READ-ONLY

The forum is now a read-only archive.

For bug reports & platform questions → [email protected]

Our community lives on Discord and YouTube — come join us!

Risk of weekend gaps

12 replies

Tradine

Customer, bbp_participant, community, 27 replies.

Visit profile

11 years ago #114205

Hi, 

 

the best strategies that I´ve found in the last few month was on H1 with not closing on friday. 

 

In the moment I´m a little bit insecure with the risk of weekend gaps.

 

This are my thoughts:

 

Should I hedge the positions over weekend?

Is this a intervene that could damage the hole portfolio?

Does the backtest consider weekend gaps good enough and so I´ve nothing to do?

Should I only generate strategies with closing on friday?

And so on…

 

How do you handle this things?

 

 

 

0

Patrick

Customer, bbp_participant, community, 422 replies.

Visit profile

11 years ago #132706

Hi,

 

if you are affraid, use close all possitions on Friday, its easier than hedging, btw hedging cost additional fees and work. Just check, that UTC at SQ is same as UTC of your broker to not have difference in data.

in case of unexpected situation i think it could damage your portfolio, remember that USDCHF issue at the beginning of the year.

Personally i dont use this option yet.

 

Hope that helps little bit.

Patrick

0

Tradine

Customer, bbp_participant, community, 27 replies.

Visit profile

11 years ago #132708

Hi Patrick, 

 

thank you for your thoughts.

 

I can´t just close the positions on fridays. The strategies are generated and tested with the weekends. If I would do this they don´t have the trade duration they need. In this way I would trade total different systems live. I only could do this with new buildings. But on H1 the strategies without close on friday delievers better results.  🙁

 

They USDCHF issue was intraday. Thus closing on fridays don´t have any sense for risks of this kind.

 

What do you mean with “Personally I don´t use this option”? Does it mean you are staying over the weekend? How do you handle your risk? Are the backtest results (with weekends) of the SQ enough for you?

0

Patrick

Customer, bbp_participant, community, 422 replies.

Visit profile

11 years ago #132710

USDCHF was intraday but anything other can happen in the weekend (war etc) THIS was only thing i remember that could damage your account¨…

 

I dont use it, because i am focused on other things like earn on real account. After i can think about it. I have also generated all strategies without this function. So now i am staying over the weeekends. BUT i can make you sure usually are most of positions closed, time to time is 1 trade opened during weekend. (i have 10-20 strategies H1)

 

Anyway when there is gap, you can get bigger loss but on the other side also bigger profit. SQ handle this – if there is a gap, it count it like there is a movement. So the backtest is correct in this.

0

Tradine

Customer, bbp_participant, community, 27 replies.

Visit profile

11 years ago #132712

I also get the impression there are not so many open trades over the weekend (running portfolio with 15 strategies and live testing with several other strategies). But I will have a special look for this in the next months and I will do some separate testings in TradeNavigator for finding the biggest gaps in the last ten or fifteen years for better calculating of my position sizing. 

 

Thank you Patrick! This exchange of knowlegde is very helpful for me.  🙂

0

stearno

Customer, bbp_participant, community, 379 replies.

Visit profile

11 years ago #132884

You can retest in SQ and choosing to close on Fridays.  See how it impacts the performance.

 

But one thing I want to mention.  As a guy said on podcast yesterday, “you will see yourself in your trading.”  Meaning that your personality shows through on your trading.  

 

Therefore, we need to keep in mind our personalities when choosing the settings.  

 

For example,

1. I am not a patient person.  So in my manually trading, I will enter a swing trade off the hour chart. I will check on that dang trade every 5 minutes.  I cannot just “set and forget it.”  The same applies with automated trading, short trade duration fits a lot better with my trade personality.  

2. Also, I cannot stand potential risk and being exposed. That is why my automated strategies HAVE to trade with a stop loss.  Therefore, I will not even test strategies that don’t have a stop loss or that don’t close on Fridays.  

3. Drawdown is another big one.  People think they can handle 25% drawdown, or what it feels like to have a drawdown period of 270 days.  They see their account balance go from 10,000 on March and then in Decmeber it shows 7,500.  How would they actually react.  What if drawdown of the strategy is 50%.  How many people can actually see half of their money go and not see anything recover for 270 days (talking in reality as it happens).  They think their strategy has stopped working and stops the EA, to stop loosing anymore money.    (good article talking more about this is found at (I am a customer, but otherwise not related to this website: http://mechanicalforex.com/2011/01/measuring-a-systems-psychological-pressure-the-ulcer-index.html) 

 

But those are examples from my personality and/or my personal opinion.  Everyone must find what fits them, otherwise, we will not feel comfortable with the trades and maybe start messing withe them (which does then affect the results).

 

Now that aside, with regards to being exposed: statistically you are not protected by any one event; but you are protected by the statistical probabilities across all the events.  Which means, if it gaps one time, you will see the loss.  But as Patrick points out, it could also play in your favor.  So as long as you backtested over a long enough period to reach a statistical significant sample size, then your strategy has a statistical probability that over all the gap situations you come out a winner (assuming your strategy is a winner).

