High Low Stats by 30 Minute Periods

[dbntina 10]

All,

 

I wrote some code in tradestation and outputted it to Excel.

 

Using 5 minute data on the ES.D I basically calculated the stats that a high or low or either was created in the different time periods.

 

I kept it simple using periods:

 

A is 9:30 to 10

B is 10:01 to 10:30

and so on to

N is 4:01 to 4:15

 

Hope some of you find this helpful.

 

:thumbs up:

ES High Low Stats.xlsFetching info...

[Guest forsearch]

Why not keep it simple and use 30 minute data on the ES.D (not 5)?

[Frank 10]

hi dbntina,

 

lets discuss.

 

you can see how the data is skewed such that a key high or low is made in the first 1-2 hours. that is good information as you will not be fighting the key 'bigger-picture' movement for the day if you keep this in mind.

 

but the other question is how often does the market extend such that it chooses a direction and continues in that direction for the rest of the day.

 

for example, can you run the data with a filter in it?

 

ie, if the highest bar for the day came before the lowest bar for the day (high made first), what are the statistics on when the FINAL low will come? which period is the market likely to make its final low?

 

the second case is then, if the lowest bar for the day comes before the highest bar of the day (low made first), what are the statistics on when the FINAL high will come?

 

I have tracked this for the past year and curious if my data lines up with last 5 years. My data shows that there is about a 2 in 3 chance of the FINAL high or low not occuring until after 2:30pm, with actually many occurences not until after 3:30pm.

[cosmic 10]

thank you dbntina, do you define the IB for the ES as the first hour of day session trading?

 

thanks!

cosmic

[dbntina 10]

Yes I do comic

[dbntina 10]

forsearch, I sure could have used 30M bars and saved some time. It's funny how I can miss something so simple :doh:

[dbntina 10]

Frank,

 

I think that is easy for me to run. I will do that and post.

 

Let me make sure I understand:

 

Show a table just like the others that basically shows the period that the opposite high or low forms after the first high or low?

 

I can run that and post if that is what you mean.

 

dbntina

[Frank 10]

right, look back at the day in retrospect and see if 'highestbar>lowestbar' (> means 'after' to Tradestation) or lowestbar>highestbar....

so:

if highestbar>lowestbar, then which 30-min bar was the highestbar made?

 

if lowestbar>highestbar then which 30-min bar was the lowestbar made?

 

frank

[dbntina 10]

Okay Frank here it is:

 

Same spreadsheet but I have added a 2nd spreadsheet to find out what periods the following corresponding high/low was made if the other was formed first.

 

Enjoy

 

ES High Low Stats.xlsFetching info...

[Guest forsearch]

  dbntina said:

Okay Frank here it is:

 

Same spreadsheet but I have added a 2nd spreadsheet to find out what periods the following corresponding high/low was made if the other was formed first.

 

Enjoy

 

 

Nice update. You might want to add a column to show the cumulative stats, same as in the first one.

 

-fs

[dbntina 10]

Frank,

 

I had it in there initially but deleted it out because it didn't make sense. You would actually probably want to cumulate from the last period forward. It might be confusing to some so I basically left it out.

 

Take it easy,

 

dbntina

[Frank 10]

1 other stat I keep is the final 'high to low' range for the day --- but kept as a ratio vs the range of the first 60-mins. ie, how often is the final range more than 2.0x the range of the initital 60-mins. this 2.0x guideline can sometimes help give a bit of a guideline to help 'see' the structure of the day.

 

dbntina, can you run those stats if you get a chance?

 

total range / initial balance range...

[dbntina 10]

Frank,

 

Yes I think that will be easy and quick. I will try to do that tonite.

 

You are asking basically what % of the time for each day of the week and all days that the high to low range is >= to twice the IB?

 

That should be pretty simple if that is what you are asking.

 

dbntina

[Frank 10]

yes.

 

if its easy, check out what is the actual ratio per day of the week too... ie, 1.89 for Wednesdays vs 1.67 for mondays?

 

thanks dbntina, I am hack programmer -- I can make indicators that do these things and then count them but I don't know how to do statistical research studies like this.

 

frank

[dbntina 10]

Frank,

 

Here it is updated. Check the third tab for what you wanted.

 

dbntina

ES High Low Stats.xlsFetching info...

[Frank 10]

hi dbntina,

 

I have been referencing these stats a lot and they are great.

 

I was wondering if you would be willing to run the same code on an ETF to see if its statistical characteristics are consistent with S&P's.

 

The security is 'EFA' -- which is an extremely popular ETF that tracks all the major international markets. Its very liquid and is interesting because its a foreign ETF and was curious how often foreign indices continue to trend on intraday basis relative to the S&P's.

 

mostly interested in when the FINAL high or low is made (whichever comes LAST)...

 

thanks for any help,

 

frank

[dbntina 10]

Frank,

 

I am glad that is helpful. Unfortunately, I had access to TS because my company had a subcribption. They closed the account and I no longer have access to TS to be able to run any code or anything. Sorry about that. Good trading to everyone,

 

dbntina:shrug: