A Sustainable Rally? … the AI Bots say No!

After the carnage of yesterday where bitcoin triggered stops all the way down to the 1800’s (yes it’s still unusual to talk about a collapse in the market down to 1800, a level it only broke above 2 weeks ago) the bitcoin market has had a decent rally back to above 2200, dragging the other cryptocurrencies up with it.

Is this the start of a new up-leg already? Well the AI Bots certainly don’t think so. They have used the rally of the past 24 hours to sell their remaining long positions; selling etherium (ETH) at 146.68, ZCash (ZEC) at 175.74, Ether Classic (ETC) at 15.518 and DASH at 96.855.

The models definitely appear to be in the mode of grabbing short term profits when it looks like the market will turn up then locking those positions back into cash. As discussed previously the models are not trained to go short, they simply take an unleveraged long position or switch back into cash so all cryptocurrencies being flat is as negative as they get.  Lets see if they’ve got this right and the rally fizzles out.

CurrencyStarting BalanceUSDCryptoRateUSD EquivTOTALGain/loss
Ether Classic$203.00$1,012.29014$0.00$1,012.29398.67%
Hold BTC$23,938.00$0.0022.907177032198$50,349.98$50,349.98110.33%

Note – apologies for the last 2 posts, they had the wrong position and performance figures attached (the last 2 trade updates hadn’t fed through). This has now been fixed for this post and the prior posts.

— Wintermute —

3 thoughts on “A Sustainable Rally? … the AI Bots say No!

  1. Just out of curiosity, for which timespan is the ai-bot looking for a long position? From the trading actions you posted so far it looks like days or even weeks?

  2. Hi Nils,
    We don’t explicitly set a trading timeframe for the model – it adapts as part of the training process. There is, however, an implicit timeframe “guide” which is set by the drawdown cost parameter fed into every historic decision. A higher cost parameter will have the tendency to create a longer term trade bias.
    We originally set this cost parameter to 1% and have since increased it to 2%, this will have the effect of creating a longer term bias for the models.
    We have selected these parameters to reflect the relatively high transaction cost base of trading cryptocurrencies relative to conventional markets.

Leave a Reply

Your email address will not be published. Required fields are marked *