What about going back to more prosaic discussion of the current price movement? Where do you think is support and resistance? What chance the triangle will resolve up or down?
Not sure what you have in mind. This one?  Daily RSI, Williams %R and MFI closed below 50, and don't seem to move for now. I can see us break out downwards, but I'd expect we find support at $533/$534 (3d BB, 23% fib $1160 to $340)  |
|
|
|
|
Of all the dubious log linear extrapolation models, the least dubious one.
(yes, I mean that as a compliment.)
|
|
|
|
You mean you have few enough model parameters to make overfitting the given data unlikely. Yes?
That's not possible for a neural net. The weights assigned to each node of a neural net can hold an enormous amount of data. A trained neural net by definition fits to the input data. You don't see it if you examine the node data because that data is spread throughout the entire network. I didn't really want to say it's likely, but you can in principle set the no. of parameters low enough, given the points of data, s.t. the risk overfitting is arbitrarily small*. Keep in mind that this was my responsne to K128kevin2's claim that "overfitting is impossible with this model", which I found (and still find) hard to believe. * If you challenge me on this, I'm afraid I'll have to go through my trusty old Bishop to find the answer  |
|
|
|
Just out of curiosity: the data used to determine the profitability of the network's outputs wasn't used in the training phase of the network, correct?
I've tested it on the same data and on different data, and the results are very similar. The issue with training and testing on the same data is generally overfitting, but overfitting is impossible with this model. The testing I've done with neural networks on future data has yielded pretty much the same results as tests on historic data. You mean you have few enough model parameters to make overfitting the given data unlikely. Yes? This still allows for subtle effects how well your network generalizes to new input, especially for something as error sensitive as trading. There's a reason why in ML the data is usually partitioned into a training set and a test set (or three, including a validation set), and the two are kept separate. |
|
|
|
|
@BitCoinWisdom
It's starting to look like Mintpal is the major exchange for altcoins (other than LTC). I noticed they're already added as an exchange.
Could you consider adding Mintpal's top 5 currencies by volume, whatever they are at a time, and update that list, say once a month or so? Altcoin volume changes quickly, and you probably don't want to list all of their currencies, but the 5 or so biggest currencies on the perhaps biggest altcoin exchange seems reasonable, no?
|
|
|
|
|
2014-06-23
trend: down
intraday median daily price: $591
intraday daily MACD hist: -13.3
RI = 38
comment: last day of trading didn't exactly help the reversal case. RI further away from 50 again. MACD still pretty far from turning positive.
|
|
|
|
|
Just out of curiosity: the data used to determine the profitability of the network's outputs wasn't used in the training phase of the network, correct?
|
|
|
|
I know nobody likes to hear this, but looking at the daily and 3 day MACDs slowly moving into a position that counters each other, I can see us in an uneasy balance between 500 and 700 for another month, maybe two :/
I agree, some known Bitcoiners are investing in other projects (Bitcoin unrelated), I was also discussing the price with a friend of mine and he said he doesn't think that a new rally will be that easy or that quick after the MT.gox fiasco... and he also said that people doesn't ever realize how MTgox will effect the future of Bitcoin regarding regulation and adoption. We see the same thing, but draw very different conclusions then. This is the first time in a long while that I recommend friends that asked me about it to buy in, if they had any intention to do so. In previous periods, I always considered the downside risk simply too great to make that recommendation in good conscience. Now, I'm telling them that they shouldn't expect huge rallies immediately, but I also consider the risk to the downside comparably controlled. I used to tell people to invest at any price if they are thinking about a long term investment (couple of years) but at this point I am losing faith, and the last time a friend asked me about investing I told him that I don't know and he has to study the whole market himself, I don't want to see friends thinking how disappointing this investment was because of me. Obviously, the "do your own research" and "all risks are your own" rule applies always. But this is a case that plays out similarly for many of us, I presume: someone who's not yet into Bitcoin asks someone who's balls deep into it for advice. And I'm a lot more comfortable with them being slightly unsatisfied because their investment didn't take off immediately, than if they followed my recommendation and would sit on a substantially underwater position for half a year. And, like I said, I see a lot less potential for the latter to happen than at any other point in 2014. |
|
|
|
I know nobody likes to hear this, but looking at the daily and 3 day MACDs slowly moving into a position that counters each other, I can see us in an uneasy balance between 500 and 700 for another month, maybe two :/
