Blackjack Notebook

[k_c]

I was contacted by Hugues Hoppes who describes himself as a computer science researcher and previously Principal Scientist at Google and Research Manager at Microsoft Research. He has created a Blackjack Notebook in which he has referenced my online program.

This was the message sent:

Hi,
Thanks for your nice website at http://www.bjstrat.net.
I did some blackjack analysis and was able to match your hand calculator results,
which seem to be more accurate for SPLIT than WizardOfOdds.
FYI, my notebook can be found at:
https://colab.research.google.com/gi...lackjack.ipynb

Cheers,
Hugues

Evidently he has put a lot of work into this. The link will not load on my old laptop but I can view it on my cell phone.

k_c

[DSchles]

Impressive body of work.

Don

[iCountNTrack]

This indeed very good work. A few quick thoughts:

a) I love Python but i dont think it should be used for BJ analysis because it's much slower than C++, some of CPU runtimes are really high

b) it appears he's using sims to compute split expectations

c) No reference to this forum or QFIT products (that's blasphemy)

d) It's 2022 a nice simple GUI would be good

[Norm]

I'm not a fan of any new general purpose languages. But, R is really neat for data analysis.

[iCountNTrack]

I'm not a fan of any new general purpose languages. But, R is really neat for data analysis.

Yeah R is great and it's fast. Of course, the speed comes from using modules written in C or libraries written in Fortran, but that's the beauty of a scripting language . Sadly these fast libraries are not useful for Blackjack because we are not doing any Fourrier transforms or curve fitting, just some old testament loops and conditional statements and basic arithmetic which are significantly slower with a scripting language.

[Norm]

I had to convert a program from MatLab to VB a few years back. It included a multi-linear regression; while VB has no math functions other than some trig and logs. I used OLS regression using matrix math (X’X)^-1X’y, also not included in VB, and requiring a matrix inversion. One line in MatLab took 76 lines in VB. Once done, I told the guy who wrote the MatLab code I see why people pay a pile of money for MatLab. But then, it would take one line in R for free.