percentage of up days every 12 days

Question

maha al 2020-6-15

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/548745-percentage-of-up-days-every-12-days

评论： dpb 2020-6-15

Hello I have a data of stock prices and I am trying to write a code that calculate the pecentage of up days for the past 12 day.

equation as follow:

SPY= (number of times the closing price was up/ period)*100

thanks

2 个评论
显示无隐藏无

dpb 2020-6-15

Twelve calendar days, trading days, ...???

Moving twelve day window, consecutive 12-day periods, ...???

Data include dates or just closing price???

So many questions, so few details... :)

maha al 2020-6-15

I have only closing price. I want to take the numbers of days the stock went up for the past 12 days and devide it by 12. For example: if the number of times it went up was 4 times we had green market then 4 devided by 12.

lets say I have 100 closing price = closep.

didx = 2:length(closep)

% if closep(didx) > closep(didx-1)

% up=up+1

% elseif closep(didx) < closep(didx-1)

% up = up

% end

I dont know how to make it does it for 12 days by counting the numbers of up market taking the the first closep(2:13) then (3:14) then (4:15) and so on

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Adam Danz 2020-6-15

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/548745-percentage-of-up-days-every-12-days#answer_451602

Assuming your data are daily stock prices, you can use Y = diff(X,1,dim) to compute the difference across the n_th dimention of your data. If your data is a vector, you just need the 1st input.

Y will have 1 less data point than X since Y(1) is the result of X(2)-X(1).

Then look for positive values in Y. If Y(n) is positive that means the stock from X(n+1) increased from the previous day on X(n).

The rest should be straightforward.

2 个评论
显示无隐藏无

maha al 2020-6-15

thanks for your reply. How can I do it for specific period. Can you please elaborate. I want to do it for every 12 days.

Adam Danz 2020-6-15

编辑：Adam Danz 2020-6-15

I have the same questions dpb asked in the comment section under your question.

It sounds like you have a vector of daily stock prices (7 days a week). If you want to compute a 12-day moving average of the daily change, you can use M = movmean(A,k) where A is

A = diff(x)>0

A will be a logical vector of 1s and 0s indicating days that the stock increased. To get the avg number of days in increased within a 12 day period, assuming the input vector is daily values,

movmean(A,12,'endpoints','discard') * 100

See the movmean function for more option to control window size and how to handle the end points.

I really can't give more details without knowing more about your data.

请先登录，再进行评论。

Answer 2

dpb 2020-6-15

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/548745-percentage-of-up-days-every-12-days#answer_451635

编辑：dpb 2020-6-15

"counting the numbers of up market taking the the first closep(2:13) then (3:14) then (4:15) and so on..."

OK, so a running number...

One way amongst many...assume variable CP is close price vector...

nD=12;              % number days for moving percentage
dCP=diff(CP)>0;     % 1 if +, 0 not (following day higher --> T)
pctUp=arrayfun(@(i) sum(dCP(i:i+nD-1)),1:numel(dCP)-nD))/nD*100;   % percentage

NB: The above incorporates the difference vector dCP in the anonymous function -- if the data change, then the difference must be recalculated and the new result computed.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

Answer 3

dpb 2020-6-15

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/548745-percentage-of-up-days-every-12-days#answer_451647

However, the neatest solution is probably

nD=12;                                       % number days for moving percentage
dCP=diff(CP)>0;                              % 1 if +, 0 not (following day higher --> T)
pctUp=conv(dCP,ones(1,nD),'valid')/nD*100;   % percentage

2 个评论
显示无隐藏无

Adam Danz 2020-6-15

编辑：Adam Danz 2020-6-15

That's not too different from my answer and produces the same results.

nD = 12;           
CP = rand(1,1200); 
% conv method
dCP=diff(CP)>0;                              % 1 if +, 0 not (following day higher --> T)
pctUp=conv(dCP,ones(1,nD),'valid')/nD*100;   % percentag    
% movmean method
dCP=diff(CP)>0;                                                                           
pctUp2=movmean(dCP,nD,'endpoints','discard') * 100;   
% Compare results
figure()
plot(pctUp, 'bo', 'DisplayName', 'conv method')
hold on
plot(pctUp2, 'rs', 'DisplayName', 'movmean method')
legend()