@acer I would never have guessed with(:-Calendar):

@Christian Wolinski This is a much better way than my convoluted method...

@acer 

Thanks for your insights once again, you fixed it.

I did think about DateDifference, but on my maple its a bit mercurial.

Its seems previously invoked Finance package upsets it. 

with(Finance):
DayCount("13/1/2019","1/1/2019");
               
with(Calendar):
Warning, Finance:-Calendar is not a correctly formed package - option `package' is missing
Error, (in with) module `Finance:-Calendar` has no exports

@acer but I would never have come up with your formulation. I happy it works...

@tomleslie Thanks Tom

The only problem is sometimes I get this error when i attempt to look at the interim_file in Wordpad. I have to close Maple completely and reopen interim to see its contents.

@tomleslie re returns. Yes. The simplest answer is usually the correct one...

@tomleslie thanks for your effort.

I don't have the GOT package either, Certainly I would use the linear transformation, but it states its only valid for Omega Ratios of over 1. The link which explains the linearisation technique is no longer valid...

I have a suplementary question. I'm not sure about the monthly returns data. I'd assumed they just subtract 1 to get the percent monthly return, Cell R1C1 is 1.283 (28.3% -seems high).next month R1C2 is 0.429 (- 57.1%). But R1C8 is -0.311....??

I want to put my own data though this for 3 stocks, annual return for each (4 years worth):

Matrix(4, 3, [[0.049214060475074796, -0.039495815964258746, 0.11449725779665731], [0.10937360103809103, -0.07403192172846383, 0.05145193083198468], [-0.02002071584993901, 0.032795876292319816, -0.137325940737269], [0.061433054336773145, -0.07926813859959723, 0.5713767521086268]])

@Carl Love 

This is a picture of the excel worksheet (Evolver). Note the low variance... It is calculated independant of the budget. It obtains (slightly) differerent covariant matrix to Maple..

Your method of unitary budget solves the problem of maximizing growth while keeping variance low.

Using your second method: 0 variance produces only moderate growth (3.2%). Have I missed somethng?

@Carl Love You're right of course. I mistook a returns matrix for the covariance matrix. The clue was the 3x3 matrix for Q, whereas a returns schedule of only 3 years would be considered inadequate (would require 7 minimum)...

What i'm after is a procedure to maximize the return g (rather than the stated 10%), and at the same time minimize the variance X^t*Q*X.

@acer thanks 

@Carl Love Whatever size you define in Kitonums code, the resulting gif is exported as a 400x400

@tomleslie I wanted option 3 in Carl Loves post, works well!

@Carl Love edited. It seems to me there are two probabilities. The probability of the particular configuration occurring, which is 1/1001. Then testing whether that configuration was a result of proper randomisation. You applied Fishers test and got probability 6/1001. Because that is a lot less than 0.05 it is significant. i.e there is evidence the pick wasn’t done randomly. I didn’t really think about a test for significance when I originally posted the question.

I’m not sure if you are tearing your hair out at my interpretation but I’m going to bed now.

@Carl Love I don't have Maple 2019. But thanks anyway. I put it through your 2018 code and it works....

So my old maths prof was correct for the probability...1 / 1001 = 0.000999

@mmcdara 

Thankyou for your effort. Just to clarify, each contestant was given an opaque bag containing a buff.

Once they were dispensed, they were told to reveal their buffs (simultaneously). So that would be Scenario 2.