tag:blogger.com,1999:blog-5902574258306547143.post3896905387065750016..comments2024-01-18T12:48:27.027-08:00Comments on Shifting sands: A visual look at market randomnessPetehttp://www.blogger.com/profile/09556415961629445879noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-5902574258306547143.post-67746864476354444432012-07-14T19:14:58.761-07:002012-07-14T19:14:58.761-07:00It really depends on what you are doing. I have lo...It really depends on what you are doing. I have looked at lower timeframes with FX trading, but not yet utilizing machine learning. The very high frequency stuff like HFT is more like market making which is not really what I am focussed on.Petehttps://www.blogger.com/profile/09556415961629445879noreply@blogger.comtag:blogger.com,1999:blog-5902574258306547143.post-44394482277448160732012-07-12T11:24:15.942-07:002012-07-12T11:24:15.942-07:00As far as working with Machine Learning, have you ...As far as working with Machine Learning, have you found the same to be true---that you tend to get better results working with lower freq. data?Williamhttps://www.blogger.com/profile/03006742452208934542noreply@blogger.comtag:blogger.com,1999:blog-5902574258306547143.post-12406359973104538142012-07-12T05:54:53.637-07:002012-07-12T05:54:53.637-07:00Hey thanks, I haven't actually looked at highe...Hey thanks, I haven't actually looked at higher frequency data, partly because I don't have it available. I do have some intraday forex data and it might be interesting to see what comes out of that. <br /><br />I think lower frequency data is more clear about what is really "going on" in some respects, the lower the time frame the more noisy things get, and more subject to distortions like big orders going through or whatever. That is why I tend to focus on daily/weekly timeframes. <br /><br />Also with a lot of data, charts like the above get very big :) But I think I'll give it a go.Petehttps://www.blogger.com/profile/09556415961629445879noreply@blogger.comtag:blogger.com,1999:blog-5902574258306547143.post-52297426133393206062012-07-12T01:37:22.428-07:002012-07-12T01:37:22.428-07:00Very cool. In your research so far, it seems like ...Very cool. In your research so far, it seems like the lower frequency data you use, the less 'random' the data appears to be. <br /><br />Have you applied these tests to minute/hourly/etc. data? I would guess that the higher the frequency, the more random/noisy the data would appear.<br /><br />What do you think?Williamhttps://www.blogger.com/profile/03006742452208934542noreply@blogger.com