 

-Stearno

0

seaton

Customer, bbp_participant, community, 157 replies.

Visit profile

11 years ago #132888

@Stearno Well Said! 

0

Tradine

Customer, bbp_participant, community, 27 replies.

Visit profile

11 years ago #132907

Hello stearno, 

 

really well said! I can see you have more experience just few months of trading…

 

The jouney in trading are the most interesting thing for me at all. The last 3,5 years have been a very hard time but also very educational at all levels.

 

Fortunately I have some experience with mechanical systems and so it isn´t a big problem for me to wait some time for new equity highs. The first SQ systems are already winners and confirm my first backtests. And so I´ve nothing to complain. 

 

Every week I learn something new and my backtests and my live results are taking a step forwards. I´m really very happy with the oportunities of StrategyQuant and I´m looking forward to all further steps I could go in the future. 

 

Thank for the interesting article.  🙂

 

Tradine

0

stearno

Customer, bbp_participant, community, 379 replies.

Visit profile

11 years ago #132921

Tradine,
Glad to hear you are rocking and rolling. It is always a pleasure to share in success with other traders!

By the way, you said you check live trading against backtests. Would you mind expounding more on your method?

Because I am considering the best way to do this. I was a little concerned to not compare 10 live trades against 10 years of backtest averages. So I wanted to see what others have found worked.

Second question: have you decided what level of variance (difference in the two numbers) is okay to you when comparing the live numbers to the backtest numbers? So for example, your back tests show 60% profitable trades with 1.78 return on drawdown. Live tests show 51% profitable trades and 1.49 return on drawdown.

-Stearno

Sent from my HUAWEI MT7-TL10 using Tapatalk

0

Tradine

Customer, bbp_participant, community, 27 replies.

Visit profile

11 years ago #132931

Stearno, 

 

I´m not very advanced in in comparing the backtest results to the livetrading results now. This is one of my main topics for further development.

 

In the last months I´ve had a lot of changes in parameters. In every week I learned something new from algorithmic trader and so I could better understand the single parameters and have to change it. Because of this I haven´t enough data for a significant statistical analysis for comparing. The trades were profitable but this needs sometimes several weeks and in August I have got a small loss. Futhermore I´m looking on winner/looser rate, number of trades, profit factor and max. Drawdown and so on.

 

I haven´t rules for max. variance. But I think this is very important. I need more practical experience for finding the right rules. If it´s always a 80% confirmation of the single numbers in livetrading for example I can calculate with it. In the moment I don´t know how exact the livetrading results vs. the backtesting results are. So I think in the next step it´s very important to collect enough data to find the right answers for this. 

 

And last but not least for me it´s very important to check the SQ-results at first in a small trading live account with real money. This is the last filter for my strategies. If they are not good enough I will don´t trade it on the normal account. In the moment all of my SQ strategies are in this step not final live.

 

Sorry for my bad english. I don´t use the language very often.  😀

 

Tradine

0

stearno

Customer, bbp_participant, community, 379 replies.

Visit profile

11 years ago #132933

The language was just fine. 😉 I too am figuring it out at this point. Thanks for sharing!

Sent from my HUAWEI MT7-TL10 using Tapatalk

0

geektrader

Customer, bbp_participant, community, 503 replies.

Visit profile

11 years ago #132946

Just to clarify:

 

most strategies in SQ look better with keeping trades open during the weekend. BUT, just like MT4 backtests, SQ does not consider if there is a gap, it still assumes your StopLoss / TakeProfit was filled, even though that price never existed during the weekend. SQ and MT4 handle it this way: say the market closes on Friday at 1.1320 and you have a EURUSD short that was opened at 1.1310 with a StopLoss of 1.1340. Now market opens on Sunday at 1.1410, SQ and MT4 in their backtesting enginees do still asume you got your order closed at the SL of 1.1340, which of course is impossible during a weekend gap. So while in the SQ / MT4 backtest your order had a loss of 30 pips, you have suffered a loss of 100 pips in reality! That ALSO happens if you use real tick data, SQ always assumes your SL was filled at exactly what you specified.

 

Another problem with the weekend exposure is that the spread widens at market close with any serious broker (ECN´s) as liquidity begins to disappear (of course). This means your StopLoss might just get triggered by the widened spread right before market close.

 

For this reason I generate all my strategies without any weekend-exposure, because the strategies that SQ generates with weekend exposure, are simply just wrong and it depends on a lot of luck that they really work like this because the weekend gap is small or there was no gap at all. Just imagine if there would be a 200 pips gap during a weekend (and we had that already!)… no thanks! 🙂


🚀 Unlock Your Edge in Automated Forex Strategy Development 🚀

Historical Forex Data Starting From 1987, 28 Pairs, M1, 99% Error-Free, Lifetime Free Updates

0

stearno

Customer, bbp_participant, community, 379 replies.

Visit profile

11 years ago #132959

Thanks for sharing. I did not think of those aspects.

Sent from my HUAWEI MT7-TL10 using Tapatalk

0

Viewing 12 replies - 1 through 12 (of 12 total)