My efforts to use an arbitrage trading bot went bust. It was fun to look through though and I appreciate Butor very much for sitting me down and helping me fine tune things. this was a valuable lesson not only in bots but in the way opportunities are found. In the 5 weeks I had the bot running I maybe entered into two trades and I closed the bot on the 1st one as I had typed cat and not tail – to keep the output going. The 2nd time it was running I realized that there was about to be a trend reversal and I would rather just keep the money as all bitcoin if I can. I think Instead I would
like to use coinapult as a rudimentary strategy so that in times of collapse I could hammer that address with bitcoin and hedge against bitcoin value (of course I won’t bother doing that so much if I’m holding onto cash. As it stands I’ve been living on coin for several years now so I do have fiat but only to cover services that refuse to take coin such as Living and Gas and Groceries. (sadly). Well back to the topic. I’ve known of bot trading since I was working with cryptsy, but a good close python proficient expert friend of mine had shown me that bot trading was not only risky but could be gamed to you lose morey from other hostile traders and that you may be burning yourself in fees if the spread is not present or is unaccounted for or if there is constant slippage as you have no shorting opportunities for alt-coins (that I am aware of). I decided to give this C++ Arbitrage bot a little go ahead because I’m interested in C++ and ultimately bitcoin development overall. Blackbird is the name of the software and it is available on Github. I found it easy to use and implement. There were very few dependencies but those are simply a matter of following the instructions.
This bot would work best on either a cloud server as Butor setup or a remote workstation such as a pi or a laptop – I happen to have the laptop with a broken screen for this purpose that I usually ssh into. It was actually more of a hassle to get the different exchanges up and running – I also took the time to exchange the funds back into fiat which was no fun because I felt like I’m just losing on fees as I wasn’t trying to speculate at all. Butor’s reasoning for keeping the rates in fiat was that it kept you away from the danger of the market crashing and you losing a substantial position in your investment. When you start Blackbird there is a feature that causes the system to close if you keep your funds in Bitcoin so you must sell everything and turn it into dollars – I had setup my experiment to run across 4 different exchanges – BTC-E, bitstamp, Bitfinex, and Kraken. To run the experiment I happened to be sitting on 60 Litecoin which I traded for BTC to split up across the different exchanges. Originally my plan was Bitfinex and Kraken. But after thinking it down on paper a bit more I decided to dust off my old phone (and its 2fa keys) to unlock some other exchanges I signed up for a while back. So now the list bubbled to four (and would have been five but Gemini was retarded with the amount of paperwork and first born blood they wanted so I Noped myself away from that one) and Despite this, it turns out that one of them was off anyway.
BTC-E has a forever open trailing spread which remains higher than the other markets. As a result you forever are locked on an open trade position that the bot fails to escape. Being busy with life I didn’t really grasp the wasted effort of using btc-e until after the experiment ended. I should have converted the cash on btc-e back into bitcoin and withdrawn it back into the other exchanges once it I realized the module was deactivated (it had been installed into the software but was not configured … likely because of the issues with price separation I mention above)
Beyond BTC-E, my other qualms with arbitrage trading were the fees lost from trading my BTC into fiat balances. They were a small nag but not a problem ultimately. I was more upset over the fact that to really utilize arbitrage trading properly, you need a huge sum of money on all the different exchanges and even then, you’re limited by the lack of market movement. Basically, your earning power is limited by the smallest sum of money available within any given exchange in your lineup. It isn’t like you are making a 7% return on 10,000 USD. It’s more like a 7% return on 2K because you have divvied the money into each exchange. Even then, consider that I had set my spread level to trade very low at .20 % edge above fees – trade fees are .2% usually. You must pay them twice once to enter then again to exit the trade. So you have a total of about .80 % of your trade to consider then add in your profit. For our example here with .2% ROI added to .8% I needed to see a minimum spread difference of 1% between Bitfinex and either Kraken or Bitstamp (BTC-E was disabled remember). Splitting up my entire investment into 3 different exchanges and only being able to have the opportunity for three different trade types was very limiting even though it was always on and monitored (I could have more if I were willing to sign up new exchanges and/or if I were willing to find other exchanges that offered shorting options, beyond bitfinex) As I said, in five weeks I only saw two trades and they had never closed. Whereas there was a large bump and drop last week where the Bitcoin price had grown to around $460 then gradually back down to $370…I would have much rather had kept the entire balance as BTC and shorted it using coinapult once it seemed to reach the peak – giving into the fees and slippage lost I would easily have had enough headroom to clear .2 BTC in profit. The trade off would have been a requirement of more monitoring to look for trade opportunities.
Needless to say, I lost money through spending fees and making trade loss moves to changeout my funds back into coin and transfer back to my own private wallets offline. I’m going to try out using coinapult instead with these funds. Their service offers instantaneously hedging of Bitcoin without requiring you to login to an account and trade. That tradeoff of no need to login should not be ignored however – you are spending about .8 of a percentage point on each trade when being lazy and sending money to a coinapult address. Have you run your own Trading bot or had better luck automating your trading activities?