This is an ongoing problem that I have had with support. Any of my assumptions are based on tedious e-mails and conversations with SD support.

It appears that we are not receiving consistent answers.

Let's keep asking questions until our analysis verifies their answers

It's very lonely here!

--------------------------------------------------------------------------------

Using SD...there should be a blog with FYI about SD. It could be questions asked and answered by support or different ways to use SD

for example;

It is my understanding when using SD that:

1) when using Action button and selecting Buy = Open and Sell = Open SD will alert when strategy conditions are met and trigger an execution on Open of next bar

2) when backtesting and checking 1-minute bar inside the interval bar this will "closely" reflect actual results

3) or using "lookback" 1- bar will "closely" reflect actual results

4) when using Saved function will accurately reflect the actual execution prices if program was set on auto

5) backtesting results will accurately portray the drawdown amounts as if utilyzing auto

6) you cannot use Num Triggered or Last when backtesting

7) there is a maximum number of entries one can add to the custom wizard

Comments welcome

FYI

Are you aware that there is a maximum amount of custom field items that you can save in SD? I reached my maximum and had to delete a few to make room.

They will all appear in the wizard but not in the Custom Fields area.

Mattie,

Yes it is a vast topic and very dependant upon subjective conditions.That is exactly why I believe it should be tackled.

Too many traders, including experienced traders listen, pay big $ for information and never ever realize all of the emotional and psychological demands that weigh upon them as they develop and prepare every day.

I realize that it is near impossible, sans being clairvoyant, to address all of the issues, but I also know that it is possible to address some of the most pertinent points that can be developed in order to lessen the burden of excessive and sometimes unending analysis.

Most canned indicators are either regurgitations of earlier ideas or useless time wasters. In this volatile marketplace it is imperative to generate methods that will reduce entry and exit times. This is my goal.

Maybe we can find a way of having a forum after the close and invite all members to join

I enjoy this venue and will continue to post and share but being a full-time trader I am looking for a more interactive venue where the participants can develop and share ideas more spontaneously.

If anyone can recommend or are of like minds please let me know.

I have created and /or found many useful formulas, as a matter of fact,information overload. This is why I am designing my own market map utilyzing screener columns to decrease latency issues.

If a trader designs a method to generate viable trading candidates and then inputs them into Level 1 and than is able to watch their personal "market map" I believe it will aid in overcoming the fear or delay of entering and reduce need for further analysis. Thus, quicker decisions based upon each trader's personal barometer.

This exercise is based upon my hypothesis that "curbs" emotional ties to every trade and hence will allow quicker executions.

Traders must find the least amount of indicators to rely upon in order to remain competitive in this professional marketplace.

I also like to see price position to Bollinger Bands



{ % position of price between Bollinger Bands} (BollingerBands[Upper,Close,20,2,D] - BollingerBands[Lower,Close,20,2,D]) * 100

How best to use;


Bar[Second,0] >= X AND Bar[Second,0] <= Y


or is it useless?

Comments??

1YearHigh or Bar[High,W,52] or Bar[High,M,12]

Let's discuss screeners and whether they can be used in intraday trading and what appears to be the simplest and most reliable alerts in columns.

Start with a simple premise; the reason that a stock goes up is more buyers than sellers, now how can we best determine when there are more buyers than sellers (or vice versa) ? Examples: Higher highs and higher lows (the opposite for sells e.g., lowest low 30' PriceRangeChannels[Lower,30,0,1]). How can we determine whether the move "appears" sustainable? {Volume Blast} Bar[Volume,5] >= Bar[Volume,5,1] *1.50 or {Vol Pop} Bar[Volume,3] > (Bar[Volume,3,1] + Bar[Volume,3,2] + Bar[Volume,3,3]) AND Bar[Volume,3] >= 100000
I want to continue this discussion by first determining what each trader believes to be the most important component to motivate a trade and then code it for a column in screener.

Let's try to ascertain what motivates others to enter or exit the markets and develop a strategy around that premise. All comments should be welcome and attached code if possible. If not then we should try to code it and decide and discuss its' viability.

TY Tony!

To expand this conversation, if I set-up my entry to buy on ask and my exit to sell on bid this may result in more realistic results.

Also adding Bar[Second,0] >= X AND Bar[Second,0] <= Y may qualify codes to be closer to actual results.

My reasoning is; as long as trend continues your trade results will not be compromised but if trend changes within your chosen time frame it is anyone's guess to your actual entry/exit price.

I am not foolhardy to go "auto" and compete with the advanced technos who have lines directly to the various exchanges and the technology to execute at +/- 10 miliseconds. I simply want my backtesting and actual results to "closely" mirror one another, in order to eliminate some time latency.

Once I get to that point it becomes a visual exercise to corroberate alert as well as a cerebral decision whether market conditions warrant the trade.

Remember that last month 30.7% of all trades on the NYX were electronic arbitrage and probably another 20 - 30% were based on algoritimic strategies, so if anyone doesn't realize who you are trading with or against, think again. This is not a game, this is business. There are no Marquis of Queensbury rules to protect your capital, you are on your own!

