I over coded my simple 2D homogeneous diffusion model with one injector and producer wells using insane amount of for loops. Where was this when I was struggling :D
Great Video. Is it possible to also discuss how the greeks and the PnL are calculated intra-day . is this a approximation instead of a recalculation as calculating as eaxh market data ticks come in can be system intense so wondering what techniques are used intra-day.
The notebook is at: ru-vid.com?event=video_description&redir_token=QUFFLUhqbndQblgtTFJUNV9vYl92SWg5X0JhUEFQa2RzUXxBQ3Jtc0tsWDM1LTJIckZuVkRvZlJiNm0zQldFYkNLRktUbW5QRXhPOHJ6WGtNUkRYaGxjU1dKWGszUUpTUTZkUWlWdGpicU44RzFVNnJ3bVM2Tnl1dlhMX1BWaFVvNVhaSDI3eDVPVERxZGktTmx3bHNkN2owTQ&q=https%3A%2F%2Fgithub.com%2Fkpmooney%2Fnumerical_methods_youtube%2Fblob%2Fmaster%2Fbinomial_model%2FBinomial%2520Model.ipynb&v=W_f3UL8xHDw
That's real nice explanation, but I realized that this is working for any guess of initial slope, is it the equation or any kind of flaw in shooting method? If you're there, a reply would greatly help. thanks
It needs an initial guess to start the calculation. 0.5 was just the value I chose. In principle, it doesn't matter, though the algorithm works better if the first guess to close to the real value.
I am a pharmacist and I feel so stupid compared to the knowledge that you have shown here . Pharmacy school was sooooooo much easier than what you have done here .
Hi Kevin, Thanks for the explanation that's great. I have done a customized version of bisection in my code, in which I will be setting the low to 0 and high to 40 assuming that IV lies around the range of VIX value (maybe 25) if the method exceeds the lower limit (0) I will return 0 and if exceeds upper limit then I will return upper limit (40)
Hey great video! There is a single thing that doesn't quite add up for me, and that is the np.sqrt of the standard deviations in the code, but the relationship actually talked about the variance, so shouldn't we really take the square instead?
You're right in that it is a very clumsy way to do it. You can drop the sqrt, and replace std with var. the slope would then be 2*H so instead of multiplying by two in the return statement, you'd have to divide by two. It was written clumsily unless there is something I am missing.
Hi, very interesting video. I am trying to do a similar fit with data of cyclic voltamperometries in fuel cells, however I have tried to use the assimulo package with RungeKutta4 but when I print my results I just get the initial value repeated over the whole array of results. Would you recommend using another method of assimulo package of Implicit type or try to change the parameters of the method (for example the step size, time of simulation, etc.) I hope you can help me with this, thank you a lot .
It's hard to say if another solver would do a better job without seeing and playing with the code. Can you integrate the system with some parameters you know are close to the correct values? Are the parameters being fitted scaled so they are roughly the same order of magnitude? Are any parameters in an exponential function and are causing the system to explode to infinity or move wildly when adjusted? Those are some places to start looking.
Hi Kevin, you must have downloaded Close Price for 300 days data of some stock like Apple and shown. Even Data can be easily available in Google sheets using GOOGLEFINANCE function. I would love if you would use Google sheets and shown from actual past 300 Close price as to what is Probability of Apple Stock's price going up by 20% in coming year.
In principle, no, but in reality poor guesses can cause either poor performance or lack of convergence. All choices of initial values will give a slightly different result, but the difference is small since we set the tolerance.
This is awesome, I'm finding that option APIs which return greeks won't do this calculation for you and we probably dont want it anyway. Question: can you use the underlying beta value if you downloaded it as opposed to calculating it? I'm thinking they would be similar enough. Also do you use the fast beta calc or the standard one in the platform? thanks - great video! It took me about 3 years to figure out that I needed a beta weighted portfolio! Esp with UVXY and other inverse ETFS!
I played with them three ways 1) using recurrence relation 2) using ordinary differential equation 3) by orthogonalisation of basis {1,x,x^2,...,x^n} with inner product
It's been awhile since you've posted a new video. I'd love for you to continue your content. It's great. I tried to check for other socials to make the same comment but I didn't see any. Hope you come back and provide more future content
@@kpmooneyYou've done a good amount of IV. What about IV for the entire option chain? You can then show term structure. material prep from the Vix White Paper. I love watching and sharing the videos as an engineer who is in finance now you bridge both my passions so well haha.