Market Statistics - Finding a statistical Edge in your market

[CanOz]

I'm keen to know some basic facts about a market that i might trade. I've had some time so carrying on from the Inter'l thread, here is some data on the FESX regarding when the session high / low is made. This may surprise a few people....

It is during the opening few minutes of the pre-cash open that the session high or low is made, obviously on a gap up or down. You can see the shift in daylight savings on the chart as well.

There is also another concentration of highs and lows around the US open, likely a bit spread out for news.

 

[skc]

I'm keen to know some basic facts about a market that i might trade. I've had some time so carrying on from the Inter'l thread, here is some data on the FESX regarding when the session high / low is made. This may surprise a few people....

It is during the opening few minutes of the pre-cash open that the session high or low is made, obviously on a gap up or down. You can see the shift in daylight savings on the chart as well.

There is also another concentration of highs and lows around the US open, likely a bit spread out for news.

May I make a few suggestions to your chart.

1. Shift the data with daylight saving by an hour. So your X-axis should become hours from cash pre-open.

2. Plot in 5 or 10 minute blocks. I can't tell what your interval is but 5 minute blocks should provide a clearer chart.

3. Plot vertical axis in percentage of occurrences. Whilst there is a concentration of highs/lows being made in the first few minutes of open, it's hard to tell whether it's 20% or 50% of the time.

4. Plot same data by day of week and see if there are any major differences.

 

[pixel]

Thanks, CanOz, for sharing that study.
Somebody spent a lot of effort to compile and chart that data. Would be interesting to see a similar graph for ordinary stocks - individual ones as well as groups or entire Exchanges.

Especially for the small, speccie end of the ASX, I get the impression something similar may occur. And it's easy to explain too: Usually, "surprise" announcements are published before Open, and then it's the eager beavers, aka "Amateurs", that open the trading day full of exuberance or despair. And then things settle down and price drifts back towards a mean.

Have others observes the same?

 

[CanOz]

SKC, will try and do some more work tomorrow or give it to an Excel wizard that can manipulate some more.

In the meantime here is a webinar that explains the point of using market statisitcs, i think this is quite relevant and may help some understand what we're trying to do.

Any excel wizards, if you would like to help out, i can provide reams of intra-day and EOD data.

If some has ideas, i can also get an Amibroker coder to help on the EOD data.

 

[McLovin]

Interesting chart, can. Thanks for posting.

Thanks, CanOz, for sharing that study.
Somebody spent a lot of effort to compile and chart that data. Would be interesting to see a similar graph for ordinary stocks - individual ones as well as groups or entire Exchanges.

Especially for the small, speccie end of the ASX, I get the impression something similar may occur. And it's easy to explain too: Usually, "surprise" announcements are published before Open, and then it's the eager beavers, aka "Amateurs", that open the trading day full of exuberance or despair. And then things settle down and price drifts back towards a mean.

Have others observes the same?

I have observed similar. I've also observed, but have in now way tested this that they run hard out of the blocks then retrace then slowly rise again. At the speccie end announcements are usually poorly interpreted (as an example in biotechs regulatory approval is a really misunderstood concept), and companies often use announcements as a marketing exercise not having much to say, but putting in a few positive phrases that keep the sp ticking along. I think in smaller less well followed (researched) companies, the amateurs are often trading with each other and that keeps things rolling along (fomo) much longer than it does in larger mc companies. I'm pretty sure you could build a nice little living by buying stocks with the most threads/posts on the other forum.

 

[kid hustlr]

Can,

Any chance you want to also do this for the FTSE

Working off the assumption the FESX has follows the FTSE quite closely, I 'm surprised there's not more of a 'U' shape, I wouldn't expect many new highs and lows in the junk period after the first say 2-3 hours.

You should split this into highs and lows as well imo

 

[CanOz]

Can,

Any chance you want to also do this for the FTSE

Working off the assumption the FESX has follows the FTSE quite closely, I 'm surprised there's not more of a 'U' shape, I wouldn't expect many new highs and lows in the junk period after the first say 2-3 hours.

You should split this into highs and lows as well imo

We'll it actually is spilt, but as skc suggested some work on the interval needs to be done to allow that visibility....

 

[Roller_1]

Plot the high above and lows below the axis

 

[fiftyeight]

CanOz gave me some hourly data for the SPI for the last 3 years, including ETH.

This shows how many times the weekly high or low occurred on a particular day.

Next step is to find out if changing the start day of the week has any significance.

We can then look at if the market are trending or range bound shows anything interesting.

Hopefully look at some other markets as well

 

[CanOz]

Plot the high above and lows below the axis

Yeah that's a good idea....

 

[Wysiwyg]

If statistics prove that price moves greater than a 50/50 bet, i.e. 51/49 or better, does this make it an edge? If there is another metric other than ratio/percentage (edge?) then please enlighten.

 

[DaveDaGr8]

It's win ratio * win percentage.
given your 51/49 split

IF your average win % was 10% and average loss % is 12% and your trade is $1000

51 times you make $100 $5100
and 49 times you lose $120 $5880

You would have a losing system.

 

[CanOz]

Watch the video....

 

[DaveDaGr8]

Hey Canoz,

I'd be happy to have a look. I can do excel and matlab if i need to.

 

[Wysiwyg]

It's win ratio * win percentage.
given your 51/49 split

IF your average win % was 10% and average loss % is 12% and your trade is $1000

51 times you make $100 $5100
and 49 times you lose $120 $5880

You would have a losing system.

That is a position sizing and trade management scenario and I suppose, over a number of trades, would be a statistical result. Not sure if that is the aim of the discussion.

 

[Wysiwyg]

Watch the video....

"It's because all those results that you're being given are of a statistical nature and they do not describe what might happen to a trade when it (ahh) actually goes off." 52.45 minutes onward. Big Mike.

 

[CanOz]

"It's because all those results that you're being given are of a statistical nature and they do not describe what might happen to a trade when it (ahh) actually goes off." 52.45 minutes onward. Big Mike.

That wasn't big Mike....that was morad speaking. What do you want wysiwyg? A silver bullet?

 

[Wysiwyg]

That wasn't big Mike....that was morad speaking. What do you want wysiwyg? A silver bullet?

I have gone deeper and deeper into the rabbit hole and found nothing more than one thing. Following the big dollars. Get in when they're in, get out when they get out. It is only the weight of money that move prices no matter how under valued or how crazy a valuation. is.

 

[cynic]

That wasn't big Mike....that was morad speaking. What do you want wysiwyg? A silver bullet?

A lead one will do! We all know how to transmute that stuff into gold, don't we?

 

[cynic]

And even if we don't, lead bullets certainly have their uses!