>The other thought would be to put in a trailing >stop at the target rather than a limit. If the stock >wants to run 5 points, let it. One of the variations to the original system that I coded did just that. There was a slight improvement, but not nearly enough to make it even close to profitable. IMO it certainly can only make sense to include that in any such algorithm. You would be exposing yourself to additional risks such as halts etc, but I would bet that the testing would show a pretty consistent though not large improvement. JB
other thoughts... - What about starting with a small base size and then adding once it appears you are on the correct side of the trade with a breakeven stop? - Wait until the stock is at the lowest range in the past x days. Start your program the next day. - Alternatively (or additionally), just skip any day where the previous day had a range greater than the last x days. (Mixed feelings on this one) - Only take trades in the direction of underlying trend. (Don't reverse.. just get out until another cross.) - To minimize chance of catastrophe you have to constrain the number of flips (which hurts the system). But with a few filters it seems like you could increase your chances of volatility occurring and overcome that limitation. There's an equation here.
A lot of good ideas Fire. >To minimize chance of catastrophe you >have to constrain the number of flips Apparently, abogdan doesn't agree with us on this on since his lastest system posted contains no such limit. I will code his latest posted system up when I get time, but without this limit I already am confident of the results. JB
i dont think abogdan's original intent was to post a complete system, it was just to make a point and generate some discussion. "anal"ysis. heh heh
>dont think abogdan's original intent was >to post a complete system, it was just to >make a point and generate some discussion. You know, that's what I thought too and as you can see I posted my initial gut thoughts regarding his original posted system. As I have said repeatedly, I have no problem with a partial system being posted. Just don't respond to the tester like he is a moron and an ass when the numbers on the system turn out horrible. Just say "Yep, that system is a loser and I have rules that turn it into a winner" And no, he doesn't owe us those rules. JB
abogdan, when would you stop trading a symbol? and could you give us a real example of a trade you did? the best would be a trade with some flips. thanks
Dear ET members who are still following this thread. In order to prove my point I had to do something that I normally hate doing. I used TradeStation to write a system that is based on the conceptual model that described at the beginning of this thread. As you know, TradeStation has a lot of inherent limitations such as: #1 It does not allow optimization on Bid/Ask BidSize/AskSize data #2 It does not have an ability to program some of the best digital filtering technics like FIR or IIR filters. # 3 It does not allow multiple entries/exits on the same bar. #4 It does not have a capability of having multiple stocks positions etc. Knowing all these limitations I still wanted to illustrate my concept. So instead of delta I used the length of 15 min price bars, I also used "0" line that is straight and not adjusted by Sizes. I had to omit a lot of other technics that normally work well for me due to TradeStation limitations. With all of this in mind please take look at the following Easy Language code that I slapped together strictly for illustration purposes. To test the system please follow the following instructions: Copy and Paste the code in a new strategy window. Inputs: Cnannel(0.03), ShareMultiplier(1.45), GuaranteedSwing(2.6); Vars: OpenPrice(0), DefaultShares(1000), Flip(0), MaxProfit(0), Done(0); If (date[0] <> date[1]) or (OpenPrice = 0) then begin OpenPrice = Open; Done = 0; Flip = 0; End; If Close > (OpenPrice + Cnannel) And Done = 0 And (MarketPosition = 0 or MarketPosition = -1) Then Begin Buy DefaultShares*(Power(ShareMultiplier,Flip)) Shares this bar on close; Flip = Flip + 1; End; If Close < (OpenPrice - Cnannel)And Done = 0 And (MarketPosition = 0 or MarketPosition = 1) Then Begin SellShort DefaultShares*(Power(ShareMultiplier,Flip)) Shares this bar on close; Flip = Flip + 1; End; If MaxProfit < OpenPositionProfit Then Begin MaxProfit = OpenPositionProfit; End; If CurrentShares <> 0 and MaxProfit <> 0 And OpenPrice <> 0 Then Begin If (100*MaxProfit/(CurrentShares*OpenPrice)) >= GuaranteedSwing Then Begin If ((MaxProfit - OpenPositionProfit)/MaxProfit) >= 0.08 Then Begin Done = 1; Flip = 0; MaxProfit = 0; If MarketPosition = 1 Then Begin Sell this bar on close; End; If MarketPosition = -1 Then Begin BuytoCOver This bar on close; End; End; End; End; Verify the strategy. Bring up 15 min bar chart of KLAC with at least 250 days history Apply the strategy. Adjust the properties of the strategy the following way: In the strategy properties window please set Fixed Shares = 1000, maximum Shares/Contracts per position = 4000, max bars that study will reference = 2, commissions = $0.005 per share, initial capital = $100,000, do not allow multiple entries. In the inputs window use channel = 0, ShareMultiplier = 1.45, GaranteedSwing = 2.6 I'm looking for all of you who would be able to test the strategy to share the results you get with all of us who are following this thread. Mind you again, it is just a rough, crude version of what I'm actually running. Have Fun! P.S. You can, obviously modify the system to fit any stocks you like.