Bitcoin seems to be all the rage at the moment, and the massive boom/bust it went through recently only seemed to fuel interest in the fledgling peer to peer crypto-currency. It seems like a pretty interesting thing. Many people seem pretty convinced they can get wildly rich off it, and some no doubt will, but there seems to be no end of speculators and bitcoin gambling sites, and places to get free bitcoin for doing menial tasks (like generating artificial views or traffic etc).

The high volatility of the bitcoin value does make it look as if there would be the possibility of making small profits in riding the short term fluctuations, (I guess that’s day trading? Although I have almost no knowledge of finance.) This wouldn’t be about long term speculation on the value, but just hoping that if in the short term the value is fluctuating about some mean with a large enough variance, you should be able to buy BTC when it dips low and then sell when the price creeps back above. As some people argue rampant speculation is bad for the bitcoin economy, the above type of short term trading might be good for it, if enough people did it, it would help to smooth out some of the short term fluctuations of bitcoin value, maybe.

Not that I would actually have the balls to put my own money in the hands of bitcoin and my own poor programming skills, but I thought it would at least be interesting if you can make an algorithm that could perform ok on historic bitcoin trading data.

Basic concept

The basic idea could be something like follows:

• Buy when the price looks to have fallen below the current mean

• Sell as soon as you can make a small margin

• Be wary about buying anything when the market looks like it is doing anything other than going fairly steady.

Possible algorithms

So just to get some rough idea about how this approach might look I coded a simple implementation that takes a long and a short look at the trends in the current USD/BTC. I take two windows, one something like a day long, and the other just a couple of hours and fit a straight line to the BTC price over those two periods. The algorithm then decides what to do based on that. If the longer duration slope is relatively stable then it looks at the short, if this indicates a recent fall in the price a small amount of BTC is bought. The algorithm also continuously checks the current price against that paid for previously bought bitcoins and sells if the price exceeds some threshold above this value (to account for fees and a profit margin)

Rule 1 – When the long term trend (green) is fairly stable, and the short term (red) has fallen, buy some bitcoins.

Rule 2 – If the bitcoin value (purple) goes above a certain amount above the value at which you bought the bitcoins (x) then sell them for a little profit.

Running with test historical data

I downloaded some historical bitcoin trading data from the last two months at 15 minutes intervals, and ran the test algorithm. Simulating transaction fees and a couple other bits about not buying too frequently etc.

The green line represents the USD in the account, BTC aren’t shown, but the total value of both USD and BTC (at current market rate) is shown in red. So what do you know, it looks like it works. The total value of USD and BTC at the end of the run is about $130 which makes for an increase of 30% over the course of 2 months, equating to 480% over a year. Better than a savings account I guess.

Of course I am full of doubts as to whether it would actually work given the turbulent ups and downs and unpredictability of real life. Still, it’s an interesting exercise to think about.

A chocolate version of scotch eggs, creme eggs wrapped in doughnut and deep fried. I made some creme egg filled doughnuts last year, too late to get my hands on any more creme eggs to have a second go, but now the time has come! On reflection I might stick to mini ones in future, as a full size egg wrapped in sweet doughnut and rolled in sugar is a pretty hefty calorie and sugar hit, but they look good in an egg cup.

New places, new photos…

I got some pear essence from The Asian Cookshop, and it smells amazing, if incredibly chemically and artificial. But that was always one of the brilliant things about pear drops, which I guess are just a bunch of whatever ester makes that nice smell and some sugar. Because cupcakes are the thing at the moment (and they cause less anger when I spend ages photographing them before we’re allowed to eat) pear drop flavoured cupcakes seemed like a pretty good thing to make. And hell they were, everything best about pears, cupcakes and sweets were in those things.

I thought slices of candied pear would make a good topping, but finding reliable instructions for candying is surprisingly difficult. I took a little guidance form David Lebovitz and various other random Google searches. The overall idea seems to be simmering fruit in a syrup, and the bigger the pieces the longer it will take. Also gradually increasing the sugar concentration seems to be pretty common. So as long as you simmer the fruit long enough, in a high enough concentration of sugar I don’t think you can really go wrong. My rough procedure is below and it seemed to work out.

Candied (glace) pears

1. Peel and slice pears
2. Simmer in water for 10 minutes until tender
3. Mix 250 g sugar and 250 ml water, dissolve, then add pears and simmer, uncovered, for 20 minutes
4. Leave overnight, bring to a boil, simmer 20 minutes
5. Repeat until the syrup reaches the soft ball stage
6. Drain, let dry on a rack for 24 hours, then coat in sugar

I also added a chunk of ginger for flavour and also didn’t bother with overnight, a few hours seemed like enough. I think it took a total of four 20 minute simmers for the syrup to reach the correct concentration.

Having been working with magnets lately I was going to try and make a toy simulation of magnet domain formation, but in the end made something that looked a bit more like the sea sloshing around. Still a first try in using the javascript mode in processing 2.0, shame about all those libraries that wont work now (though I’m happy to see a toxiclibsjs out there), but apart from that it makes life with the web much easier! Press the mouse to randomise!

The sun is setting on my time in Oxford, it’s almost time to leave. It’s been a pretty place to live for a while but probably high time to move on. The stone university buildings around here make for some great texture to shoot HDR photos at night, I just wish my Sony alpha 230 would have a wider auto-bracket mode to make such things a bit more convenient. The two HDR photos here were tone mapped in Photomatix Pro, and then finished off in Photoshop.

The cows on Port Meadow are incredibly docile, apparently they’re pretty bored of people coming out early on a frosty morning and taking photos of them eating, pooping and standing around looking cold. Still they make good subjects for photos even if my hands were almost too cold to operate the camera.

It’s pretty nice living a few minutes walk from Port Meadow, especially as winter sets in, leaving icy fields, a cloud that clings to the damp grass and fog that hangs over the rivers. It also makes it a little easier that sunrise is so much later this time of year.

Next time more cows!

After seeing many amazing water drop photos all around the internet I tried out some quick attempts at droplet collision photos. Armed with a cheap speedlite, a tray of water, and a syringe it was actually fairly easy to get some decent splashes. It still takes a lot of lucky timing, and many failed attempts, so building an arduino controlled water dropper sounds like a pretty fun project and evidently leads to some amazing pictures. There’s also an amazingly large difference between milk and water at these scales, but the shapes both make are incredible.

Sombrero

A hand?

Mushroom

 

???

Another mushroom