Is it safe to assume (I know ass of you and me) that setting SD "Test every "1" minute bar inside the Interval bar on virtually any strategy will produce results that closely emulate actual" results?

TY in advance!

Mattie,

It makes sense and I have utilized the Saved feature and then checked it versus the backtesting.

This is why I have been so frustrated because oftentimes there is little resemblance with regards to the results.

I will continue to trade manually and work with SD and find a way to implement it with my manual/visual models.

Is it SD's position that checking the test every "n" minute and using a fast interval, e.g., 5 or 10' in interval (not checked) will be the best way of emulating actual results? That are very, very close to actual results and backtesting?

RELATIVE VOLUME


Ratio between current volume and 3-month average value OR "N" month

mattie,

do I have to manually enter the industries?

Industry or Sector group

Is there a way of coding the industry or sector of stocks in SD Screener?

for example if symbol is SHLD ...Retail-Department Stores

Mattie,


My BT results are attached already, I am using it in Saved area and comparing signals now.

I want to put a realistic system out there first and then develop comparable columns data in scanner in order that we can develop a blueprint for trading.

One post at a time in order that we may all grow and profit!

Time to share, compare and develop profitable systems together!

Here is one that I have been working, if need be let's analyze it and improve it for our beneficial use. Check to see if BT mirrors actual trades in Saved function.

216a Intraday
{Intraday entry} Bar[Close,15] > (Bar[Close,15,1]*1.01) {buy on candle close if this 15 min candle closes 1% higher than the last 15min candle }

{Intraday exit} Bar[Close,1] < (Bar[Open,D]*.98) OR Bar[Close,1] > (Bar[Open,D]*1.08) OR EntryShares * Bar[Close,D] <= (EntryShares * EntryPrice) - 1000 OR (Bar[Hour,1] * 100 + Bar[Minute,1] >= 1550) OR {True at end of period} ( (Bar[Minute,1]=14 AND Bar[Second,1]>55 AND Bar[Second,1]<=59)OR (Bar[Minute,1]=29 AND Bar[Second,1]>55 AND Bar[Second,1]<=59)OR (Bar[Minute,1]=44 AND Bar[Second,1]>55 AND Bar[Second,1]<=59)OR (Bar[Minute,1]=59 AND Bar[Second,1]>55 AND Bar[Second,1]<=59) )

May 1 -Jn 16 ,2009 P/L $38573.70 378/210 64.29 2.04( profit factor) largest loser $1285 MaxDraw -1735


with True @ end of period added same as above
P/L $38573.70 378/210 64.29 2.04( profit factor) largest loser $1285 MaxDraw -1735

I am testing this in Saved function, ultras, 500 shares.
Apr 1 -Jn 16, 2009

P/L $88238.20 798/440 64.46 2.09 ( profit factor) largest loser $1590 MaxDraw -2049

March 1 Jn 16, 2009
P/L $418737.80 1451/706 67.27 3.90( profit factor) largest loser $2735 MaxDraw -1100


No BS Only $ucce$$ !

vegas,

As indicated previously thatis exactly what I was told.

I used it in auto trading and it did not mirror back testing.i am still testing

AND ( (Bar[Minute,1]=14 AND Bar[Second,1]>55 AND Bar[Second,1]<=59)OR (Bar[Minute,1]=29 AND Bar[Second,1]>55 AND Bar[Second,1]<=59)OR (Bar[Minute,1]=44 AND Bar[Second,1]>55 AND Bar[Second,1]<=59)OR (Bar[Minute,1]=59 AND Bar[Second,1]>55 AND Bar[Second,1]<=59) )

and it appears to be an improvement, the 1 second limitation appears to be too restrictive.

Strategy Desk Support

FYI This is another answer that I received from SD.Is this the same idea as presented in post and by Tony?

Thirty minute signals tend to be more reliable than 5-minute signals because there is less noise on the higher timeframe. It also benefits us because there are fewer chances that an alert will be triggered prematurely due to intra-bar movements. Other trades are reliably getting real-time signals to match backtesting. I cannot give you any specifics due to our compliance rules, but I can tell you that crossover strategies can be more difficult to mimic because of the intra-bar action. Things you can try are pushing the formula back by 1 bar, and trying a time-based exit such as:

Bar[Second,0] >= X AND Bar[Second,0] <= Y

Replace X and Y with seconds so that if the bar lands between this range, the criteria will be true. This can be used to have alerts trigger at the beginning of the bar or end of the bar, for example.

TY all,

vegas, I will add code to strategies and see whether it may afford me closer results to backtesting.

Mattie,

This is all well and true if I was testing a daily, as you indicated but I am coded for a 10' interval.

Therefore, according to all of the diatribe with support etc.

my strategy should closely replicate the backtesting results.

It doesn't!

Another response from SD support;

With regards to your two formulas and how they would work in real-time trading:

AND {30' High + .05} Bar[Close,D] < Bar[High,30,1] - .05

We're referencing an open bar in Bar[Close,D] so this in real-time is effectively the last trade. It'll take the high from the previous 30 minute bar and compare it with the real-time price. In backtesting, this will not work properly because Bar[Close,D] will reference the closing price of the day. You'd have to try it with Bar[Close,30] instead.

