Squawk /
09 April 2018 at 7:34 GMT
Thanks for the very knowledgeable explanation on the NFP post. I'm a ML guy trying to get into this field, but I lack the economic knowledge and intuition. I am now looking into developing a reinforcement learning automatic trading algo to trade the major US indices. Could you please point me to some resources to fill in these gaps I have regarding the macro economic calendar news? Thanks :)
1y
Algos and macro economics are diverse bed fellows IMHO. The latter operates slower for policy effects but the knowledge of potential macro events means the markets price the effect fully before hand and remember algos operate real time and faster than any economic macro policy. However as you wish to pursue algo scenarios may I point you to market profiling as an initial trade programming scenario ? I operate 35 algo programmes but they are BAYES based on the whole.
1y
The remainder are based on std devs and distributive data such as market profiling. The statistical confidence values are high.
1y
I agree to some extent. I think of them as algo features at different resolutions. You can very well have a Bayesian inference model which uses both macro economic data and intraday market data. I think non-parametric models using normal distributions are a very good way to attach the prediction problem, the best part of this is as you mention confidence intervals. By the way, I don't really understand what you mean by market profiling. My trading lingo is equivalent to a 2 year old. Please explain. On a different note than algos, what clicked for me lately was using Fibonacci extension patterns + RSI. I feel like I was blind before I read about that. I've just been eyeballing the NASDAQ 100 futures and have traded Fibonacci levels successfully in the last couple of weeks. Anyway, my goal is to get an intuition about everything that moves the markets, and the economic data news is one of them. Basically I want to be able to tell what impact a certain reading will have on equities.
1y
Something for you to also consider mate :-)
https://www.tradingfloor.com/posts/sentiment-momentum-and-the-perfectly-timed-trade-7019540
And an algo for you !!
https://www.tradingfloor.com/posts/may-the-force-be-with-your-algo-7928840
Just google Market Profiling (Steidlmayer) basically he uses std devs...there are two articles of mine describing algo's and BAyes and there is also Markov Property values as trade scenarios. I eschew fibbonacci traditional scenarios as curve to fit theorem as there are tighter trade functions IMHO. We can use Benoir Mandelbrot or Lyaponov or Freeman Zeroes formatting or Fourier functions. The Bayes addition of ANOVA or the classic Null Hypothesis give far tighter scenarios. Just a few pointers that might be of worth mate. Hussam Al Fayyad on site is statistician and far more qualified than me can give even more scenarios.
Finally matrices on FX is brilliant as the interpolar and correlations can be superb for daily revenue streams :-)
https://www.tradingfloor.com/posts/sentiment-momentum-and-the-perfectly-timed-trade-7019540
And an algo for you !!
https://www.tradingfloor.com/posts/may-the-force-be-with-your-algo-7928840
Just google Market Profiling (Steidlmayer) basically he uses std devs...there are two articles of mine describing algo's and BAyes and there is also Markov Property values as trade scenarios. I eschew fibbonacci traditional scenarios as curve to fit theorem as there are tighter trade functions IMHO. We can use Benoir Mandelbrot or Lyaponov or Freeman Zeroes formatting or Fourier functions. The Bayes addition of ANOVA or the classic Null Hypothesis give far tighter scenarios. Just a few pointers that might be of worth mate. Hussam Al Fayyad on site is statistician and far more qualified than me can give even more scenarios.
Finally matrices on FX is brilliant as the interpolar and correlations can be superb for daily revenue streams :-)
1y
As for data release impacts to current pricing watch the distribution of yesterday 15min or 30min and look for voids (Reeman Zeroes etc). Perhaps the classic TICK to TRIN and TRIX give a better level of market real time balance and bullish/bearish bias whilst being able to avoid most bull or bear traps.
1y
A loot of new words here for me, it's gonna keep me busy for a few days, thanks :) I'll most probably come back with more questions.
1y
April 16th the site will change here. My assumption is that you will still be able to access the articles posted :-) There is a whole year of them describing various parameters and reasonings along with data for results.
1y
I found a bunch of your articles, but they are not listed neatly under your profile, it's a bit hard to see if I missed any.
1y
Btw, do you only post stuff here, or can I follow you on twitter or something like that... ?