I agree, some known Bitcoiners are investing in other projects (Bitcoin unrelated), I was also discussing the price with a friend of mine and he said he doesn't think that a new rally will be that easy or that quick after the MT.gox fiasco... and he also said that people doesn't ever realize how MTgox will effect the future of Bitcoin regarding regulation and adoption. We see the same thing, but draw very different conclusions then. This is the first time in a long while that I recommend friends that asked me about it to buy in, if they had any intention to do so. In previous periods, I always considered the downside risk simply too great to make that recommendation in good conscience. Now, I'm telling them that they shouldn't expect huge rallies immediately, but I also consider the risk to the downside comparably controlled. |
|
|
|
|
I know nobody likes to hear this, but looking at the daily and 3 day MACDs slowly moving into a position that counters each other, I can see us in an uneasy balance between 500 and 700 for another month, maybe two :/
|
|
|
|
Nice corrective trend so far, down from the poloniex ATH ("ATH" by dismissing the very first trades at 0.01). Almost perfectly within the maximum bounds of corrective retracement, without turning into a dump. Either the result of a remarkably mature market, or we have a benevolently guiding whale  EDIT (a day later): boy, that escalated quickly... down to 0.0048, but already on the rebound half a day later. Next stop, 0.0068? |
|
|
|
To Tzupy, MaxwellsDemon, zimmah -thanks for the appreciation. I wish I was good it TA, but I've just started learning it. To cbeast - special thanks. The sign of real popularity is people parodying you  You made an analogy that TA is like math and physics. Math and physics make 100% accurate predictions. You imply that TA makes 100% accurate predictions. Please demonstrate your predictive science. No such implications was made. The table is only about ways of processing raw data. The way we calculate MACD from price and histogram from MACD is similar to way we calculate speed from coordinates and acceleration from speed. The part of physics that deals with it is called kinematics and, being pure descriptional one, makes no predictions. To make a prediction we have to introduce idea of force. (Which would convert kinematics to dynamics. Dynamics does make predictions. To make TA a predictional science we have to introduce idea of " market forces". Which I didn't do, because it would be an analogy rather than science. Hehehe. Nice. One remark though: I like your table, but your original equivalence was: divergence - second derivative - acceleration. You sure about that one? I'm still trying myself to wrap my head around how to best conceptualize 'divergence' (in the TA sense), but I'm not sure if it's as straightforward as it being the second derivative. In the paragraph above you seem to substitute 'histogram' for 'divergence', which I think is more accurate. 'Divergence' is, imo, at least another step removed from that. |
|
|
|
I'm really enjoying reading this thread. It appears that you have a nice work in progress that may develop into a useful tool.
Back when I was still trading/speculating vs. hodling; I wrote a python script that was intended to do something similar - warn about reversals. Several variables were used and assigned a value of -1, 0, or +1 (like counting cards in blackjack). The scale was -7 through +7. It worked pretty well before I broke it. I'll be following this thread with some interest. Good work.
Even though I no longer speculate in the short term, I still enjoy reading this sub and marvel at the detailed analysis that many of you guys do. Quantifying and predicting new markets is valuable beyond the scope of just bitcoin. Tip: weight your variables, add assumptions (be willing to remove them if they prove wrong), and be very critical of your process. Thanks for posting.
Thanks :) What do you mean though you "broke" the script? You changed the parameters, and didn't keep a backup of the ones that worked? Anyway, I like the idea of combining very different types of analysis, some chart based, some maybe news/sentiment based, aggregating them into a single value (if I understood it right that this is what you did.) Could then even be an interesting optimization problem to find the best weights for the different types of input. * * * 2014-06-22 trend: down intraday median daily price: $596 intraday daily MACD hist: -14.5 RI = 50 comment: 50, means RI is getting nearer to indicate a reversal (upwards). MACD still far from it, which makes it all the more interesting. |
|
|
|
I've been playing around with an experimental indicator in here, but I wouldn't really suggest anyone yet to make any decisions based on the output. I know I don't take it too serious yet. I lost interest in trading the swings a bit in the past weeks... "trading fatigue", is there such a thing? That's why I didn't post in here more often. For the record though, I enjoy all the other posts in here (justanothersheep makes a lot of good points imo, for example). Also, this is one of the few threads where traders/TAers don't have to defend themselves against the accusations coming from the "pure holder" mentality constantly. I did take smaller profits twice during the swing down since May 31st/June 1st, but went long again afterwards. Both trades were pure momentum based, so nothing that warrants an analysis post. So perhaps the following is not a really deep or interesting answer, but: I'm still long as a default position because I'm not really worried yet about the longer perspective upwards (say, looking forward a month from now). That general positive outlook could change of course, but as long as we stay above what I think are key support lines at $560 (1w BB, 50% fib $440 to $680), and $535 (23% fib from $1160 to $340, 3d BB), I'm okay with the situation... I think we are simply very cautiously moving ahead, in a "2 steps forward, 1 and 1/2 steps back" rhythm, if for no other reason than that the sentiment of the 2013/2014 bear market is still lingering, and because we are still waiting for that next "fiat floodgate" to open (an ETF, a new exchange associated with names from traditional finance, or even just a new national market that suddenly develops a taste). The USMS auction might play a role too, but I think its effects are overrated. |
