An interesting tool - StrategyDesk from Ameritrade

[dwperley]

Originally Posted by JK100

i believe that on strong bull markets you do better with buy&hold, but you have no insurance against a crash or volatility.

what are the risks of actually trading with these rules with strategy desk?

any thoughts?

JK

You may be right that a Buy & Hold strategy may fare better than the 3 Tools strategy in a bull market. I also wholeheartedly agree that the Rule #1 3 Tools strategy will save you from major losses in a crash or period of extreme volatility. My million dollar questoin is "Is the cost of the loss insurance provided by the 3 Tools strategy too high?". My one year backtesting on a handfull of stocks show that the Buy & Hold gain was dramatically higher than the Rule #1 3 Tools strategy for the same stock during the same period. I am just wondering if there is a better way to practice Rule #1 (do not lose money) without giving up so much upside. For example, maybe a strategy using the 3 Tools to get in and then use a 10% trailing stop for the exit.

To answer your second question, "what are the risks of actually trading using StrategyDesk?" As far as risks if the tool works perfectly (key word perfectly) the risks of trading with the tool are the same as trading manually. I personallly would be a little leary of letting a tool do my trading for me. If it were me I would start by using the alerts (color bars, e-mails, or sound) and manually trading based on those alerts. Once I was comfortable with that method I might consider automatic trading perhaps on a small amount at first. As a somewhat side note the StrategyDesk tool does not allow the user to reinvest the entire proceeds from the previous trade as I mentioned in an earlier post on this thread.

[JK100]

well, frankly, i don't know anything anymore. i'm reading jim cramer's book now. he says you are plain dumb to buy and hold. technical trading seems off. people here hold aapl, which is way way off what phil was talking about.
so what to do?
give my broker a call?

JK

[jasin]

A side from the R1 formula posted here, does anyone have any worthwhile formulas they have been experimenting with?

I've been writing tons of formulas, then screening out stocks that adhere relatively well, and then going and trying to plug leaks.

I can tell you there is no single formula that is going to give you the proper signals most of the time for a large number of stocks. Every equity has its own unique personalities reflected depending on its volatility, news movement, institutional holding, etc.

Right now I don't have anything I would really consider a solid formula that you can take to the bank for any one stock. So many indicators and situations need to be coded for any single stock.

But, if you have some decent formulas I'd be interested in running down some numbers and seeing how well it can be tweaked. If you have any shorter term plays you are looking at, let me know and I'll see what I can come up with. Who knows, it might be more predictable (in terms of the indicator logic) than the ones I've been trying to dial in.

[tstpilot00101]

Limitation with StrategyDesk:

I discovered what I consider to be a limitation with Strategydesk. The backtesting results are NOT what would have happened in the past. Here's the problem:

Under the Strategy Setup box in the Entry/Exit section there are drop down boxes labled as "Action". This is where you tell Strategy Desk to buy or sell. However you can ONLY specify to buy/sell at the HIGH, LOW, OPEN or CLOSE. That's the problem. Here's an example:

I set up an alert to BUY 100 shares of ABC when it crosses a moving average, and it fires at say 11am at $50 (so I buy 100 shares at 50). At the end of the day I can copy/paste that same formula into a backtest for the same day, but because in the backtest setup window, I have to specify to buy at the OPEN, CLOSE, HIGH or LOW, let's say I select buy at OPEN. Remember, the "real life" price crossed the moving average at 11am at $50. However, the backtesting result will show a BUY at the OPENING price (which can be much lower) at 9:30am!!

Therefore the backtesting does NOT reflect "real life" in the past. Has anyone else had problems with this?

I called Strategy Desk and the guy on the phone knew exactly what I was talking about, but all he could say was "Gee, you're right. We'll have to look into that".

The point of this elongated post is this: The backtesting CANNOT show the ACTUAL TIME OF DAY (along with that price) a certain formula would have actually fired. It can ONLY show the OPENING, CLOSING, HIGH or LOW price that IT WOULD HAVE USED in the past... not the actual price. BIG PROBLEM. I'd love to hear other thoughts on this matter.

Thanks!

--Tim.

[dwperley]

Originally Posted by tstpilot00101

Limitation with StrategyDesk:

I discovered what I consider to be a limitation with Strategydesk. The backtesting results are NOT what would have happened in the past. Here's the problem:

Under the Strategy Setup box in the Entry/Exit section there are drop down boxes labled as "Action". This is where you tell Strategy Desk to buy or sell. However you can ONLY specify to buy/sell at the HIGH, LOW, OPEN or CLOSE. That's the problem. Here's an example:

I set up an alert to BUY 100 shares of ABC when it crosses a moving average, and it fires at say 11am at $50 (so I buy 100 shares at 50). At the end of the day I can copy/paste that same formula into a backtest for the same day, but because in the backtest setup window, I have to specify to buy at the OPEN, CLOSE, HIGH or LOW, let's say I select buy at OPEN. Remember, the "real life" price crossed the moving average at 11am at $50. However, the backtesting result will show a BUY at the OPENING price (which can be much lower) at 9:30am!!

