zdo

BF, .coming over here from your other invitation post and this is the first thing I noticed... As far as pairs go you might consider substituting EurJpy for the UsdChf. EJ has plenty of bang for the buck / action, a reasonable spread, etc while UsdChf is almost perfectly (negatively) correlated with the EurUsd (so even if a setup occurs in the swissf it will follow what the EurUsd does.) hth zdo

MetalHead, I'm not familiar with zeroLag MA. If you have the Sierra code, please post it and we'll see what's going on with it. Thanks. zdo

Here is a comparison of JurikMA (cyan) and QuickMA (yellow) The arrows show spots where JMA beat QMA to the turn (Note, though, there may be instances off this chart where QMA leads JMA) Enjoy

Metal and all, see http://www.traderslaboratory.com/forums/f46/quick-ma-4504.html#post48873 hth

And here is the source in readable text Indicator // // Plots a fractional-bar Quick Moving Average // with the option to select gapless daily opens. // // copyright 2008 John McCormick jmactrader.com // // feel free to copy and use this code royalty free // as long as it retains the above acknowledge // inputs: Price(Close), Quick_Length(9), displace(0), gapless(0); // set gap to a non-zero value to skip over opening daily gaps Var: Length(maxlist(1,Quick_Length)); // GapLess Vars: // gapless O,H,L,C where O=C[1] RelO(0), RelH(0), RelL(0), RelC(0), gap(0), Accum(0), WtMean(0); if date<>date[1] then begin gap = O-C[1]; Accum = Accum+gap; end; RelO = O-Accum; RelC = C-Accum; RelH = H-Accum; RelL = L-Accum; // Gapless bars - end WtMean = (RelH+RelL+RelC)/3; // End Gapless if gapless=0 then plot1[displace](FMA_Smooth(price,length),"FMA_Quick") else plot1[displace](FMA_Smooth(RelC,length)+accum,"FMA_Quick"); Function // generates very smooth and responsive moving average // copyright 2008 John McCormick jmactrader.com // feel free to copy and use this code royalty free // as long as you don't remove the above acknowledgement // Inputs: Price(numericseries), Length(numericsimple); Vars: j(0), workinglen(maxlist(1,absvalue(Length))), peak(workinglen/3), tot(0), divisor(0); Array: val[100](0); if workinglen>100 then workinglen=100; // use larger array to handle lengths over 100 tot=0; divisor=0; for j=1 to floor(workinglen+1) begin if j<=peak then val[j]=j/peak else val[j]=(workinglen+1-j)/(workinglen+1-peak); tot=tot+price[j-1]*val[j]; divisor=divisor+val[j]; end; if divisor<>0 then FMA_smooth=tot/divisor;

Here is a 'very inexpensive' alternative to Juriks Moving Average contributed in Easy Language by John D. McCormick (TS handle = JDM, see jmactrader.com). Full thread is at https://www.tradestation.com/Discussions/Topic.aspx?Topic_ID=77583&SearchTerm=a%20better%20moving%20average&txtExactMatch= Posted with his permission. Please retain his 'copyright' info in the code. Many thanks to JDM. QUICK.ELD

Trader333 JMA‘s ( and adaptive MA’s in general) main advantage is improved smoothness without introducing significantly more lag… mostly a benefit to automated systems to produce fewer 'false' setups or triggers… JMA's weaknesses include a tendency to overshoot... For years the only use I have of JMA is that it quickly gets flat in congestion and stays more stable and flat (relative to other measures) and thus is a better indicator to build congestion bands around... yada yada A comparison is attached of 1 min YM with (c,4) settings. Red is EMA, White is JMA… up arrows show some of the extra jiggles in ema….hth. zdo

Yes Jurik MA is great! Has it ever been beaten? ...btw for a 'low cost' alternative that tracks as smoothly and only occasionally lags JMA turns by one bar - see https://www.tradestation.com/Discussions/Topic.aspx?Topic_ID=77583&SearchTerm=a%20better%20moving%20average&txtExactMatch= zdo

