The flipist method

[rebuilder]

I'm thinking of starting a trading experiment based on randomness. I tend to believe markets, like complex systems in general, are mostly chaotic and unpredictable. If this is true, then a random trading strategy should be as good as any. So, I figure it's time to put to practice an idea I've considered, as dictated by fine, ancient Persian customs, both sober (earlier) and inebriated (now): Starting 7th January 2012, 10 AM GMT, I intend to start a trading account, fund it with, say, 100 dollars (dollars to keep better protected from BTC volatility, 100 to make it a bit more interesting), and choose, by flipping a coin, every 24 hours whether to buy or sell. If I hold BTC and the coin says sell, I'll sell, if it says buy, I'll hold on, and vice versa.

I'd like input on a few things, keeping in mind I'm obviously willing to take total loss on my investment here:

-Is there any difference to starting an account on Bitcoinica or Mt. Gox, given that I intend to take a very low risk profile to prevent some sudden zhoutonging from mucking up the experiment. Currently I'm leaning towards simply making a new Gox account, but having a 1:1 Bitcoinica account would let me just take a screenshot of the current net value every day.

-How much should I trade with? I guess without leverage I may as well go all in, simply trading my entire balance at a time either way, as dictated by the coin toss.

-Apart from "this is a tremendously bad idea", is there anything I'm not considering here?

[old_engineer]

Why not apply your strategy to historical market data?  You can run a full simulation of how well your idea would have performed.

[teflone]

I love the idea..  curious as to the result..

How will you decide the start of this?  This should be a random hour of the day too ?

[rebuilder]

Why not apply your strategy to historical market data?  You can run a full simulation of how well your idea would have performed.

That's what I thought of first, but where's the fun in that? If I do it live, we get something to look forward to each day...

Anyway, I'd need to run several simulations, each of which would likely yield completely different results. What I'd need is historical data of many people using, and actually sticking to, other strategies to compare the random strategy to. I don't think anyone's likely to have such data, or if they do, to share it.

If someone feels like outlining a strategy now, though, and showing that they can beat my random strategy, that'd be interesting, though. Of course really we'd need several examples of the random strategy being played out to really compare, though... Maybe I should start several smaller positions instead?

[antoineph]

Why not apply your strategy to historical market data?  You can run a full simulation of how well your idea would have performed.

That might be useful, but it wouldn't fully test the theory because in real practice his trades affect the market (even if they are small)

And if markets ARE inherently chaotic and unpredictable, there's the butterfly effect to consider.

[rebuilder]

I love the idea..  curious as to the result..

How will you decide the start of this?  This should be a random hour of the day too ?

True, but in the interest of actually seeing this through for any length of time, I think it's best it's a time when I'm likely to be awake and ready to execute the transaction. Until I can be bothered to figure out how to code a simple bot to take care of the whole thing for me, anyway.

[Cluster2k]

I'm thinking of starting a trading experiment based on randomness. I tend to believe markets, like complex systems in general, are mostly chaotic and unpredictable.

This is a variation of trading techniques based on randomness.  I've read about people picking stocks by using a dart board, seeing on which stock ticker their cat would walk and where their pet bird would poop.  Quite often these stories hit the media when the stock market is booming, and it seems no matter which stock is picked there are sure to be plenty of winners.  The stories disappear when the market turns.  

Bitcoin is in a mini boom at the moment, and the random trading returns  

[rebuilder]

This is a variation of trading techniques based on randomness.  I've read about people picking stocks by using a dart board, seeing on which stock ticker their cat would walk and where their pet bird would poop.  Quite often these stories hit the media when the stock market is booming, and it seems no matter which stock is picked there are sure to be plenty of winners.  The stories disappear when the market turns.  

Bitcoin is in a mini boom at the moment, and the random trading returns  

Yes. My intuition is that, assuming I take a low-risk approach without leverage, and avoid getting wiped out completely, the returns should in the long term closely follow the market trends in general. That's actually the biggest objection I have to this myself - it might not be that interesting, at least without the ability to compare to more analytical trading approaches.

[Otoh]

I'd use a random generator rather than a coin flip to distance it from your sub/un-conscious influences as much as possible, e.g.:

http://www.random.org/

there're also cool ones based on atomic decay for free out there

& with a 3 option outcome: sell, hold or buy

start with a float of say $100 then each 24 hrs s/h/b $10 worth

& bitcoinica for the screen shots, but constant buying selling there will just lead to being eaten up by the spreads, so perhaps factor in a 1 in 10 chance of either selling or buying & a 9 in 10 chance of just holding your position

- I'd be interested in mirroring your trades too, perhaps with added variables

[DarkEmi]

The answer is quite obvious, the average profit you should expect is half the performance of the BTC / USD since on average you will hold bitcoin half the time and USD half the time.

So it is not really useful, you would obtain as well the same average by keeping half your money in BTC and the other hald in USD without the variance

The idea is still fun however xd

[sgbett]

Why not apply your strategy to historical market data?  You can run a full simulation of how well your idea would have performed.

It's random though! It would be different each time

[trogdorjw73]

I'd stay away from Bitcoinica for a randomness experiment -- even at 1:1 leverage, you're still likely to get Zhoutonged. You could try two tests, though: one is to do a 50-50 coin toss, and the second would be a 33-33-33 (really 67-33 or 33-67) with the sell/hold/buy strategy. That would give you a 67% chance of continuing your current strategy (e.g. sell and hold are the same if you've only got USD; buy and hold are the same if you only have BTC).

[StewartJ]

When will you go public and offer the IPO for the new "FlipCoin" investment concept ?

 

[FreeMoney]

I expect your results will depend more on what % amount and % time you spend in coin than anything else.

If you flip in and out in a way unbiased by price you'll do about half as well (or half as bad) as buy and hold, less fees.

[runlinux]

hmm... interesting idea.

wouldnt be too hard to test this though as stated above.

tempted to give it a try myself with as much data as i can find and excel...

[bb113]

Null it, man. Null them all.

I think this is a great use of an extra 100 bucks. Imagining how it would scale though is more interesting.

[damnek]

Maybe we can bet btc on the outcome

[zhoutong]

I really love this idea.

Bitcoinica might be a good solution for you because you can hold negative coins if you want. You can buy/sell say 5 BTC at a time, so that you can buy/sell multiple times in a row. (You can use our RESTful APIs for this too! It'll be even better.)

The account ledger feature can show you the balance changes everyday very clearly.

I'll match $25 if you deposit $75 into a new Bitcoinica account. Just PM me the username.

Good luck with your experiment! :-D

[rebuilder]

OK, it seems 10 GMT was optimistic for getting this show running.

I intend to take zhoutong up on his generous offer, so the trading will be done on Bitcoinica.
I need to sell a few coins to raise the full 75 dollars of my investment in this, and Gox seems to be pretty sluggish at the moment, taking a long time to execute transactions. Once that's done, though, we'll be live, probably in the next 10 minutes or so.

I guess this is one way of randomizing the time for the experiment...

I'll be using random.org to choose a number between 1 and 2. If it's 1, I go USD > BTC, if it's 2, I go BTC > USD. If my position already is the one indicated by the "toss", I hold my position.
That's about it, I guess. The experiment runs as long as I can be bothered with it. I make no guarantees of any kind. I don't know what I'll do with any money left once I stop the experiment.

I hope to post the first results in a moment.

[chunglam]

Hi Zhoutong, I suggest you to host a Bitcoin speculation competition. Every participant start a new account with 100 USD and no further deposit and withdraw allowed. After one week or one month, the most net balance participant win the competition. What do you think?