I have a question for the counters (and Dr. Z):

When the S&P makes a new, all-time high, what has been the historical reality of additional closing higher highs in the subsequent 1, 5, 10 days? The theoretical answer is obvious. But what is the historical reality?

My gut says that there will be more additional closing highs than the theory predicts. But what is the real answer?

Kim Zussman writes:

The attached plots waiting time (in trading days) between new all time highs in SP500 (1960-present).

Obviously you're more likely to get a new ATH tomorrow if there was one yesterday as opposed to 100 days ago.

Over this same period new ATHs were followed by another ATH the next day 56% of the time. This is slightly higher than drift, as over the period up days occurred on 53% of all trading days.

Victor Niederhoffer adds: 

Perhaps this will add some stats to Rocky's question. Since 1996 there has been a remarkable consistency to the distribution of moves after big highs. For example, after 1000 day highs, there were 172 of them. The expectation the next day was - 1/2 an S&P point. Or about minus 1/10 of 1%. The average duration to the next 999 day high was 2.3 days. The expectation 3 days later was about -2 points or minus 1/3 of 1%. On average, considering that you were not in a 1000 day high, the average duration to the next 1000 day high was 58 days. The expectation was 2/10 of 1 big point the next day or 2/10 of 1% a day. The expectations for 100, 200, and 500 day highs are remarkably similar, i.e. random to slightly, slightly, insignificantly, meaninglessly, negative. All this is based on continuously adjusted futures data so that the actual changes, not meaningless adjusted changes, were used.

Kora Reddy writes:

I am doing the below study on dividend adjusted $SPY ETF closing prices (regarding continuous future contract prices and I need an opinion on how to avoid what the chair says "it's better not to roll". Do we trade the next month expiry futures while taking the signal on current month futures on the expiry days…) but if after closing on an ATH, and then if $SPY presents a dip after 10 trading days, at close, the expectation from t+10 th day to next 10 trading days (non-interleaving trades).

#    24
% wins   63%
avg     1.05
med    1.21
avg win 2.99
avg loss       -2.18
max loss      -5.41
stdevp        3.29
t-test  1.57

For the dip presented after 20 trading days after $SPY closes at ATH, expectation for the next 20 trading days (non-interleaving trades).

#  15
% wins   80%
avg % 2.54  ( vs 0.8 % for any 20 trading days , or 1.26% barring the trades in 2013 year  )
med % 3.34
avg win %      4.03
avg loss %        -3.41
max loss % -4.96
stdevp  3.61
t-test 2.72

below the 15 prior trades since Jan 2000 ->
Date    t+20    t+20 %
01-May-14    ??    ??
27-Mar-14    1.71    0.93
24-Jan-14    6    3.37
12-Dec-13    4.5    2.55
15-Oct-13    7.19    4.28
15-Aug-13    2.9    1.77
05-Jun-13    0.85    0.54
25-Feb-13    6.48    4.46
10-Oct-12    -4.96    -3.57
02-Nov-07    -3.06    -2.32
09-Aug-07    0.59    0.47
07-Jun-07    3.92    3.05
02-Mar-07    3.34    2.8
05-Jan-07    3.71    3.07
14-Apr-00    7.12    6.82
14-Feb-00    -2.2    -2.06

that 1-May-14 trade is the result of $SPY closing at ATH on 2-Apr-14 and 20 trading days later $SPY closing lower, as on 1-May-14 ( lost -0.29%).

ps: no major edge for the dips presented after 1/2/3/4/5 trading days after $SPY closing at an ATH.

Victor Niederhoffer replies:

Just adjust algebraically. Or do it the best way. Maintain the original prices for percentage calculations, and work with algebraically adjusted changes for all other price changes and independent variable calculation. Fortunately, the market is evil, and likes to take the same amount from the poor lower feeders all the time. So it moves 5 or 10 bucks with impunity whether its 850 or 1850 with the same frequency. 


WordPress database error: [Table './dailyspeculations_com_@002d_dailywordpress/wp_comments' is marked as crashed and last (automatic?) repair failed]
SELECT * FROM wp_comments WHERE comment_post_ID = '9342' AND comment_approved = '1' ORDER BY comment_date




Speak your mind


Resources & Links