a judgement based on a number of trades whether it is worth pursuing or not. You simply can't spend 5 years on a system only to see that it doesn't work, so you have to have some cut off point earlier than that. I think 100 trades is a good middleground.

 

[meanreversion]

Probably beyond my backtesting ability - I wouldn't know how to define what I see on the chart in the backtesting software - Basically just tight ranges and wedges looking for breakouts. I don't have a formula to say exactly what constitutes a tradeable range... I just see it.

This particular strategy wouldn't work with larger stops\targets or at least I don't think it would, I haven't actually tried\tested it.

As you know, I am looking to trade longer TF's but this is more of a little project - Something to trade when I'm on late shifts or have days off in the week.

You could try trading some fairly simple trend systems on a daily timeframe. One I saw recently which looked interesting was a Bollinger trend system.. enter if price crosses 2 s.d. away from 20 day MA, and then exit if price crosses 1 s.d. (that's the trailing stop), provided the signal is in the direction of the underlying trend (defined by a short MA and a long MA). This sounds more complicated than it is.. it should be fairly easy to code up through the charts platform on IG and I imagine it would have worked fairly well this year.

 

[meanreversion]

Agree with meanreversion and pi, but at the end of the day, you NEVER will know if it will work forever, what exactly the edge is, or whether that edge will alter over time. There isn't going to be a guarantee, so you have to make a judgement based on a number of trades whether it is worth pursuing or not. You simply can't spend 5 years on a system only to see that it doesn't work, so you have to have some cut off point earlier than that. I think 100 trades is a good middleground.

Weeeeeeeeeeelllllllllllllllll.............. I understand entirely what you are saying, BUT I have found that systems tend to start making decent coin at precisely the point of maximum despair for the trader - this is the point at which you are ready to switch the thing off. So it's a little tricky.

 

[LiamH]

Weeeeeeeeeeelllllllllllllllll.............. I understand entirely what you are saying, BUT I have found that systems tend to start making decent coin at precisely the point of maximum despair for the trader - this is the point at which you are ready to switch the thing off. So it's a little tricky.

Like you say, it's tricky. I agree with the middleground of 100-ish trades, I suppose that trading such short TF's this would cover various market conditions.

I'm not a fan of automated trading in all honesty, I like that you can set something up that will do all the work for you but, perhaps I'm weird, I quite like the chart watching and manual excecution of trades. I like when I make a profit based on what I can see happening there and then rather than the cold and calculated way in which an automated system goes about it's business.

Cheers for the MA\Bollinger strat... I was looking at something similar a couple of weeks back and it appears to work well.

 

[meanreversion]

Just don't trade too often. At some point you'll find yourself close to a mental stop loss. The wife will see you sitting there, "doing nothing", and hand you the baby with the smelly nappy. Chaos will ensue.

 

[Black Swan]

You've done well not getting your ten pip stop smashed through a lot more often tbh...I thought you were going to state 20 pip stop and 10 pip target but heh well done. Even with 0.5 pip spread on the Euro (fxcm on active-trader) most would struggle to get your day trading results straight of... Carry on. What could change? Well going back to work is one thing, secondly moving up to money that's needed for a yearly supply of nappies not just one packet..

BTW if you're on paternity leave does that mean you and the wife have had a new baby? If so congrats, typical of the miserable ba5tards on 'ere not to say Yay!! :clap:

 

[LiamH]

Just don't trade too often. At some point you'll find yourself close to a mental stop loss. The wife will see you sitting there, "doing nothing", and hand you the baby with the smelly nappy. Chaos will ensue.

Ha ha ha she's quite supportive... Before handing me the baby she asks two questions - Are you watching a setup and are you in a trade! She's a good'un.

 

[LiamH]

You've done well not getting your ten pip stop smashed through a lot more often tbh...I thought you were going to state 20 pip stop and 10 pip target but heh well done. Even with 0.5 pip spread on the Euro (fxcm on active-trader) most would struggle to get your day trading results straight of... Carry on. What could change? Well going back to work is one thing, secondly moving up to money that's needed for a yearly supply of nappies not just one packet..

BTW if you're on paternity leave does that mean you and the wife have had a new baby? If so congrats, typical of the miserable ba5tards on 'ere not to say Yay!! :clap:

Cheers for the congrats, we had a boy on the 9th

Unfortunately the spread is more like 3 pips on FXCM Micro which is a bit annoying but yeah, it seems that the strat is right 50\50 or maybe a little better without me making mistakes. It appears I do need the odd 'runner' to show a profit... Had one today so now 12% up yay! lol

In all honesty, what you've said is one of my concerns... I worry that this strat is just working now and will fall on it's backside at some point soon. The annoying thing is not knowing.

Going back to work will slow me down but I work silly shifts so will still get 50-60 hours of trading in each month during the London session.

 

[Black Swan]

Cheers for the congrats, we had a boy on the 9th

Unfortunately the spread is more like 3 pips on FXCM Micro which is a bit annoying but yeah, it seems that the strat is right 50\50 or maybe a little better without me making mistakes. It appears I do need the odd 'runner' to show a profit... Had one today so now 12% up yay! lol

In all honesty, what you've said is one of my concerns... I worry that this strat is just working now and will fall on it's backside at some point soon. The annoying thing is not knowing.

Going back to work will slow me down but I work silly shifts so will still get 50-60 hours of trading in each month during the London session.

Yay, nice one, wish I could go back and re-live my times as a young Dad over and over, great times enjoy..

The strat is one thing, but you've overlooked what's really working; sound MM and discpline will always win through whatever the conditions. Even if you get a drawdown if you keep the MM sound then you'll recover when conditions return. Remember its not you "its the market stoopid", if it's not happening you can't force it, something we still all have to remind ourselves of. Unless your strat is a bit weird/highly optimised then the comfort is (IMHO) it'll always work if you're aiming for ten pips.

Just a word on the Euro, it's the only pair I day-trade...it stuck around that major 1400 level (in a bit of a range) like Rooney to a 1200 quid a night gold digging chav. 2 things in relation to that; firstly it's one of those times that price makes a 'kin total mockery of all us hard core TA guys watching indicators, the world and every trader on the planet obviously had orders to sell and or buy at that level also it made that tight range a good 'play' screwing 10-15 pips out of the bounces nice and regular, now it's moved off that sticky area you may find things change..but you'll adapt no doubt.

As for fxcm micro if you get a 2.2-3 pip spread that's similar to the standard account spread and the micro is also NDD so you won't go wrong there, good platform, marketscope is a good package and codebase is something you may want to look into when you get back to work...

 

[wackypete2]

we had a boy on the 9th

Congrats!!
Be prepared for plenty of sleepless nights....might try to make them productive if you can trade with an eye shut and a cup of coffee in one hand.

Peter