The Day Of Week effect, does it still work

Do you believe in day of week bias in the S&P500 ?

  • No it's a myth

    Votes: 42 61.8%
  • Yes, but only Monday being a up day

    Votes: 4 5.9%
  • Yes but only Monday and Friday are reliable

    Votes: 10 14.7%
  • Totally, I use each days effects in all my S&P500 patterns

    Votes: 12 17.6%

  • Total voters
    68
Murray, I have been testing stupid ideas since before you became famous. Run that test on rolling 90 day periods incremented every 30 days. You'll get wildly divergent answers. Back in 2001 I briefly traded a system I called "Buy Wednesday" (you can look it up here) for as long as it worked after the backtest, which was all of one week.
 
Quote from hypostomus:

Murray, I have been testing stupid ideas since before you became famous. Run that test on rolling 90 day periods incremented every 30 days. You'll get wildly divergent answers. Back in 2001 I briefly traded a system I called "Buy Wednesday" (you can look it up here) for as long as it worked after the backtest, which was all of one week.

Yes, I remember the Wednesday effect , mid week relationshops are not stable. I am going to publish rolling window results in the soon in this thread. I am just getting the topic up and running.
 
I'll be looking forward to it. I abandoned the idea about six years ago. WTFK? Maybe it works again. Of course after you publish it it won't.
 
The monday friday deal is affected by the time of year and how it coincides with government reports, earnings. Also if the monday is the last day of the qtr or the first day of the qtr or month then it is affected.

All mondays are not equal nor are all fridays.

John
 
It is like the fucking idiots who tested Tuesdays long term (back when the Fed announcement was on Tuesdays) and found high volatility then!
 
Quote from jficquette:

The monday friday deal is affected by the time of year and how it coincides with government reports, earnings. Also if the monday is the last day of the qtr or the first day of the qtr or month then it is affected.

All mondays are not equal nor are all fridays.

John

Good point, What you are talking about are things like Day of Week in Month effects. I might also show this analysis but am worried about the attacks from ET people for discussing what they feel is curve fitted nonsense results. I already planned on doing Third Friday analysis since that being a special case is obvious to anyone who trades.
 
FWIW,

I track the daily open/close of the nearest S&P futures contract (8:30am - 4pm).

I crunched my spreadsheet (April 11, 2006 to yesterday) and these are my results for Day of Week in S&P points:

Monday average: 0.88 median: 0.85
Tuesday average: 0.8 median: -0.84
Wednesday average: 0.46 median: 2.10
Thursday average: 1.15 median: 0.55
Friday average: -0.19 median: 0.15
 
Market's VERY Basic Pattern

Weekly Trend

In an up market: Mondays and Tuesdays are UP and corrections are WED into Thursday morning with Friday closing at the high of the week.

Basically strong Mondays and Fridays, and weakness mid week.


Bear Pattern:

Just the opposite-down Mondays and down Fridays and strength mid week.

Friday weekly close is usually the high or low of the trend.

Intra-day is smaller fractal of weekly.

Bullish days an up opening, mid day pull back and close at the high.

Down days-down opening, short squeeze till midday, then down into the close.

Hourly chart each morning will confirm likely days pattern.

Down and dirty quick calculation: 61% directional Stable price=Square root(high*low)

rules:If tomorrow's opening price is above todays stable price ,sell at market.

If tomorrow's opening price is below todays stable price ,buy at market.
 
These are the exact rules I tested as follows:

If we just simply buy at the open on monday in S&P500 and exit on the close your average trade is $114.72. This was over the period 1982 to date. Remember my calculation was done at $250.00 a point. When Joe tested it a point was worth $500.00 , which would have been well over $200.00 a trade in old point value dollars. If we look at the results from 4/22/82 to 6/1/93 we average $116.06 at $250.00 a point. So the results have been statistically the same for the 14 years since Joe published this system.
 
Back
Top