Scalping methods, or other short-term methods, always require strict backtesting discipline because the effects of slippage are such a large % of the average profit. I've seen many a would be great emini scalp system do poorly in actual trading because the designer didn't take into account the realities of the bid/ask spread when placing profit/stop loss targets. Also, many of the best scalping opportunities are the "quick thrust" periods which often result in an extra tick slippage. There is also the usual issue of not knowing whether your profit target or stop loss were hit first if both occur on the same bar. Tick data is required in this case... or you can always assume the worse case scenario in your backtesting and see if you still have a good system (which can only be better b/c not all those exits would have been stop losses, some would've been profit targets). I haven't looked at the posted system so I can't comment. Just speaking generally.
Yup... what may appear great in theory may only be executable in a low % of situations because of the inability to get an appropriately optimal fill...
Very much the case. Interesting you should bring that up. In backtesting scalp-type trades, I always add 1.5 points for slippage and double the charges. This came from having expensive broking, but it now stands me in very good stead because if a strategy is profitable consistently with that kind of margin for error then it is more likely to work in real life, and has a good chance of being more profitable than expected. In fact i use those kinds of rigors for all my tests that i run. It probably puts some potentially profitable strategies out of the running, but you do only end up with the best of them in the long run. So it's not just me taking that kind of view... Natalie
I can't seem to get a H.G. type strategy to work especially well on daily bars. I have a rather complex version of the code, but I was playing with a "Speghetti Code Version" and I still couldn't get it to work very well on a reasonable universe of stocks. I'll post the code I whipped up so people (if inclined) can rip it apart and make improvements / suggestions. I tried a few different exits including Trailing ATR, Parabolic, and some other stuff. Here goes: inputs: ADXLength (14), CUTOFF(30), XMALEN(20); VAR:ADX14(0), XMA20(0), LongLoss (0), LongProfit (0), ShortLoss (0), ShortProfit (0); ADX14=ADX(14); XMA20=XAVERAGE(C,XMALEN); LongLoss = lowest(low,3) ; LongProfit = Highest(high,10); ShortLoss = Highest(high,3); ShortProfit = Lowest(Low,10); if (marketposition = 0) and (ADX14 > CUTOFF) and (close < xma20) and (TLSlope(Average(Close, 5)[1], 1, Average(Close, 5), 0) > 0 )THEN BEGIN Buy ("L.Smurf's H.G. LE") next bar at high stop; end; if (marketposition = 0) and (ADX14 > CUTOFF) and (close > XMA20)and (TLSlope(Average(Close, 5)[1], 1, Average(Close, 5), 0) < 0 ) then BEGIN SellShort ("L.Smurf's H.G. SE") next bar at Low stop; end; { if marketposition =1 and barssinceentry >= 1 then Sell ("L.Smurf's LX Loss") next bar at LongLoss stop ; if marketposition =1 and barssinceentry >= 1 then Sell ("L.Smurf's LE Profit") next bar at LongProfit stop ; if marketposition =-1 and barssinceentry >= 1 then BuytoCover("L.Smurf's SX Loss") next bar at ShortLoss stop ; if marketposition =-1 and barssinceentry >= 1 then BuyToCover("L.Smurf's SXProfit") next bar at ShortProfit stop ; }
I am confused also. Are we still referring to the ADX system presented by stocon to start this thread. If we are, the link to the no fear zone presented this method with a daily chart as it says a buy stop is placed above yesterday's high. Maybe I missed something.
I can't figure much from the wl script maybe pretzel could break the scalper method down a little to help clarify.