Therefore the backtesting does NOT reflect "real life" in the past. Has anyone else had problems with this?

I called Strategy Desk and the guy on the phone knew exactly what I was talking about, but all he could say was "Gee, you're right. We'll have to look into that".

The point of this elongated post is this: The backtesting CANNOT show the ACTUAL TIME OF DAY (along with that price) a certain formula would have actually fired. It can ONLY show the OPENING, CLOSING, HIGH or LOW price that IT WOULD HAVE USED in the past... not the actual price. BIG PROBLEM. I'd love to hear other thoughts on this matter.

Thanks!

--Tim.

You may have to ask Phil Town in order to get the right answer but I believe in a real life scenario a Rule #1 investor following the 3 tools would most likely buy on the 1st trading day following a buy signal on all 3 tools. In addition a sell would occur on the 1st trading day following a sell signal on all 3 tools. I think I chose "CLOSING" which should come close to this if you place a market order just after market close (which should execute first thing next trading day). So in my understanding the StrategyDesk backtest should emulate a real life trading scenario relatively well.

In real life I can't imagine a Rule #1 investor sitting at his computer the entire trading day watching all 3 signals and then immediiately issuing a buy or sell order when the 3 tools indicate to do so.

[belgian85]

Originally Posted by tstpilot00101

Limitation with StrategyDesk:

The point of this elongated post is this: The backtesting CANNOT show the ACTUAL TIME OF DAY (along with that price) a certain formula would have actually fired. It can ONLY show the OPENING, CLOSING, HIGH or LOW price that IT WOULD HAVE USED in the past... not the actual price. BIG PROBLEM. I'd love to hear other thoughts on this matter.

Thanks!

--Tim.

The Interval default setting is set to Daily, but you can change it to 60 minutes, 30 minutes, 10, 5 &1, and also weekly and monthly. So if you set it to 30 minutes for example, it will "check" every thirty minutes.
The one caveat with this is that even though daily data(open, close) is available from 1/1/2000 to today, the "minutes" data is only available for the previous 24 months. But it's still pretty useful though.

[drace1]

I've been backtesting various formulas and strategies since early April using SD (on a daily basis; I'm retired and have the time). About a month ago something changed in their data base and all of my previous testing became invalid. I reworked my formulas but could not get similar results. Therefore I had to start all over.

I've been corresponding with the SD software team regarding a number of issues with their data and software, but since they get data from an independent source they have to approach the source to make the correction. One example of a data problem is where an opening price for the day in SD reflects the close of the previous day or early trading close (see their Alert.xml file). I suspect if SD placed a limit order based on that value it would be in error.

For the past month I have been testing real-time because elements in my formula (DI) don't work when backtesting formula to the 1 min level (activated by clicking on the Advanced Detect intra-bar signals box). My latest formula use the DI and RSI, but here too when there is a data glitch, an incorrect trigger might occur (had an RSI drop to zero the other day for no apparent reason).

They say a new version of SD is in Beta testing and will be released in a few weeks. Hope they have resolved more of their software problems. No sign the data problems will be resolved soon. Until the data is fixed I'll have to challange every result I get from this program.

IMHO it's to early to use real money to trade automatically using this software.

[drace1]

Regarding the use of Rule #1 3-Tools for trading with SD..... I would advise you set up the buy and sell formula as Studies in the Chart window. There you can see them toggle between one and zero over the past chart period. From that you will see that these tools could possibly be used as guides, but you certainly would not want to trade based upon them alone.

[jasin]

Welp, as I think I've said before, I use a Ameritrade account for my cash investment broker. I use IB for my day trading (stocks and futures).

The 3tools rule backtests extremely well in a shorter window, such as 60minute for many many stocks. You cannot apply a single rule to everything, but what I did was backtest the entire index baskets against it to come up with stocks with max expectancy and lowest drawdown in each time window. You will come up with many many stocks that it has rediculous returns for, well beyond any buy and hold.

Ameritrade is great in the sense they are very liberal with their data. IB is quite stingy, and I can only get 1 minute bar data a few days back. The good thing about IB is I can buy 300 shares of a $100 stock for $1.50, so each round trip is only $3. If you're swinging or scalping or shifting positions around a lot you can realize how fast that extra $7 adds up.

