https://i.imgur.com/fXNmuDK.png -Gox
https://i.imgur.com/sKvNTWM.png -China
Looks like we're setting up for the capitulation to flesh out. It's been a strange few days, it seems like there's an appetite to go down but the whales are holding for better prices. As usual gox is bouyant but china retains the initiative. For me that 6hr mass index reversal bulge is the main signal of an impending drop, bollinger's low and stochRSI can accelerate down for a while to make up for the past month. Also despite all the fighting volume is still relatively low compared to the main bubble. I'm not good enough to guess timing but my outlook remains bearish.
posting price targets creates a confounding effect that, if enough people know about and "trust" a specific target, and trade according to their best possible actions, the Nash Equilibrium will necessarily invalidate that target as players try to undercut each other for better and better prices. for this reason i usually only provide them in private or in limited release paid reports.
exactly, good signals from deep analysis can negate themselves if released, setting a bounty for this is perfectly reasonable, especially if the methods have worked better than chance recently.
I bet knowledgeable eyes watch the blockchain to see where and when large amounts of bitcoin flow in and out (are the addresses public? all of them?)
but no one will know the amounts of funds going in and out which would show appetite for support/pump/crashes
i've totally seen a massive surge in btc transferred when a crash stalled midway, i think in that case you can be pretty confident that most of it's replenishing exchanges to dump, which is some idea of the "appetite". But it's probably one of the only cases where watching the blockchain helps. Better than nothing though, and it's a decent signal as btc flows fast compared to fiat.
Charts on bitcoin exchanges is fun and I got roped into them for a while but all these hours would be best spent finding out the inflows/outflows of BTC & fiat per exchange on an hourly basis.
i hear what you mean. i find a lot of the trappings of the trader universe like emas, donchian channels, and marking a chart up with every possible fibonacci retracement to be a little distracting.
i work with a small set of classic indicators including the ones presented in the OP, as well as volume data, and fractal analysis, which is related to Elliot's wave model. fractal analysis attempts to find consistent patterns in the price function which can be used to better understand the "price environment". these patterns include triangle consolidation patterns, double-tops, cusp-tops, and bubble patterns (oscillation model). as for classic indicators, i find that the best indicators are the ones that incorporate volume heavily into the algorithm, because volume is one step closer to the inflow/outflow data you talk about.
and while i'm sure that these data would be a useful indicator in some sense (i'd expect it to look like a transformation of the mass index, peaking at times of of market uncertainty and reversals, correlating with peak in- and outflow), i think actual trade volume is far more important. i have often said that the only two important sets of data are the price function and the volume. i treat these data alone with a scientific approach and strive towards the simplest models that are effective at anticipating price behavior.
that being said, i have found empirically that they have "yielded returns better than chance during the time period which i have employed them". the problem of induction, of course, forces me to consider that this may simply be due to chance
so i'm not claiming any superpowers here regarding predicting what is an inherently stochastic function, the
price function.
--arepo
Awesome, have you seen this book by Benoit Mandelbrot:
http://amzn.to/1afxNeA ? I found it a great read, finished it just before he died (sadly). Without giving away too much can i ask do you look for self similarity across different scales? I'd like to give it a try sometime, probably start by minimising squared residuals between plots of varying length within a range. Def agree with sticking to actual trade vol and price as pure datasets, days destroyed/blockchain trans vol/fiat exchange rates etc would be fun to look at but their relevance might only be occasional and otherwise add noise to normal forecasting.
