Aaaa no creo que puedo explicarlo en espanol lol... lo siento... pero puedo explicarlo en ingles, y se puede usar google translate :/. Lo escribire en una manera que probablamente sera facil para traducir.

I use a neural network with 3 layers. The inputs are the percent changes in price at each hour over a certain number of the most recent hours. The number of inputs corresponds with how many hours it looks into the past to make the prediction. The outputs are percent changes in price for future hours. Each output corresponds with an hour into the future. For example, the 24 hour prediction neural network has 24 outputs (and 60 inputs). This means that it looks at prices over the past 60 hours and tries to predict what prices will be during each hour for the next 24 hours.

The activation function is the Sigmoid function, and edge weights are adjusted using backpropogation. The average error is calculated by making predictions throughout the historic price data, calculating the percent error of each prediction, and averaging all of these percentages.

Espero que esto ayuda, aunque es escrito en ingles lol. Digame si tienes mas preguntas!

Well once the neural network readjusted the prediction after the dip in price at midnight EST last night, it predicted the general shape of the graph over the past 21 ish hours correctly. The peak occurred about an hour earlier than it predicted and it wasn't quite as high. The prediction was that the average price at the peak would be around $474 ish, but it ended up being something more around $465. Not a perfect prediction, but it definitely got the graph shape right.

It's predicting for a pretty intense drop in the next several hours, getting down into the $420s from around 1am to 3am EST, followed by a rise back up to the mid $400s. Let's see what happens!

I think you are calculating yours pretty much the same way I do mine? I just calculate what percent off each prediction is from the actual price, and then average all of these percentages, having it make predictions at every point throughout bitcoin's historic data.

Your suggestion about classifying everything as uptrend or downtrend is interesting, I might try that. That could be insightful to seeing if it predicts the graph shape well even if it gets the magnitudes of rises and falls incorrect.

Also I do not use a separate data set for testing - I train it and test it on the entire historic data.

lol I mean, obviously performance over one day isn't going to be a solid measurement of overall performance. I was just noting that since it's been accurate recently, it may be more likely to be accurate in the immediate future.

But also doesn't your program prove that mine is effective? You said that by guessing the price will always stay the same you had an average error of 2.68%, and that was only looking at the price since November. Mine has an average error of 1.3%, or often less, by looking at all data over the past 3 years. There's no way this would be possible if it wasn't guessing trends correctly more often than incorrectly. Honestly your post just gave me a lot more confidence in my own software lol

gracias xD

Cuando no hay algo que pasa en realidad que afecta el precio de btc, los predicciones generalmente son buenos. El problema ahora es que a veces cuando algo como lo que paso en China afecta el precio, los predicciones no son precisos. El neural network que uso ahora tiene 3 estratos. Voy a trabajar para mejorar los predicciones y crear un neural network con 4 estratos. Yo creo que esto ayudara. Vamos a ver!

So it has re-adjusted its prediction and, like I said might happen in my previous post, it still predicts a peak around the same time, possibly an hour later. Only difference is that the road to that point looks a little bumpier, and the peak isn't quite as high.

Hmm so in about 30 minutes it is going to re-adjust it's prediction based on this drop we just had happen. I doubt it's still going to predict a price peak at that time but we'll see. It might still call for a peak around then, just at a lower price.


Like arepo said, it trains a neural network on the historic price data. It seems like it's mostly been better at predicting shapes of graphs and trends rather than specific prices. It may know when the price will peak or dip, but sometimes it will get the specific value of the peak/dip wrong. Luckily though, this is more important than specific numbers.

Yes, I mean short-term. As in, it being the highest price in the next 24 hours. Seems to have dipped down all of a sudden in the past like 5 min or so though, so idk what this is all about.

Chancellor - The 24 hour and 5 day prediction charts are not always coherent for a couple reason. The first reason is because the 24 hour chart predicts average prices over 1 hour periods and the 5 day one predicts average prices over 6 hour periods, so the 24 hour chart is more likely to show slightly more volatility. The other reason they aren't always coherent is simply because the predictions are created by different neural networks with different inputs. In the case of significant differences between their predictions, most of the time the 24 hour chart is more accurate.

Darkstone2 - The error is how far off the predicted prices are from the actual prices on average. That means that if the 24 hour chart predicts a price of $500, the price will, on average, fall between $493.5 and $506.5. Sure it's not 100% accurate, but you will almost definitely not find a human who will do that well. Plus, the error includes when real-life events that can't be foreseen by software affect bitcoin prices, and during those times the price is further off than those average errors. This means that normally, it is actually more accurate than that. Over the past day ish it has definitely had an average error of less than 1% for the 24 hour chart.

I can assure you this does much better than random dice rolls lol... I was testing a trade simulating program that made decisions based on these predictions for a while, and it was earning a decent amount of money until the China news came out and prices plummeted.

Oh and the 20 day prediction is more for kicks and giggles. That's why I put the little note above the chart explaining that it isn't accurate... plus it says it has like 8% error.

