A Consolidation Breakout/Breakdown strategy was the topic of today's chat. The main idea behind this strategy is that it looks for periods of relative consolidation, then looks for prices to break out of its Bollinger Band range. This idea presents a two-step scan that searches for opportunities for instruments that "decide" a price movement direction. When using this strategy, you should also consider other market factors, such as the broad market direction.
After the chat, I was able to tweak the parameters to set up signals to my liking for the example that was dicussed. It's always important to run Raw Mode back-tests so that you can tune your scans with the correct parameters that meet your criteria. I've attached the long and short strategies below. Simply right-click and "Save As" to your computer, and "Import" them right into the Wizard Lab tab of Prodigio.
Great trading, and see you next week!
| Attachment | Size |
|---|---|
| Cons. Breakout (Long).wiz | 2.25 KB |
| Cons. Breakdown (Short).wiz | 2.27 KB |



charles garner on Sun, 09/18/2011 - 17:00
Title: Back testing engine
I tried to get in contact with Mike at Prodigio chat board on the following date 04/28/2011 . Still no contact from Mike. Why am I having so much trouble to get someone to identify and verify this problem. I feel if I could get someone on the phone this problem could be addressed and corrected.
I talked to berry on this very subject and he also informed me that yes there was a problem. Stilll nothing has been accomplished. He also informed me that he sent material on up the chain of command about this problem
There is no use going any further in describing exactly what the problem is until I can get someone's attention.
Chuck
Hi Chuck, if you give me a brief description of the problem I might be able to give some insight. Thanks, Bryan.
My understanding is that when you backtest in paper money, the engine picks up a signal-according to your strategy. Next day-you have an order. What actually happens--sometimes the signal and the order are at the same bar on the same day. I ran a backtesting scan today. Out of 24 symbols, 2 were correct. By correct, I mean the signal was picked up at the correct time. For the remainder of symbols, the signal was not picked up due to strategy criteria-but the order came up the next day. Let me know if you need further info or screen shots. The above examples came from chart observations. Please review and let me know what you think. Thanks, Chuck
Hi Chuck, yes please attach some screenshots and a copy of the exported strategy to the forum and let me know the title and I can take a look at it.
Bryan,
As you requested, have attached the following examples:
Symbols FLIR, LRCX. Signal and order execute on same bar. Bar to the left of white arrow should have been the signal bar, but it did not meet the rule criteria.
Symbol CHKP. Does not give a signal, but, does have an order. The backtesting engine picked a bar that does not meet the scanning rule. There are multiple examples like this one if you need to see more.
Symbol MXIM. Meets strategy criteria. Signal bar - 9/12. Yellow line goes thru the bar where signal hits. Order bar - 9/13 (white arrow)
According to explanations from the Prodigio Reps-the MXIM example is how it is supposed to work.
Also attatched the rule, strategy scanning rule and symbol list from backtesting scan. All was done in paper money mode.
Chuck
Hi Chuck, I ran your example and I think it simply needs some explanation. The first thing to be aware of is that the charting can be a bit deceptive in this version. Usually, the arrow matches the open time of the position, but in this example it's tacking itself to the signal generated by the daily bar (this is one period to the left than usual). We began a project that dramatically improves charting so that it's much more intuitive; we're in the middle of it but switched to a new project as some important business-requirements came in.
That said, it's important to do two things: first is always set the parameters of the chart study to be the same as the WL rule that you're running; in these examples the rule uses a 14-period Bollinger Band but the charts are displaying 20-period Bands. This is a common problem and the new chart build handles this automatically. Second is to always prefer checking the calculation numbers instead of the lines drawn on the chart. If you check the examples on FLIR, LRCX, CHKP, and MXIM, the close price is greater than the 14-period upper Band in each case. But look closely at MXIM on 9/12 (after setting the Band period to 14). Here, the chart draws the line cutting into the OHLC bar, but if you look at the actual values, the close price 24.00 is indeed above the upper band of 23.81, so the signal is correct. The band value draws its "dot" in the center of the bar, whereas the close price "dot" is drawn to the right of the bar. This is confusing but is unfortunately characteristic in just about every charting package. Keep in mind that the chart is a module that is simply a "view" of the underlying workings of the system (which is mostly server-based scanning). The system is producing accurate values/signals but we plan on improving on our displays, etc.
Thanks, Bryan
Very Interesting to say the least.
Chuck
2 Questions
It seems like the ATR itself should be above the ATR moving average for the signal? It seems like that would indicate a spike in short term volatility.
The strategy gives good signals but the whole thing seems backwards to me. I have watched the video twice unfortunately the sound seems to go out occasionally and I miss some important points. (fyi the sound loss is a common theme in the archives)
Can you briefly explain your use of the standard deviation?
Keep up the great work
Stefan
Hi Stefan, sometimes I try to mix up the TA oscillators when trying to measure different similar concepts. I try to do this for demonstrative purposes. I like using Standard Deviation because it categorizes the data into ranges that are somewhat like the bell curve (a "Normal Distribution"). In other words, even though the behavior of the market isn't perfectly "Normal", you can consider it a good rule of thumb that 66.7% of the values are within 1 standard deviation, 95% of the values are within 2 standard deviations, and 99.7% of the values are within 3 standard deviations. This behavior is slightly different than the way the ATRs behave, but both can be used effectively to measure volatility spikes.
When using the ATR, you can certainly compare it against it's average to look for spikes; it's a very sensible measurement. We actually have an out-of-box study called AATR, which is simply the moving average of ATR. IMO, This technique in general is a good way to look for spikes in data.
HI Bryan,
I am very encouraged to use LR Slope in my strategies because of your recommendations in video chats, but I have had a serious problem with it. I called support for answers but did not find out what I need to know.
My question is, if a a flat line has 0 LR slope, and the max it appears is 1 = 45 degree angle. What LR slope values should I expect from 45-90 degree angle from the Prodigio's implementation?
Sorry for distraction. May be I can benefit from your help here.
Best Regards
Amir