Home » Uncategorized » That MIT Paper That Supposedly Knows the Future of Bitcoin.

That MIT Paper That Supposedly Knows the Future of Bitcoin.

The paper is right here: http://arxiv-web3.library.cornell.edu/pdf/1410.1231v1.pdf

The quotes are from here: http://newsoffice.mit.edu/2014/mit-computer-scientists-can-predict-price-bitcoin

Devil’s Advocate: So MIT Guy. I bet you are a billionaire now. I mean, knowing the future and everything. You must have sold your house and borrowed every last penny you could get your hands on, the better to invest it in bitcoins using your fool proof algorithm.

MIT Guy: Give me your money and I’d be happy to invest it for you.

DA: Did any Austrian Economics go into making up your computerized predictor? You know, trying to actually understand the humans who are buying and selling, and what makes them tick?

MIT Guy: Can we explain the price variation in terms of factors related to the human world? We have not spent a lot of time doing that.

DA: One thing AE says about all these computer predictions is that they work perfectly until they don’t. So I’m sure you tested your gizmo in all kinds of situations, bitcoin going up and down, over a long, long time period, to make sure you weren’t being stupid and cherry picking one tiny interval, say two months,  where bitcoin was going in one direction only.

MIT Guy: Let me show you a summary in graph form:

See? The blue line is the price of bitcoins, and the black line is how much money we made. So 50 days

DA: Oh, I see. 50 whole days. And I notice the ups were huge and the downs were slight.

MIT Guy: Hey, it’s not my fault we picked the one time this whole year bitcoin was going up. And I resent you calling that cherry picking. Give me your money and I’d be happy to invest it for you.

Smiling Dave: What’s worrisome is not that the guy doesn’t know how to do statistics. Plenty of statisticians don’t know their job. What’s worrisome is that he thinks statistics is where to look in the first place.

DA: Take your articles, for instance, Dave. With zero statistical data, using sheer logical reasoning from first principles, Mises and Timothy Terrel and Peter Schiff and others have predicted that bitcoin is going to die, and explained why it was doomed from the get go. And you summarized their work right here at https://smilingdavesblog.wordpress.com/2012/08/03/bitcoin-all-in-one-place/

SD: Yes, the current state of mainstream Economics is sad indeed. But hey, it always has been. I mean, the whole idea of having universities teach economics in the first place was to make sure the students learned govt propaganda. Read up on what happened to Say and Bastiat.

Advertisements

6 Comments

  1. derekmc7 says:

    The funny thing about mathematical models, is that you don’t have to directly encode all the information about the real world into them, they just need sufficient complexity to capture the real information from the data.

    That is exactly the benefit of these kind of mathematical models. You don’t actually have to know how the world works, you just have to demonstrate that the model has all the information that it needs to correctly mirror real world phenomena.

    Understanding how the world works can help you create better models, but sometimes it can lead to worse models as well.

    These kind of models are usually applied to predicting short term variations and not long term viability. They are most effective when applied this way, and that can make them incredibly profitable.

    Neil Tyson has a video of him walking a dog, explaining the difference between weather and climate change.

    Scientists study energy systems in order to analyze long term climate trends, but trying to predict short term weather is more about finding statistical correlations.

    I don’t know if I agree with your predictions about bitcoin, but I don’t know if it’s relevant to this kind of statistical modeling that this MIT guy did.

    Like

  2. Smiling Dave says:

    “… they just need sufficient complexity to capture the real information from the data…you just have to demonstrate that the model has all the information that it needs…”

    Your education is…?

    Like

  3. derekmc7 says:

    I was studying Mathematics but dropped out.

    The main point of my comment is that predicting long term and short term trends are completely different skills. He is making money using short term prediction.

    Like

  4. Pramod says:

    I don’t get this short term vs long term. Long term is nothing but a series of short terms. If something is sure to work in the short term, it has to work for the long term too. There is a lot of difference between saying ‘it works in the short term’ and ‘it might work in the short term’. People saying the former actually mean the latter.

    Like

  5. Smiling Dave says:

    Plenty of things only work short term. For example I can safely predict I will live for the next few minutes, but not that I will live for the next five hundred years.

    The thing is, lots of things have very small changes over time that do not mean much in the short term, but add up long term.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: