Please do continue! If you experts don’t like what you see or read go to another thread or go count your money!
LOL now with the ignore function. trolls = freebumps.
Hey MeiHua… Have you ever done any statistics on weekend gaps in your quest to find quantifiable edges? I’m thinking since more unforeseen bad things (natural disasters, un predictable political turmoil etc) would seemingly happen on the weekends then would good things… An edge could exist with Usd or JPY denominated pairs where one would sell before Friday close, and see if something bad happens over the weekend that would cause safe haven flow and a bearish gap on the pairs… Just throwing it out their if you got bored
I have studied gaps in instruments that gap more frequently. SP500, crude, gold, etc. I can do a study on it but the sample size is will be small so we have a larger standard error. But I can try it and see what I can come up with anything interesting.
[QUOTE=“MeiHua;485591”]
I have studied gaps in instruments that gap more frequently. SP500, crude, gold, etc. I can do a study on it but the sample size is will be small so we have a larger standard error. But I can try it and see what I can come up with anything interesting.[/QUOTE]
Sweet… How far back do you have data for?
I guess maybe you already use ProRealTime or similar, since you trade full time, but if you don’t, I really recommend ProRealTime. It’s a great product and you get high quality data for years back.
(Not trying to plug anything! Just a tip since it’s something I find very useful myself)
usually between 7-10 years.
Yes I have my own pro level data feed, but thanks for the recommendation.
I am a big fan of seasonal patterns. Both long term cyclical patterns measuring months or years to the more frequent patterns of days. I had touched upon that in a different thread, which i may repeat here because that thread has fallen off, but it will also give me a chance to re update that work. But here is something that hits close to home. This is the 5 day weekday patterns and percentages. it is very simple monday is always and ‘S’ which means SAME, ‘D’ means different. I know this is my own notation but just follow me here. I am using EUR/USD data going back to 2001 to present day, daily data. You can see in the list form how many samples of each type of week was taken.
Explanatory examples:
If monday is an up day, all up days for the week become ‘S’, if monday is a down day all down days become an ‘S’. Monday is the KEY DAY.
So all patterns start with S
SDSSD means that tuesday closed opposite to monday, wednesday and thursday closed in the same direction, and friday closed in a different direction. The up down is not relevant, it is a RELATIVE measure.
This is the same information just displayed 2 ways. One is a tree style map where you can trace the day by day. one is a day list. Just depends on what you like to look at.
How do you use this? How does this help?
Alright I have highlighted specific day patterns that have a larger probability to occur. lets take the wednesday example of 56%
I am going to walk through how this may occur and why it may be important.
Monday is a down day, than tuesday retraces an is an up day. So basically this week has been chop. What do we do for wednesday?
Well this is where the pattern comes out, so far the weeks pattern has been SD, so we need to look for SD?. Well if this pattern of SD occurs there is a 56% chance that wednesday is going to be the same type of direction as monday and create a new S.
Lets walk through the logic of these in price, We just had a move in 1 direction then a retrace. Where are we probably going to go? continue to move inside this new weekly range or breakout? Thats basically what this is saying.
lets do the multiples of each day to have the week have a total trend so all days go in the same direction. that would be an SSSSS week. (.48*.49*.43*.55 = .056) So 5.6% for all days to work in the same direction, so not very often. I know thinking about these patterns arbitrarily with letters seems strange but if you think about the price patterns that it actually relates to it makes more sense.
So lets go back and continue with the SDS example. Lets look at the thursday, we just had an SDS week. so thursday is only going to be SDSS or SDSD. So look its about a 50/50 shot. now think about the week just happened, we just have been chopping around. Going up and down and up and then what? the market in an indecision mode. Which is also shown with the statistics here, there is no viable directional markers.
But there are some patterns that occur with good frequency. SDSS becoming SDSSS is 63% of the time, which is a good edge in anyones book. So your going to be able to have a statistical bias. Again this IS NOT AN ENTRY. but a way to use statistical measure to help form a bias on market behavior.
By using the larger time frame, repeating patterns. Can allow us to get insight into how frequently they may repeat. As always markets must be traded forward but only studied backward. By adding these statistical measures into your toolbox, then compound that with your traditional techniques. For example trend lines, HHs HLs, MAs or whatever tool it might be. Will allow you to have numbers based confluence to help boost your edge.
Really interesting stats. The issue with statistics is that it’s one dimensional in the forex world. Add risk management to the mix, that makes it two dimensional, psychology makes it three dimensional. So by this, I mean if you looked at your technique and looked for SDSS becoming SDSSS, you might find on Friday, you have a non farms report and you have a 300 pip range that takes out your stop. Or you panic when you see you are 100 pips down, close your position and things flip.
In saying that, really enjoying this thread, and the insights you show.
Risk management and psychology are independent of this data. That is your ability to execute on an edge, the onus is on the trader. The probabilities are unaffected by those additional 2 dimensions you raise. Why would statistical measures of patterns be invalidated by a personal lack of skill? Just because I can’t pilot a plane when I sit in the pilot’s seat, doesn’t make the plane any less a plane on its own. Based on my inherent lack of ability to use it to its potential as a flying machine.
The last day will always be a friday, so yes NFP, and all major reports are factored into these statistics. All news spikes are accounted for, actually every trading day that has occurred in the past 13 years has been factored in. If you really think about the diverse markets, from 2001 until today. Its gone through fundamental shifts, large economic cycles, and many different volatility phases. Actually the euro didn’t really count as an accounting money before then I believe. Before then it was a ratio of the different currencies heavily weighted deutschmark. So really I have done it over the euro’s entire lifetime, and every large or small scale change in environment, economic climate, or volatility spike has been taking into account.
On another note, do you have any observations regarding volume (real volume as in futures, not tick volume) and edges?
Absolutely, however because no one here trades futures I havent share them. There are a lot more tools out there for people who have real volume, order books, time and sales, delta. Which is why i prefer to trade futures. imho real volume is a key element to trading, both discretionarily and algorithmically.
Could not agree more. Which is why I’m moving from spot to futures. I almost feel stupid for taking so long to figure out the major advantage that futures have.
I think that it would probably be advantageous for “retail” traders, like the ones on Babypips, to learn a bit about volume for instance. It’s the third dimension that makes everything come together imho.
My trading has improved noticeably since I decided to get a real data feed.
I’ve been wanting to switch to futures. Problem tho. It takes more capital. What about opening a futures account and analyzing it but trading spot? Meihua of still like to see the data on futures volume :-D.
Out of interest, do you use Bloomberg or another provider for your data feeds?
problem with that is (and I could be wrong) most futures brokers delay there price feeds in demo. Also the price feed itself will generally be close but off a little.
If trading higher timeframes it could work I guess as a small delay in feed wont really hurt you to much
Yea I mostly just look at daily
I’ve been using ProRealTime, but I’m in the process of switching to Saxo Bank now where I will get everything in one place.
Yes, they usually delay ten minutes or so.
To me that hasn’t mattered since I trade daily charts. But I still thought it was worth the extra money to get live intraday data through PRT. Since Saxo is a futures broker I will now get everything in one place.
Then you should be fine. Not sure of the broker you are thinking of looking into. I used to use think or swim and they were pretty good. I am sure there are better out the. There demo has a 10 minute delay but is free forever as long as you use it like once every 30 days or something like that.