Satuki @ Home

Here are the final results of the trading automation experiment I conducted about 3 weeks ago.  The two stocks I chose for the experiment is FAZ and SRS.  Both are leveraged ETFs.    FAZ ended up making 30% on the first try.  However, SRS ended up having 5 losers in a row.  So the net is up 2.5%.   Here is the whole series 1 2 3 4 5

Here is what I learned so far from this experiment.  100% machine trading is probably not feasible.  If I had turned on the machine when FAZ and SRS were close to 100, then I would have lost my shirt trading this reversal pattern.  Nevertheless it does not mean that this reversal pattern is bad or not even tradable.  We need to find out WHEN we can turn on the machine. Preferably, it should be close to when a true reversal will happen.  It does not have to be very precious.  Using this experiment as an example, if a true reversal started now and the machine opened a trade, it would hit the 25% target.  This would cancel out all the 5 previous losers, which would result in a break-even in SRS.   That would leave us with a net 30% profit in the FAZ position.  So it is my next major project to find out when I can start to automate the trading of this reversal pattern.

Here is SRS’s chart with 5 losers on it.   Click on the image to have a better view

Machines can beat a lot of traders out there since machines do not have 4 very lethal problems that human traders have

1:  hold onto losers
2:  cut winners
3:  fear
4:  greed

However machines are susceptible to the following.

If my machine is based on trend following, it might be whipsawed to death without even knowing it being in a range-bound market.  Because it has no fear, it will keep taking fake signals.

If my machine is range-bound based (mostly using some type of oscillators), it might trade against a trend when it keeps shorting a overbought signal in a bull run. Because it has no fear, it will keep doing that until it has no money left.