UrmaBlume, re: What do you mean by phase? Many thanks. zdo

darth ?And who's nitpicking? It really doesn't seem like 'self interest' has all that much fired up precedence when a programmer like AgeKay posting is actually 'outing' the inefficacies of hiring off the shelf programmers for trading applications. btw, wouldn't 'good' apply to AI too? "...just throw as many ideas at ( a lousy inappropriate, not 'good' (but still probably expensive)) AI as possible" or don't do a 'good' job with the data transforms and you can train it until the wheels come off and it will never produce consistently. AI is still just a fancy word for optimization ... on hidden, unseen parameters yes but it's still just optimization at this point... and there is such thing as 'bad' optimization...

All several others have alluded to, day to day correlations (positive or negative) are bound to volatility. Using the coefficient of correlation figure for the whole population between data streams can be misleading. When indexes are above three std dev’s, just about everything hooks up / correlates(+ or -) with indexes - as fear of getting ‘caught out’ spreads through the collective – ie absolute correlation can go very high. And below 1 std dev, even those with higher relative ‘average’ or composite correlation, tend to drop off in correlation significantly and march to their own drums. Examples are Treasuries, CL, AGs, FX, etc. See The Handbook of Portfolio Mathematics by Ralph Vince circa pg 290 zdo ‘I believe we can continue the Great Society while we continue to print more money twenty four hours a day’ Weimar Bernanke

don’t have much time here and already closed TS down but 1 use iff or related conditions to make sure each of the closeData[n] is not zero and if it is, assign an appropriate value to it ie make sure each of the data streams actually has a value at 830 etc 2 make sure your data starts soon enough to have large enough number of records by 830 and maybe use currentbar condition - this prob not an issue here but... If you need it, hopefully someone can help you over the weekend.

I make the most intelligent decision I can...and back it up with a few bucks. That's poker.

janlani, using thrunner's code you could probably do something like in inputs change myDate(nDate) to MyTime(0930), and change If date = MyDate then... to if time = MyTime and change If date >= MyDate then to if time > MyTime {and barstatus(1) = 2 {if you only need it to update at end of bar}} then... and it should take off at the opening from 'rebase' hth zdo

Apply what you've learned while we're still in a more 'volatile' period and when 'volatility' calms adapt and learn to thrive in that environment. The only traders who aren't learning are the ones who have left trading completely.

lordbinder, In case candlewhisperer never answers, it looks like (uncontained to the left and containing to the right) large volume large candle bodies form his zones... first big dark body forms the 3rd. ... hopefully he will explain it more accurately and in more detail... zdo

Have you been reading Larrry William's mail ??

Kiwi, while we’re on “family oriented”, after all these years you’re still always picking on us tards who ride the short bus ? Glad you got in on the fun part - sorry you don’t get it on the serious parts… Anyway – Abraham Maslow, a human potential pioneer, noted that the gifted therapist / helper Carl Rogers had reported that successful help / therapy raises the average score of a patient on the Willoughby Maturity Scale from the 25th to the 50th percentile. Maslow then asks about going from the 50th upward and concluded “You reach a point where you have to do it all by yourself”. Do you agree? And as a group, traders seem to lean toward self reliance anyway… might there only be a small range on that Scale to which it would even occur to them to ask for help. Btwn 0 to 25 - very low chance of considering the option, blind to it. Btwn 25 and 50 - only a small segment of this range would seriously consider it, 50+ - on their own anyway. Realistic? Hopefully, my posts have not derailed AbeSmith’s generous attempts to help. I appreciate it. Hopefully he is helping a private contact as we speak… "The Regular Session has opened."

