The Day Of Week effect, does it still work

[dtrader98]

So today was a horrific Tuesday. Pretty much matches what we would expect from the worst day of the week stats for the year.

Now, tomorrow is a Wednesday. How does that day fare statistically? This is a good time to think about of statistics as a filter. Does that mean that it is guaranteed that tomorrow will be up?
Nothing is guaranteed. Also, you can't just expect every single instance to follow the average expectation, that's why it's statistics (law of large numbers approach expectation). But the statistics do show that Wednesday has a higher likelihood of closing up more often than not.
Just as it showed us Tuesday (today) had a likelihood of having the worst down swings.
No matter what our emotions show us at this moment, the statistics present historical facts and clues about events.


<img src="http://elitetrader.com/vb/attachment.php?s=&postid=1584325" border="0" alt=""><br /></font></p></font></p></font></p></font></p></font></p></font></p>

 

[ForexGuru]

Quote from dtrader98:

But the statistics do show that Wednesday has a higher likelihood of closing up more often than not.

Not enough info. Example below shows 3 "up" Wednesdays to 1 "down" Wednesday.

• S&P 500 +10
• S&P 500 +15
• S&P 500 + 05
• S&P 500 - 30

NOT GOOD even though up days out numbered down days 3 to 1.

 

[dtrader98]

Quote from ForexGuru:

Not enough info. Example below shows 3 "up" Wednesdays to 1 "down" Wednesday.

• S&P 500 +10
• S&P 500 +15
• S&P 500 + 05
• S&P 500 - 30

NOT GOOD even though up days out numbered down days 3 to 1.

I think you are still misunderstanding the point of statistics (and this example in particular). You are showing one sequence of data and expecting the earlier study to predict the specific data point that follows your sequence. The example shown was an example of averages not conditional sequences.
The sequence you show is meaningless by itself. It, however, in no way refutes the fact that, for the year of 2007 to date, on average the open to close on Wednesday was up more often than not, and that the cumulative compounded sum to date exceeded all other days. You want larger numbers for confidence? Run the data back to the beginning of available market data and see how it compares to this interval.

I think what you are looking for is a study of conditional probabilities on a sequence of events. In your case, what is the likelyhood given, uuud sequence, that next event is u or d?
I haven't run the stats, but reversion to the mean would tell me that after a run of three up sample days in a row, I would expect a down day to follow more likely than an up day, but I digress. You can run the actual stats on this, and perhaps it will add more granularity to the study.

Please post the results when your done. I'm sure everyone can benefit from seeing the actual study.

 

[NihabaAshi]

Quote from dtrader98:

...Now, tomorrow is a Wednesday. How does that day fare statistically? This is a good time to think about of statistics as a filter. Does that mean that it is guaranteed that tomorrow will be up?

Nothing is guaranteed. Also, you can't just expect every single instance to follow the average expectation, that's why it's statistics (law of large numbers approach expectation). But the statistics do show that Wednesday has a higher likelihood of closing up more often than not...

Strongly agree.

That's why its important for the trader to stick to his/her strategy and apply it within the market seasonal pattern as the link I posted earlier involved in a different tendency discussed here at ET.

Therefore, tomorrow if someone gets a bullish signal via whatever strategy they've been using prior to knowing anything about Wednesday being an up day...

Take the trade.

Simply, don't change your entry signal but you can change your trade management after entry to exploit the market seasonal pattern that Wednesday is an up day.

Some tendencies aren't meant to be traded as is and requires us to apply (merge) our already existing trade signal into the tendency.

Mark

 

[ForexGuru]

Quote from dtrader98:

Please post the results when your done. I'm sure everyone can benefit from seeing the actual study.

The best results would be to paper trade for one year.

You say Wednesdays are mostly up days. Well every Wednesday why don't you buy the S&P 500 when the market opens and sell at the end of the day. Post the trades live on ET and we then can crunch the numbers in September 2008. You could start a new ET thread for this purpose only and request no replies until the year is up to keep it uncluttered.

 

[NihabaAshi]

Quote from ForexGuru:

The best results would be to paper trade for one year.

You say Wednesdays are mostly up days. Well every Wednesday why don't you buy the S&P 500 when the market opens and sell at the end of the day. Post the trades live on ET and we then can crunch the numbers in September 2008. You could start a new ET thread for this purpose only and request no replies until the year is up to keep it uncluttered.

Although your post was to dtrader98...

So far my research shows it's not practical or not realistic to buy the open and sell the close on Wednesdays due to the fact that most traders cannot properly manage the initial stop/loss protection nor trailing stops.

Simply, lots of wiggles in between that will pick pocket most stops...at least it would pick pocket my stops had I tried to trade the tendency as is.

There's a well documented journal here at ET a few years ago by someone that used such a method...buy the open and sell the close...resulting in losses instead of profits had he properly manage the trade during the trading day.

That reason alone is why I've mentioned several times that some tendencies applied as is is not practical.

Those particular tendencies (including the Wednesday up day) requires a merge of your trade signal with the tendency to make it practical.

For example and something you can't ignore...

Take a look at the Emini ES last Wednesday August 22nd.

It's an up day that open @ 1461.25 and close @ 1468.25

That's +7 points via the market seasonal pattern tendency.

However, it should be easy to see that the intraday low was 1454.50

The difference between the Open and Low is -6.75

How many day traders do you know are going to sit there and bleed -6.75 just so they can get +7 points???

