Yes, each stock has its own parameters. But, before I consider a stock it has to perform well using a set of standard parameters. Briefly, based on the standard parameters, it needs a Win% around 50%, less than 10% drawdown, a profit factor of 2:1 or better, and stable performance over the test period that runs about 2 years.
The optimizations run in a fairly tight range, for example, I use a parabolic exit and restrict the number of ATRs to 5-8, typically using 6.
What I look for is mainly very stable performance. A quick way I check is to look for a more or less straight line equity curve. However, I do use statistics to compare the most recent 25 trades with the older trades. The most recent 25 trades must be in the Z=-+1.2 to <1.2> range.
The formula for this is:
X1-avg of "oldest trades
X2-avg of most recent 25 trades
Var1-variance of oldest trades
Var2-variance of most recent 25 trades
N1-number of old trades
N2-number of most recent trades, usually 25
Z = (X2-X1) / squareroot( var1/n1 + var2/n2 )
I don't worry much if Z>1.2 as that shows the performance is improving.
Thanks for wishing me luck.
Overnight holds make a big difference. Typically, I hold 25-35 positions overnight, with both longs and shorts. For example, I had 34 open positions going into today's open with 14 shorts. I don't try to balance net long/ short exposure, although I plan to balance out using ETFs in the future.
DS