AND {30' High + .05} Bar[Close,D] < Bar[High,30,2] - .05

An open bar again for Bar[Close,D] but in this one we're looking at the 30 minute high from two periods ago. Both formulas will operate on a tick by tick basis. The only difference is which bar to reference the previous high from.


Therefore, according to SD with the above instructions, I should mirror actual results.

It doesn't!

While I appreciate your response and intuitively it appears correct, this is nor the way SD support is presenting this situation.

here is one response from support;

When real-time trading the program is looking at your formula on a Tick by tick basis, waiting to see if the criteria has been met. In backtesting, it will wait until the end of of each interval to check to see if the criteria have been met. In your case every 5 minutes.

Often times a client will create a backtest that they want to replicate in real-time. We then recommend to back off your formula by one period in the "N" bars ago field. This will force the program to wait until the current bar is closed before it checks to see if the criteria has been met. I created an example below:

If the following formula was working in backing testing: RSI[RSI,14,D] < 70 AND RSI[RSI,14,D,1] >= 70

In real-time it would begin look like below, once the "N" bars ago is pushed back by one period. RSI[RSI,14,D,1] < 70 AND RSI[RSI,14,D,2] >= 70

Therefore, according to support using a lookback period should mirror realtime trading. It doesn't!

No false signals but not mirroring backtesting results.

As I indicated I communicated with SD support incessantly and I thought that I reached a comfort level but...

SD support told me that using my code set for 30' and interval "unchecked" at 10' would closely mirror actual results.

I have used various systems that I was led to believe would emulate my backtesting results with mediocre results.

I am completely befuddled whether anyone but those with the fastest state of the art technology can actually use SD auto.

we all want to believe that we have developed or found a viable model that will be our personal money tree.

All of my past trading success was accomplished using a simple approach. As I incorporated more sophisticated methods I became more of an analyst and less of a trader,

It is time to go back to what brought me to the dance originally.

I hope I have provided some insight into the importance of reality over fantasy.

I have been extremely frustrated with various developed strategies that show great profit in backtesting the "real" strategy (moving back one bar) and the backtesting strategy but when I "enter" orders and use the saved feature, there is very little resemblance.

I know that actual trading is keying off of tick data and backtesting off of last bar chosen in code but when I have spoken to SD support ad nauseum without any progress.

I still have doubts whether SD can actually be used in auto-mode.

Interesting possibilities or more puff?

http://emini-watch.com/about/

New Project

Has anyone worked on translating Tradestation code to SD code?

How close or difficult are the two formats?

Will just opened a new site StrategyStalkers.com

Mattie,

TY for your quick response

In that case I would like SD to enable us to create a new watchlist once our specific conditions are met, rather than have to manually create a new watchlist. Also this would save a lot of bandwidth.

Scanning for candidates

Good Morning!

I think I am suffering from brain freeze and I am looking to you for a way of thawing it. LOL

Although I have been trading for a loooong time I am finally introducing myself to screener.

My question is if I want to screen for specific conditions, e.g., stocks > $5 and < $20 with average 20day volume > 1 million (I know how to code this) do I load a symbol list unto Level 1 page for example DJ-30 with the code to eliminate those stocks that do not fit my parameters?


When I do it that way it still keeps all stocks on page and then highlights my candidates, is there a way of performing the scan and then only having a watchlist of my candidates?

TY,

JAC

I think I figured it out *1 at end of entry code + exit code * -1

Please confirm this is correct!

Is there a way of putting an entry and an exit in the scanner column?

The entry =1 and the exit = -1 ???

Thank you vegas,

I will go with your instructions.

Mattie,

I have two computers with screens, strategies etc and I would like to mirror both computers without having to do everything manually

SD support says to copy SD folder and then e-mailing it to myself.


Okay so far but what files do I transfer?

Here is what I did thus far;

Start>programs > AMTD >SD > properties > find target >

this is where I am not sure what to transfer, I am reluctant to move the wrong file and then lose everything.

Any advice?

Also SD is advising me to delete my screeners etc from one computer, I have different info on both computers and I want to mirror all my info on both computers.

TY in advance!

Great idea Mattie,

I have a brief shopping list but this is my major want;

I would like to see SD develop more reliable methods of getting backtesting results to be replicated in reality trading.

The problem appears to be that SD uses tick data for actual trades and trader's are deluded into thinking the BT results will be mirrored in real time trading.

Even when I create my RT code with a lookback feature, quite often the actual and test results are not reconizable.

Could someone help me code this idea for SD?


Weekly pivot= highest high of last 5 days + lowest low of last 5 days + close divided by 3

crossing its 3 ema 5ema and 13 ema

...when the weekly pivot crosses its 3ema = +1


...when the weekly pivot crosses its 3ema & 5ema = +2


...when the weekly pivot crosses its 3ema & 5ema & 13ema = +3

and minus figures for negative (crossing below) crosses?

TY in advance!

WOULD YOU CARE TO CLARIFY???