investor behavior
Forums › ProRealTime English forum › ProOrder support › investor behavior
- This topic has 26 replies, 5 voices, and was last updated 5 years ago by Vonasi.
-
-
01/12/2018 at 7:10 PM #59247
I am concerned with the question of how investor behaviour behaves.
In a simple code below, I have found with the 4H timebase that Monday morning investor behavior is different than Tuesday morning, compare the attachments.
If I set the time base to 4H, the result is much better.
I have the assumption that on certain weekdays and weekdays people have different investor behaviour at certain times.
I’d like to use that in trading.
So I wanted to ask whether someone here, or someone else, has an indicator or is able to program an indicator that, similar to the seasonal patterns from the Pathfinder system, reflects the timebased patterns during the week in investor behavior.P. S. The main code is the “negative”, i. e. the short version of the “DailyOpenLong”, which lives from the RepulseStop in a way that is not generally recognized.
First : Storm with timebase, german time, gmt +1, Dax 1 Euro 4H
Second (next post): Storm without timebase, german time, gmt +1, Dax 1 Euro 4H
Please check it yourself.
4H Short with RepulseStop timebased123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657//-------------------------------------------------------------------------// Maincodce STORM ... ShorT On Rising Markets// 4H Short Repulse Timebased//-------------------------------------------------------------------------//-------------------------------------------------------------------------// Haupttode : 4H Short Repulse//-------------------------------------------------------------------------// Dax 1 Euro Mini// TimeFrame 4H// maincode: 4H Short with Repulse-Stop// created by JohnScher//defparam flatafter = 210000defparam cumulateorders = trueOnce position = 1c1 = cci [21] (close) >-95 // to get out some special situationsc2 = TR (close) >24 // nice conditionc3 = repulse [3] (close) <0.1 // minet1 = dayofweek = 1 and time = 130000 // timebaset2 = dayofweek = 1 and time = 170000 // timebaset3 = dayofweek = 2 and time = 090000 // timebaset4 = dayofweek = 2 and time = 130000 // timebaset5 = dayofweek = 3 and time = 130000 // timebaset6 = dayofweek = 4 and time = 090000 // timebaset7 = dayofweek = 4 and time = 130000 // timebaset8 = dayofweek = 5 and time = 090000 // timebaseTradingTimeShort = t1 or t2 or t3 or t4 or t5 or t6 or t7 or t8IF TradingTimeShort ThenIF c1 and c2 and c3 ThenIF Exponentialaverage [1] (close) < ExponentialAverage [4] (close) THENsellshort position CONTRACTS AT MARKETEndifEndifEndifIf repulse [3] (close) >0.1 thenExitshort at marketEndif// not tested// on other major indizes// with Seasonalpatternmultipler from Pathfinder-Systems// with some Re-invest-Stategies// greets JohnScherTranslated with http://www.DeepL.com/Translator
01/12/2018 at 7:18 PM #59251And here without timebase
4H Short RepulseStop without timebased123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354//-------------------------------------------------------------------------// Maincodce STORM ... ShorT On Rising Markets// 4H Short Repulse Timebased//-------------------------------------------------------------------------//-------------------------------------------------------------------------// Haupttode : 4H Short Repulse//-------------------------------------------------------------------------// Dax 1 Euro Mini// TimeFrame 4H// Haupttode : 4H Short with Repulse-Stop// treated by JohnScher// tested 08.04.2013 - 26.11.17 with PRT 10.3// 2 Points Spread// won 161 loose 212// max DrawDown 853 Euro max RunUp 5.035// winnings total 4,473 Eurodefparam flatafter = 210000defparam cumulateorders = trueOnce position = 1c1 = cci [21] (close) >-95c2 = TR (close) >24c3 = repulse [3] (close) <0.1TradingDayShort = dayofweek = 1 or dayofweek = 2 or dayofweek = 3 or dayofweek = 4 or dayofweek = 5TradingTimeShort = time = 090000 or time = 130000 or time = 170000If TradingDayShort and TradingTimeShort ThenIF c1 and c2 and c3 ThenIF Exponentialaverage [1] (close) < ExponentialAverage [4] (close) THENsellshort position CONTRACTS AT MARKETEndifEndifEndifIf repulse [3] (close) >0.1 thenExitshort at marketEndif// not tested// on other major indizes// with Saisonalpatternmultipler from Pathfinder-Systems// with some Re-invest-Stategies// greets JohnScher01/12/2018 at 8:03 PM #59273Weren’t the … seasonal patterns from the Pathfinder system taken … taken from a website? I cant recall the website name now.
You can put Times in the Optimiser you know?
GraHal
PS Here’s a bit on Time of the Day re EurUSD
01/12/2018 at 8:07 PM #5927401/13/2018 at 12:22 PM #59333Some days of the week are busier than others, some days of the month are busier than others and some hours are busier than others. If you put yourself in the mind of a professional trader then you will see repetitive behaviour.