|
|
|
Being a HODLER has many advantages like no need to submit yourself to exchange risks and no need to wreck your brain and invest time into deciding when to sell and when to buy back. The disadvantage however is that you may spend a lot of time underwater or lose your entire investment one day if Bitcoin fails. But these people don't care as it's either all or nothing for them, and that's also a perfectly valid strategy if you ask me. There's an implied advantage that you've hinted at here without bringing up specifically. If one does not need to devote time daytrading, then that time can be used for other things. Regardless of whether Bitcoin succeeds or fails the entire investment is not lost, because the investment consists of Bitcoins plus whatever else was done with the time that was not used for daytrading. Valid point, but there's this recurring theme on this forum that you're either a "holder", or daytrader. There's obvious a huge spectrum in between those two sides, both in number of trades on average, and time commitment. |
|
|
|
Weak hands are people like MatTheCat, fonzie, and some others who are scared to hold coins for long and when the tree starts shakin' they don't know how fast to dump their coins again. Also people like the ones Blitz referred to who buy some coins to ride the imminent bubble and when they don't see a bubble within 2 weeks they will sell their coins again disappointed. Faalhaas is a perfect example of this. Then we get to the Veronicas who think they are a strong hand at first, but then get really scared when they see the price drop hard and eventually give in and dump their coins at the bottom or close to it. This is also a weak hand, but one pretending to be strong until tested and these people will naturally suffer the most. Then we have the day traders who use TA and stuff to try and buy low sell high. They are also weak hands but a small percentage of them are actually good enough to make money from the market. Most of them will miss the biggest profits during the big Bitcoin bull rides though, but they're at least trying to manage their risks. Then we get to the long term investors who are generally long on Bitcoin, but take profits out during big rallies and they also may buy more when they see dips. These I regard as strong hands with a decent sense of risk management. In terms of time investment and profitability I think this is the most efficient form of trading/investing in Bitcoin and I personally try to stick mostly to this strategy. I admit I do tend to also be a part of the last group, the true superstrong hands who are called the HODLERS of course. Being a HODLER has many advantages like no need to submit yourself to exchange risks and no need to wreck your brain and invest time into deciding when to sell and when to buy back. The disadvantage however is that you may spend a lot of time underwater or lose your entire investment one day if Bitcoin fails. But these people don't care as it's either all or nothing for them, and that's also a perfectly valid strategy if you ask me. Quoted, for future reference. Pretty much the best high-level summary of our diverse little group in here I've read so far. |
|
|
|
|
2014-06-21
trend: down
intraday median daily price: $587
intraday daily MACD hist: -16.8
RI = 41
|
|
|
|
The real weak hands are those who fantasize about some imminent 10k bubble looking at retarded exp log chart trendlines and then dump when inevitably it becomes blatantly obvious that their inflated expectations were wrong.
Agreed, mostly. I'd say having any kind of "unwavering" assumption about where price will go is a necessary condition to be a weak hand. Not sufficient yet, because some hold such beliefs, but have the (dubious) character strength to sit out deviations from that expectation without panicking. "Dubious" because they might eventually pay for it. I aim to be largely agnostic about the far away price. I gladly acknowledge growth so far was spectacular, but there's no guarantee it'll go on like in the past. If you plan to trade , you should neither resist the observation (and the necessary conclusions) when price goes up, nor when it goes down. |
|
|
|
|
Holy necro.
And agreed on what is probably the sentiment that's behind your decision to resurrect this thread... that those who are now getting to the point of declaring our little experiment a failure are as wrong as they always were...
That said, the 2011 bear market wasn't really the best example for "just buy and hold" either. It seems almost trivially easy, compared to today's situation, to identify the turning point. And if you traded accordingly you had the chance to increase your coins by a factor 10. The market was a lot more "naive" back then, I'd say.
|
|
|
|
|
Show Posts