Very few.

Most would bail for a loss around -2 points.

However, merging my method with the tendency I have 5 Long signals.

3 winners, 1 breakeven and 1 loser.

Without going into any specific details...

The last signal on that day was 2:39pm est @ 1458.75 and holding to the close of 1468.25 results in +9.5 points just on that trade by itself.

Anyways, the total profits on Long positions per contract was +2.25, 0.00, +4.75, -2.00 and +9.50.

That's +14.50 via merging my trade method with the market seasonal pattern tendency.

So far, after doing analysis on the past 5 years...

Wednesday's are extremely profitable if I change my trade managment on Wednesday to exploit the tendency.

Further, so far all I've done is kepted 1 contract Long versus 1 contract Short.

I plan to do another set of stats via inputing how I use position size managment (Longs size > Short size) to exploit the market seasonal pattern tendency.

More importantly, I will do my analysis on the Russell Emini ER2 because that's what I currently trade.

By the way, in case you ask me, I already been posting my trades live since 2000 on IRC but not at ET (many at ET know such is fact).

You can also view my brokers profit/loss statement.

All I ask is that you do the same.

In addition, I'm not challenging you because I think its just silly to challenge someone about something like this when all you need to do is look at the statistical facts about Wednesday's.

You either believe what you see or don't.

I believe it and that merits my time to review my trade history to do a comparison with how I traded on the past Wednesday's in comparison to trading this tendency.

So far I like what I see after going back 5 years of personal trade history on Wednesday's.

Mark

 

[ForexGuru]

Quote from NihabaAshi:

So far my research shows it's not practical or not realistic to buy the open and sell the close on Wednesdays due to the fact that most traders cannot properly manage the initial stop/loss protection nor trailing stops.

You don't bother with stops, you BUY at 9:30am, SELL at 4pm every Wednesday. Very simple.

PS....I only skimmed through your long winded post, the Day of Week strategy is very basic and doesn't need a long post as yours.

 

[dtrader98]

Quote from ForexGuru:

The best results would be to paper trade for one year.

You say Wednesdays are mostly up days. Well every Wednesday why don't you buy the S&P 500 when the market opens and sell at the end of the day. Post the trades live on ET and we then can crunch the numbers in September 2008. You could start a new ET thread for this purpose only and request no replies until the year is up to keep it uncluttered.

That sounds like a great idea. I'll let you do the thread, since
1) I don't want to hog all the work.
2) I think if you really have doubts about the validity of this test, then, you should convince yourself first. Once you have some kind of hard evidence, not just intuition, then share that evidence on the board to be equally critiqued by others. And by the way, if tomorrow is a down day, that in no way undermines the statistical evidence nor conclusions arrived at in the first thread I presented. It does not say every single Wednesday will be an up day.

If you want to refute the data, by all means, show some kind of study to counter it. For all anyone knows, I fudged the data or made mistakes (heaven knows ET isn't paying my secretary enough=) I trust my work, but no one else has to, that's why I make it easy enough for others to replicate. I lay these studies out in the open, so that others can critique/verify them and hopefully everyone can benefit. I appreciate any or all feedback that's constructive.

The beauty of this particular study, is that as NihabaAshi mentioned, unlike a million other TA claims, it is not subjective. The statistical facts are as they are. I only ran a year, but you can easily run it farther to corroborate/contest the conclusions.

By next year, the results may diverge wildly -- who knows? But I'll take my chances on objective statistics over subjective impulses any day.

 

[NihabaAshi]

Quote from ForexGuru:

You don't bother with stops, you BUY at 9:30am, SELL at 4pm every Wednesday. Very simple.

PS....I only skimmed through your long winded post, the Day of Week strategy is very basic and doesn't need a long post as yours.

I've been studying/trading market seasonal patterns (tendencies) since the 80's.

My research so far shows that the Wedenesday Up Day is one of those tendencies that requires another strategy to make the tendency usable for real trading conditions.

True and I do agree with you in that the Wednesday up day is very basic (very simple) but that doesn't imply it should be applied as is.

It's your (my) responsibility to find a practical application with the tendency.

Think outside the box.

Once again, you can trade the tendency as is if you want...that's your choice and statistically you should be able to make some money if you trade all the Wednesday's.

However, we all know that just because statistical facts supports such or that something backtested shows a positive expectancy...

Real trading is a completely different story.

Simply, past results is no guarantee for future results.

However, if you have no intention of researching nor applying the tendency...

Do not tell others how to apply it especially when others have already researched the tendency and discovered it should not be applied as is.

Specifically, my research shows that its much more profitable to not apply the tendency as is and that has more merits with me along with telling me not to take your suggestion seriously about not using stops.

Therefore, proper trade managment can be an edge in many different strategies including this Wednesday Up Day pattern tendency nor does it need to be subjective unless you want it to be.

P.S. I shorten this message so that you don't have to spend an extra 15 seconds in reading it because it sounds like your a trader that prefers to take shortcuts instead of reading all the facts to completely/accurately understand the book.

Mark

 

[dtrader98]

One more sample gets added to the population.

At this pt. I'm confident enough to favor an up wednesday for today

<img src="http://elitetrader.com/vb/attachment.php?s=&postid=1585866" border="0" alt=""><br /></font></p></font></p></font></p></font></p></font></p></font></p></font></p>