tag:blogger.com,1999:blog-6684450909424029692.post7810103896223800448..comments2017-06-27T18:30:50.663+01:00Comments on Quant Trader's Journey: Struggling Quant Episode 1: How I lost USD 500.000 while figuring out the link between questions, math, stats, coding and tradingQTJhttp://www.blogger.com/profile/14284461526360152361noreply@blogger.comBlogger5125tag:blogger.com,1999:blog-6684450909424029692.post-6179739622197747322017-06-26T21:06:57.037+01:002017-06-26T21:06:57.037+01:00Correction:
Two events are said to be independent...Correction:<br /><br />Two events are said to be independent if knowing one does not provide information on the other.QTJhttps://www.blogger.com/profile/14284461526360152361noreply@blogger.comtag:blogger.com,1999:blog-6684450909424029692.post-26212426562082509722017-06-26T21:03:56.161+01:002017-06-26T21:03:56.161+01:00Hi Stefan,
So you have an event with a probabilit...Hi Stefan,<br /><br />So you have an event with a probability of 0.25, this is something you observe daily and you are looking for the probability of this happening 4 days in a row. If there is independence you just multiply the individual probabilities as you are suggesting.<br /><br />Two events are said to be independent if knowing one does not provide information on knowing the other. Like knowing the outcome of your observation day 1 does not provide additional information on your observation on day 2. You know day 1 or not, day 2 probability is the same. In probability theory terms, conditional probability is equal to the unconditional probability.<br /><br />When you say when the markets are trending, you are imposing a new condition and this is a concept called conditional independence or dependence. When you say markets are trending your sample space now is narrowed down to all possible outcomes limited to trending markets so formerly independent events may be dependent now or vice versa, you need to check the following condition:<br /><br />P(AnB/C) = P(A/C) x P(B/C)<br /><br />Check this out https://www.youtube.com/watch?v=19Ql_Q3l0GA<br /><br />Btw, I do not know the problem in hand but there may also be dependence like autocorrelation in time series data.<br /><br />At one point you have to make assumptions to make life easier:)<br /><br /><br />Thanks,<br /><br />QTJQTJhttps://www.blogger.com/profile/14284461526360152361noreply@blogger.comtag:blogger.com,1999:blog-6684450909424029692.post-12142038263933634742017-06-26T19:47:32.678+01:002017-06-26T19:47:32.678+01:00Hi Jacob, I try to do similar research as you on i...Hi Jacob, I try to do similar research as you on index futurex using an application called Investor R/T which has nice statistical features.<br /><br />What are your thoughts on probability of consecutive events? Say you have an event which occurs 25% of the time over some large sample size. This event has now occured 3 days in a row; is the probability that it will occur a 4th day in a row 0.25^4? <br /><br />I've been thinking about this a lot and i'm not sure what the answer is. I think it depends on whether the events are dependent or independent of each other, however I feel that isn't easy to answer either. Perhaps when markets trend the behavior is more "dependent" and when they are consolidating/noise the behavior is more independent.Stefanhttps://www.blogger.com/profile/17251205562608969273noreply@blogger.comtag:blogger.com,1999:blog-6684450909424029692.post-53279943840939579222017-06-26T10:25:34.117+01:002017-06-26T10:25:34.117+01:00Hi Jacob,
When I first started, it was not clear ...Hi Jacob,<br /><br />When I first started, it was not clear for me what to focus, which markets, which instruments, which trading styles, which trading frequency, all were question marks. So I did a bulk reading and watching of everything that I found related for like 6 months. Then this cloud of information started to make sense. Then I decided where to focus, intraday FX for majors, fully quantitative and online. The reasons being FX trading is low cost, liquid, open 5/24, no constraints on short selling etc and easy to connect and automate. Very hard to find edges btw. And also I have made a decision on using price data only and avoiding advanced methods like machine learning which tend to overfit.<br /><br />When you make these choices, it is much more easier to come up with a workout plan for learning Python and math/stats. For example, I did not include that much stochastic processes and options pricing theory, machine learning, high frequency strategies (I can never have the infra anyway)<br /><br />So the steps you may chose to follow:<br />- go to MIT OCW and start with Prof. Tsitsiklis on Probability, watch and digest all the videos and lecture notes, it is also on youtube<br />- do the same for time series analysis with Prof. Mikusheva, links are available in my blog<br />- after doing these, start reading Quantstart, Ernest Chan and Jonathan Kinlay to start linking the theory and practice. You will find great python coding in Quantstart<br />- in parallel start listening chat with trader podcasts<br />- in parallel start doing basic coding with python but do not go into data science libraries yet, keep it basic, just google introduction to python, but do this with emphasis to pandas and numpy libraries<br />- put an average of 3 hours work per day for 6 months<br />- then make your choice, where will you focus?<br />- then create your workout plan tailored for your choice<br /><br />I hope this helps.<br /><br />Thanks.<br /><br />QTJhttps://www.blogger.com/profile/14284461526360152361noreply@blogger.comtag:blogger.com,1999:blog-6684450909424029692.post-91189424017872884242017-06-26T09:43:05.745+01:002017-06-26T09:43:05.745+01:00What steps did you take to learn Python and mathem...What steps did you take to learn Python and mathematics for quant trading?Jacobhttps://www.blogger.com/profile/04570123052061034116noreply@blogger.com