- Daily leverage (bull/bear contracts on stock) trading recommendation e-mail based on trending instruments.
- Create robot leverage instrument trader based on best traders in a Swedish stock community.
- Modify robot to use ETFs and leverage based on prices and added more stock exchanges.
- Moved on to crypto currencies, which has a much higher potential due to the lower commission (two orders of magnitude or more). Also the community is much more open, and you're free to download historical data from pretty much any exchange site out there.
- Built a rudimentary type of technical analysis bot for BitCoin and altcoin trading via the eminent Binance exchange.
- Moved over to a type of incremental swing trading algo, as outlined by this youtuber.
- Created a web view of the actual progress, etc.
- Added support for managing different user accounts, etc. and invited a couple of friends to invest, to apply some success pressure.
- Optimized the swing trading parameters multiple times, as the alt-coins waxed and vaned. Also tried out different frequencies of trading, and a small portion of the combinatory explosion.
- Ported the lot to Linux, Docker and placed it on AWS. Change underlying web DB from SQLite to flat file.
- Created another type of technical analysis, more closely related to the traditional types. Found it useful for nothing.
- Created another type of swing trading, with incredible high risk and enormous possible gains. I had miscalculated .05% on every trade in my simulation, which made it loose big-time in the live environment. Too bad.
- Implemented a kind of stop-loss functionality using traditional TA.
- Tried to train a neural net to find the buy/sell positions in any of the 190 coins available on Binance. Couldn't get it working properly.
- Instead focused on training the neural net to find streches of time where I should own any given coin. Couldn't get that working either.
- Went back to my first posit that I could get it to buy/sell for me. Didn't work the second time either.
- Tried buying the cheapest currencies which had a fair volume. Didn't do very well outside of when the influx of cache into the crypto currency market was high. A sort of high risk index leverage, I guess one might call it.
- Optimized the parameters for the above, and also built a pandas-like numpy wrapper for performance.
- Went back to using the original swing trading.
- Went back to using TA to control stop-loss.
So I've been doing this almost one year now, five months of which full time. None of the above worked reliable. My latest idea is to use TA to toggle between alt-coin trading and bitcoin trading. This is by far the best idea so far, as I've always struggled with how to handle baisse. I've currently created a much more reliable TA, which will help too.
I think this is the end of trading robots, apart from some optimization and bug fixing. This should work in most cases, and also produce a good yield.
I'm still not sure if I can live off of the returns, but as long as there are no major bugs in the simulation I sure can. Otherwise I'll have to settle with a good savings account. But I'll give it four more months before I'll let it slip so far.
