Profiting
from Trading Key Dates -- SPY -- BIDU
My core
goal is to provide quantifiable ideas and strategies that can be easily
implemented and have high probability of success. The main problem with
Seasonal trading is that it is subjective and arguably the edge is gone because
everyone knows about the Seasonal impact anyway. So I have put a study together,
where we can look at the impact and probability of seasonal trading and how we
can mitigate risk, but also have a high probability of a reasonable profit.
The way to
do this is to look at historical data of the stock. In this case we will be
using Matlab® with Yahoo data. The goal is to find trading ideas where history
shows that there is a good risk/reward payoff within a selected date time frame
for an underlying stock.
Looking at
the historical data of the SPY, I can note straight away that there is actually
not much money to be made from “seasonal” date trading. This is based on the
investment criteria and inputs I have used in Matlab®.
I will go
through my study on the SPY, and show how my analysis was made.
SN
|
YD
|
BD
|
SD
|
LR
|
HR
|
DD
|
PP
|
BL
|
SPY
|
23.1704
|
09//10
|
22//11
|
3.132
|
21.1545
|
-12.326
|
60.8696
|
-18.037
|
Outputs:
SN = Stock Name
Years = Years of data
BD = Buy Date
SD = Sell Date
LR = Lowest Return
HR = Highest Return
DD = Draw down
PP = Percentage Profitable
BL = Biggest Loss
SN = Stock Name
Years = Years of data
BD = Buy Date
SD = Sell Date
LR = Lowest Return
HR = Highest Return
DD = Draw down
PP = Percentage Profitable
BL = Biggest Loss
Looking at
the SPY I have analyzed 23 years of data. The input variables were set at a max
of 60 day time frame block within a year, with the lowest return set at 3% with
a min 60% profitable. The output is that the only date range to buy the SPY is
09/10 with a sell date of 22/11 of every year. Over the 23 years that date range
was profitable 60.86% of the time with the lowest return within that date range
of 3.132%. The highest return in that time period was 21% with a drawdown of
12%. The biggest loss was 18%.
What does
all this mean? Well it means that if I buy the stock on the 9/10 with a sell
date of 22/11 it will be profitable 60% of the time. If the trade is profitable,
the min return I can expect is 3%. But the problem lies in the biggest loss, if
the trade is a losing trade; historically the biggest loss was 18%.
Looking at
this output we can clearly see that even though seasonality may be an existing
thing as an overall trend concept, the reality is that the only date to trade where
the min return is above 3% is as noted above. Seasonality does not mean always
profitable based on historical data, that is the problem, and with this data
set it shows that even to get a small 3% return over 60day time frame, the SPY
trader may have to endure a massive drawdown. As an investor I would not be comfortable with
the risk/reward of a potential biggest loss of 18% to have a return from 3% to
21%.
What I look
for is ways to reduce the biggest loss and increase the lowest return. If we
can get those figures on an even scale, it will give us opportunity to analyze
the trade further.
So where to
from here, well the output is that we need trade ideas. And trade returns that
have a historical edge.
So let’s
look a bit further and we come up with BIDU;
SN
|
Years
|
BD
|
SD
|
LR
|
HR
|
DD
|
PP
|
BL
|
BIDU
|
10.6639
|
05//04
|
24//04
|
5.3834
|
16.6997
|
-8.3777
|
90
|
-8.4722
|
Outputs:
SN = Stock Name
Years = Years of data
BD = Buy Date
SD = Sell Date
LR = Lowest Return
HR = Highest Return
DD = Draw down
PP = Percentage Profitable
BL = Biggest Loss
SN = Stock Name
Years = Years of data
BD = Buy Date
SD = Sell Date
LR = Lowest Return
HR = Highest Return
DD = Draw down
PP = Percentage Profitable
BL = Biggest Loss
BIDU is an
example of when we look at the buy date of 05/04 and sell date of 24/04, based
on historical stock data there is a 90% chance that the return will be between
5.3% and 16%. Based on history, the biggest loss was 8%.
We have to
bear in mind that BIDU only has 10 years of data to analyze, which gives the
strategy only 10 data points.
Likewise
with all stock codes analyzed the data set we are looking at is limited, and at
the end of the day we have to work with what we have. In a perfect world it
would be good to have a 1000 data set points but with this strategy that is the
inherent flaw with the testing.
I would not
bank your house on this data due to the short time frame, but it does give the
trader another way to look at the markets and trade the key dates.
A trading
example with BIDU was just recently triggered.
The stock
price on the 5/4, was at 183.80. The share price as at the time of 21/4 is
192.74. Looking at this data and the max risk, we can note that the biggest
loss was 8.5% during this date time frame. Based on that we can look at selling
put option premium at around the 8.5% max loss level, which will increase our
probability of profit as well, or alternatively buying the stock with strategy
to exit at positive 5% return during that time frame. Either way buying the
stock or using options the strategy was profitable this year as the stock went
from 183.80 to a high of 197.5 which is greater than the lowest return of 5%.
No comments:
Post a Comment