You can set the strategy to either buy/sell on bar open or close. It's up to you.

In a bull market any long position will often test positive, regardless of the strategy. What you need to do is test out of sample. For instance, if your strategy is buying, instead of shorting, you need to test it in a bear market for it to be truely viable. Otherwise, you are just trying to maximize returns by swing trading in a market moving the same way you are. I found the 3tools, in a bull market, performs exceptionally well in 60, 30 and even 15 minute windows depending on the stock.

In a bear market, which we facing currently (short term) and likely to continue into (over the long term), the 3tools does not perform so hot.

One good thing about using the strategy in the shorter windows is it often forecasts pending dangers. If you look at certain stocks that took a major gap down, often it predicted it shortly before the close of the day to sell. Doing so would save you from 5% - 15%+ losses that I have seen happen hundreds of times.

If you are really into strategy trading, I would suggest trying out Ninja Trader. I use it for day trading, but works equally as well for longer periods as well. You write custom strategies and backtest like SD, but it has advanced order management for trailing your stops, profit targets, and scaling in and out of positions. You can take it a whole lot further, but it is also more complex.

[olddix]

Originally Posted by dwperley

I have to admit the StrategyDesk UI is not the best, but the ability to create formulas provides the user with a great deal of flexiblility (e.g. changing the MACD numbers to reflect those described in Rule #1). When I first started with the tool I created 3 strategies, one for each of the Rule #1 checks and they all appeared to signal correctly when compared with the waveforms from MSN Money. I could have used the StrategyDesk waveforms but I thought MSN Money would be a better cross check. The problem I had was that when I combined all 3 checks (MA,Stochastic & MACD) in to one strategy my triggers were incorrect. At this point I thought I would give the StrategyDesk help line a call. Steve at the StrategyDesk was very helpful and after I explained exactly what I was trying to accomplish he was able to make the corrections and we had working BUY and SELL formulas. Shown below are the BUY and SELL formulas you need to create a Rule #1 Strategy:

Rule #1 3 Tools Buy Formula:
---------------------------
Bar[Close,D] > MovingAverage[MA,Close,10,0,D]
AND
Stochastic[StocK,14,5,3,D] > Stochastic[StocD,14,5,3,D]
AND
MACD[Diff,Close,8,17,9,D] > 0

Rule #1 3 Tools Sell Formula:
---------------------------
Bar[Close,D] < MovingAverage[MA,Close,10,0,D]
AND
Stochastic[StocK,14,5,3,D] < Stochastic[StocD,14,5,3,D]
AND
MACD[Diff,Close,8,17,9,D] < 0

One feature the backtesting tool does not have is the ablilty to reinvest all the proceeds from the previous trade. The tool only allows you to reinvest the same # of shares or the same dollar amount. I have requested they enhance the tool to allow reinvesting of all proceeds. I created a crude Excel spreadsheet to compute the Rule #1 trading gain and Buy&Hold gain just to compare.

I would like to compare the results I had backtesting the example on page 257 in "Rule #1". On that page they show trades of their example on Cheesecake Factory (CAKE) from 13Mar03 through 31Dec03. Could someone complete this backtest and provide their trades, their Rule #1 trade gain, and their Buy&Hold gain for CAKE during that time period? Let me know what you get. I would be very interested to know.

HI all,

This is my first post, so be gentle with me. I found this forum because I was googling StrategyDesk Formula's. I'm looking for a trailing stop that works. But in the mean time, I want to post both of my buy/Entry and sell/Exit formulas that I have been using since July 07 with good success.

Formula 1:

Entry:

MACD[MACD,Close,12,26,9,D] > MACD[Signal,Close,12,26,9,D] AND DirectionalMovement[+DI,14,14,D] > DirectionalMovement[-DI,14,14,D] AND DirectionalMovement[+DI,14,14,D] > DirectionalMovement[ADX,14,14,D]

Exit:

DirectionalMovement[+DI,14,14,D] < DirectionalMovement[-DI,14,14,D] AND CCI[CCI,20,D,1] >= 100 AND CCI[CCI,20,D] < 100

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

Formula 2:

Entry:

DirectionalMovement[+DI,14,14,D] > DirectionalMovement[-DI,14,14,D] AND CCI[CCI,20,D,1] <= 0 AND CCI[CCI,20,D] > 0

Exit:

DirectionalMovement[+DI,14,14,D] < DirectionalMovement[-DI,14,14,D] AND CCI[CCI,20,D,1] >= 100 AND CCI[CCI,20,D] < 100

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

I would be curious how these formulas work for you guys with your own stock list. If anyone has had any success with trailing stops formulas, I would like to see them.

Thanks