data mining, indicators vs historical data

NexusFi · 2014-07-31T12:38:57-04:00

I have been considering launching a project to back test 20-40 or so indicators/price action vs historical forex data vs time frames to try and get an idea of which combinations are best. I have had some mt4 EAs running on a vps for a while now so have some limited experience, but have only tested a few indicators/strategies. I now think the best approach is to back-test a wide range of indicators and record the outcomes. It occurred to me that probably many people have done this before and I ask the community …

Discussion in Traders Hideout

Updated July 31, 2014

data mining, indicators vs historical data

July 31st, 2014, 12:38 PM	#11 (permalink)
NoiseTrader716 Buffalo, NY Posts: 64 since Jul 2014 Thanks Given: 4 Thanks Received: 38	"PAL can be used to find patterns in intraday data after data files are properly converted to the program intraday format. Finding statistically significant patterns in very fast timeframes, like for example 1 - 10 minute bars, may turn out to be very hard or even impossible due to the presence of noise and the demand for very large samples." I think that is very misleading. The problem is not "hard", it is not something that is overcome by throwing more resources at it. This type of analysis leads to meaningless conclusions because of the non-stationarity and lack of ergodicity of market data.