Why are people trying to help him portray himself better? lol... he's CLEARLY trolling

Hey guys, I made a prediction about a week ish ago about a dip in price that would occur at 4am, followed by a returning rise. The prediction was semi-successful in that the dip occurred, and the bottom of the dip occurred almost exactly when predicted, but the rise did not occur.

But anyway I would like to make another prediction now! Some of you have seen my website (http://www.btcpredictions.com). It has charts displaying predictions made by software using neural networks, and it's updated every hour based on the previous hour's prices. Right now, at 10:10pm EST, it is predicting a continued slow and steady rise in price until about 2pm EST tomorrow, when it says prices will start to go down a bit.

I believe this prediction is going to be pretty accurate. It has been making accurate predictions pretty consistently since the market calmed down a bit after the whole China ban thing. We'll see what happens though, this is certainly no guarantee!

http://www.btcpredictions.com

Hey guys, this is a website that displays the data produced by software that I've been working on for a while now. The software tries to predict future bitcoin prices using neural networks. Predictions are displayed on charts on the home page, and updated every hour.

It has been relatively accurate in the short time this website has been up. You can see the average error for the neural network that made each prediction above the chart, which is updated whenever the predictions are updated. Usually the 24 hour prediction chart is around 1.3% average error and the 5-day chart tends to be around 2.6%. There is a 20 day prediction chart as well, but that one is a bit less accurate (~8.2% average error). The further the neural network tries to look into the future, the more difficult it is for it to give accurate results. The average error numbers have been gradually lowering though as more data becomes available to base predictions off of.

Unfortunately, the predictions tend not to work as well when news events occur that significantly affect bitcoin prices. I'm going to work on implementing a more complicated neural network soon that may help reduce this problem. If anyone decides to use this as a tool for buying/selling, you need to consider this. I think that these predictions could be very useful, but they have to be combined with diligently following the latest bitcoin news and using your own intelligent judgement.

OHHH I think I misunderstood what you wrote. That is kind of a scary thought I guess. I still think it wouldn't be feasible for a couple reasons though.

1. I think they are a bit more expensive than $3k per TH/s... I didn't look very hard but after a few minutes looking around on google the cheapest I found was $5,500.
2. Obtaining 45,000 of those would be extremely difficult, and by the time you did it, you would probably need a lot more mining power to take over half of the network.

Honestly I think if somebody had the time, money, and resources to do this, they could make a lot more money doing something else like starting a business or something.

Plus even if someone did take over the bitcoin mining network and was able to change the blockchain to give themselves a ton of bitcoins, people would realize this pretty quickly. And as soon as people realized this, bitcoin's value would plummet down to 0 like literally instantly.

Well idk, I hope this helps you sleep lol

One thing to keep in mind is that if someone were to do an operation this big, you would have to factor in more than cost per TH/s to the total cost. You need a physical location to house all of the equipment, cooling equipment to keep all the equipment at a functioning temperature, electricity cost to run everything, probably a significant and expensive amount of bandwidth, shipping cost for all of the equipment, and possibly more (that's just what I thought of off the top of my head). It is interesting to think about how much it would cost, though in reality I think purchasing enough equipment to do this and actually take over the network would be near impossible. If someone were to take over the network, the best way would probably be to use "zombie" computers like in DDoS attacks (except mining bitcoins), and amass an army of "zombie" computers large enough to gather 51% of the mining network, although this is probably impossible too.

http://www.btcpredictions.com

Esto es mi proyecto para predecir el precio de bitcoin. Uso artificial neural networks para analizar la historia completa de precios y hacer nuevos predicciónes cada hora. Obviamente no es perfecto todo el tiempo, pero es effectivo un grande parte del tiempo.

Lo siento que el sitio es en ingles... y que mi español es lejos de ser perfecto lol...

How could you possibly be so sure of this? I seriously doubt you are right. All price patterns seem to indicate othewise. If you are correct then it is out of luck because your prediction doesn't seem to be based on anything...

I believe that once you buy some bitcoin, you should not even think about how much you paid for them. Honestly what you originally paid should have no bearing whatsoever on when you choose to buy or sell. If the price is going to go down, the right decision will always be to sell, even if it is for less than what you originally paid. And on the flip side of that, if the price is going to go up, the best option is always going to be to buy.

Although personally I don't buy/sell often. I'm a fan of hodling xD

I'm a bit confused but maybe I didn't understand what you are saying - are you suggesting that the government should NOT get a portion of the peoples' wealth?

Honestly imo the biggest problem I see with the idea of bitcoin being a single universal currency is that it is difficult to tax. Tax is necessary and society can't function without it...

That's an interesting chart! Looks like it's predicting a "breakout" soon - We'll probably need some good bitcoin-related news to give it the initial kick that starts the breakout, but I certainly hope your prediction is correct!

I feel quite confident that we've hit the lowest point already. I think things will go up from here.