Monday – ‘Think I’ll just sit back a bit and see what the markets are doing before getting in’.
Friday – Think I’ll just close down my positions for the weekend just in case’
Last Day of Month – ‘Think I’ll cash in so my account looks good and I’ll get a big bonus’
Market Opening – ‘Think I’ll just do a few stop raids and see what I can pick up before lunch’
Market Closing – I don’t want to hold some of this riskier stuff overnight while I’m sleeping so I’ll sell up and get back in tomorrow’.
Lunchtime – ‘Forget the trading I’m hungry so I’m off to the bistro’
Asian Hours – ‘All the big players are asleep’
and so on……
Whether it helps to make profits I’m not sure. I think probability and risk management have more chance on their own – but then there is no point trading a market that isn’t moving because everyone is at lunch I guess.
01/13/2018 at 12:29 PM #5933502/04/2018 at 2:48 PM #61431Hello again.
I wasn’t really interested in the behaviour of the BigBoys and their boredom.
I was more concerned with the fact that investor behaviour on a Monday morning is different than on a Tuesday noon.
Did you notice something like that too?
What I mean is something like this….
Code 1 is intended to effect Monday 090000 Long to Monday 130000.
Long Montag 0900001234567IF OpenDayofweek = 1 and Time = 090000 thenbuy 1 contract at marketEndifIf Time = 130000 Thensell at marketEndifResult see attachment
Code 2 is intended to effect Tuesday 130000 Long to Tuesday 170000
Long Dienstag 1300001234567IF OpenDayofweek = 2 and Time = 130000 thenbuy 1 contract at marketEndifIf Time = 170000 Thensell at marketEndifResult see attachment
Amazing or?
02/04/2018 at 4:44 PM #61443Unfortunately the future does not always pay any attention to the past.
You could easily convert your little test code above to analyse each market just by making the day and time optimized variables and then run a backtest. This would then tell you which have been the most profitable hours and in which days in your sample period. The sample may be too small to be conclusive. The only way to prove if this information has any value at all is to then forward test by running a strategy that just buys or sells in those hours and days and then you can see if the future is willing to copy the past. If it does then this could be used to maybe add an edge to a strategy by only letting it trade the most likely hours on the most likely days to be profitable.
Forward testing is boring as we would all like to get rich today based on what happened yesterday but it is a necessary evil that we must learn to embrace!
02/04/2018 at 5:37 PM #61449Na hi.
“This would then tell you which have been the most profitable hours and in which days in your sample period.”
…
That is the intention.
“The only way to prove if this information has any value at all is to then forward test by running a strategy that just buys or sells in those hours and days and then you can see if the future is willing to copy the past…”
Yes, all our plans here are based on the same principle.
We test what would have been in the past if our program had run
and then we take the program and run it in demo or live as forward testingbut
in the moment we complete the forward test, we were just testing the past.
Regardless of this, I would like to know who can program such an indicator, which shows (x-units) on the basis of past data, has the price risen or dropped between 090000 and 130000 on Monday, if so, what percentage? How did the course from Monday 130000 to Monday 170000, etc. behave?
Of course, one can then generalize such an indicator and base it on each value and each time unit.That´s it what i m looking for.
Kind
02/04/2018 at 5:51 PM #61452Not an indicator but the strategy I suggested is attached as an ITF file. Run it and the Optimize Report will show which have been the best days/hours for long positions. It should be easily adapted for short positions if you wish.
1234567IF OpenDayofweek = D and Time = T thenbuy 1 contract at marketEndifIf OnMarket Thensell at marketEndif1 user thanked author for this post.
02/04/2018 at 6:44 PM #61471Short version. ITF attached.
It would probably make sense to analyse one day at a time (by setting D to the required day) and then chart each days results in Excel to see if there is any sort of rhythm to the results. I think that the % of wins is of more interest than the actual amount won as it only takes one big bar in any particular hour of any particular day to skew the results. Any % of wins over 50% is of interest as 50% can be achieved by a monkey with a pin!
1234567IF OpenDayofweek = D and Time = T thenSellShort 1 contract at marketEndifIf OnMarket ThenExitShort at marketEndif02/04/2018 at 8:25 PM #6147702/04/2018 at 8:47 PM #6148502/04/2018 at 10:48 PM #61497I think that Forex is a probably more of a problem when searching for this sort of investor behaviour as it is more of a 50/50 casino IMHO. Indices I feel would be more likely to have some sort of investor behaviour time based bias. Having said that my attempts at using Excel on the DAX to try to find some sort of pattern have just resulted in an over-whelming feeling of randomness. I fear the data sample of 100k bars is too small to be of any use and even if I tested on more I fear the result would be similar.
I used OpenTime rather than Time to make sorting the order more logical and highlighted the top 25% best results. Chart and Excel book attached.
1 user thanked author for this post.
02/05/2018 at 1:04 AM #61506 -
AuthorPosts
Find exclusive trading pro-tools on