This detrender tracks pretty well within fixed bounds but just for snicks I put slightly dynamic / variable ones on it… those who use the extremes might find it useful… btw daedalus the skeleton for alerts is started in the fixed band case part of the code. No warranty expressed or implied but it should help get you going… hth [LegacyColorValue = true]; {Value Chart Indicators. And yes thanks and credits to David Stendahl, et al RFA is Relative Float Axis VU is Volatility Unit } Input: FixedBand(7.5), Length(7), sdLength(21 {89} {43}), StDv(1.61), {why?} bandType(2) //bType 0 = none 1 fixed 2 bb-like 2 Keltner-like (not implemented) ; // End Input Var: RFA(0), VU(0), RfaH(0), RfaL(0), medRfa(0), Avg(0), SDev(0), UpperBand(0), LowerBand(0) ; // End Var RFA = Average(MedianPrice, Length); VU = (Average(Range, Length))* 0.2; if VU = 0 then VU = 1; RfaH = (High-RFA)/VU; RfaL = (Low-RFA)/VU; Plot1 (RfaH, "VCHigh"); Plot2 (RfaL, "VCLow"); if bandType = 0 then begin // select case noplot(3); noplot(4); end; // if bandType = 0 if bandType = 1 then begin Plot3 (FixedBand, ""); Plot4 (-FixedBand, ""); if RfaH > Plot3[0] and RfaH[1] <= Plot3[1] then Alert( "UpperValueBoundTapped") else if RfaL < Plot4[0] and RfaL[1] >= Plot4[1] then Alert( "LowerValueBoundTapped") ; end; // if bandType = 1 if bandType = 2 then begin Avg = XAverage(RfaH,sdLength); SDev = StdDev(RfaH, sdLength); UpperBand = Avg + StDv * SDev; Plot3 (UpperBand, ""); Avg = XAverage(RfaL,sdLength); SDev = StdDev(RfaL, sdLength); LowerBand = Avg - StDv * SDev; Plot4 (LowerBand, ""); { if RfaH > Plot3[0] and RfaH[1] <= Plot3[1] then Alert( "UpperValueBoundTapped") else if RfaL < Plot4[0] and RfaL[1] >= Plot4[1] then Alert( "LowerValueBoundTapped") ; } end; // if bandType = 2 //keltner-like {if bandType = 3 then begin //medRfa = (RfaH - RfaL) / 2; Shift = Factor * AvgTrueRange( atrLength ) ; UpperBand = Avg + Shift LowerBand = Avg - Shift ... }

Thank you, Abe. That is a very generous and considerate offer and I may actually take you up on it. My first problem is that I’m having a problem with the problem orientation and I seriously don’t want to have a problem with problems anymore. Can you help me / us with that? "The regular session has ended…"

Isn't the win/loss ratio system / edge specific? Certain systems do just fine at 30% W/L Others need > 50%, etc Finding just any old edge will not work. The type of system one can 'manage' / 'stick to' / 'implement the appropriate mentality' is determined long before he or she begins trading...One must find an edge that really fits him or her individually - at ‘true nature’ level. That is why I recommend not separating their discovery and development at all… not making one more important than the other because it seems first temporally or in priority. grasshopper had to grab that pot with both forearms...

On one level - yes. If you are my psychic friend why haven’t you called me? … especially after I had that premonition you would be calling with help? …and even more especially since you already know what I need to change. On another level - no. If you are my psychological friend…and even more especially since you already know what I need to change…

Please help. What are my issues ?

Yep, any and all trading systems are profitable - if you sell enough copies...

Darth, et al, Correct me if I’m wrong, but I think the original post / question was making the differentiation on time frame / holding period. For example, for long time frame position trading I do seasonals and I always go home net the core position (unless I’m being stupid or get an extremely lucky intraday parabolic move). But intraday, when conditions are right, I may lift or offset part of a position to hopefully build a better cushion for the core seasonal trade. (In fact, needing to build the cushions and needing to rescue some positions was why I ‘learned’ shorter time frame trading in the first place. It takes less than zero brains to be a seasonal trader... just put on the position on the right date, then put the monkey in restraints until it's time to take the position off...) With a ‘portfolio’ of multiple systems (correlated or not), the issue is the sizing of each individual system – which I think is beyond the scope of this topic. Btw, does anyone have any articles or references on sizing in a ‘portfolio’ of strategies – I haven’t seen much on that topic… If anyone has anything on it please open a thread. Thanks. Have a great weekend all