Bill Graves Veteran   Posts: 125 Joined: 11/25/2006 Location: Phoenix, Arizona  User Profile | Hi Kris, Good questions! When I first saw your approach using the lookback function in OL, I was concerned that a lookback of 5 or even 10 periods might not capture the retracement from the most recent Swing High down to the 20 EMA or most recent Swing Low up to the 20 EMA and that too long a lookback might cause problems as explained below. However, utilizing a 5 day lookback should be adequate for another aspect of the system: for EOD trades to check how long price stays near the 20 EMA before starting a retest of a recent Swing High or recent Swing Low. All the EOD Grail examples in Street Smarts show Price touching or penetrating the 20 EMA for only 1-3 price bars during the retracement phase before resuming in the direction of the trend. In other words, once the touch or penetration has occurred, Price will resume original direction within 1-3 periods on its way to a retest. Intraday, Price can hover around the 20 EMA for more periods before attempting a retest of recent Swing Highs or Swing Lows. However, the number of periods to get to the 20 EMA, that is a retracement from a recent Swing High [Swing Low] to the 20 EMA, can easily be 15 or more EOD and many more Intraday. I wrote to Bob McFarland about the lookback approach and asked if we needed the SDK module to employ some kind of looping function or whether we could do that in OmniLanguage. In the 1 to 5 minute charts, Price can take 20 to 40 periods to retrace to the 20 EMA and another 10-20 periods to hover above and below the 20 EMA before resuming a test of a recent high or low. Please keep in mind that I am not a programmer so I probably will not be able to explain this in programmer's terms. Anyway I had suggested to Bob that, when Price made a new Swing High [Long trade], to begin some kind of looping function down to the 20 EMA. Again, this could take 10 to 15 or even more periods Intraday to retrace to the 20 EMA. Then when Price touched or penetrated the 20 EMA, begin another loop as long as Price was below the high of the bar preceding the "trigger bar", as defined in my previous post #12176. The second loop would probably not be needed EOD because Price usually starts to its retracement right away. A 5 day lookback should work here. Bob thought we could do it in OL. We have not done anything on it though. We are working on the Headley Acceleration Band project. We are doing a premarket scan, which includes fundamentals. We have completed the entry portion of the system. It has turned up some extraordinary winners both on the long and short side. I have been trading them manually with good success. We are at the exit phase of strategy. When finished and back tested, we plan to turn it over to the membership. It is possible Bob may be willing to look at the Grail Trade some time in the future; though I know he has many programming projects on his plate at this time. Therefore, I cannot speak for him. No pressure here, Bob! Truly! -:) Kris, maybe you want to take a shot at the loop idea. You are good programmer and have contributed a lot! We don't need the 8/20 EMA filter if we are to stick to the actual published system. The rules are very simple: just as I posted them at the beginning of this thread. The EOD examples in Linda's book do show that the 20 EMA can be pointing down very slightly when price touches it. Her examples do not show a chart like ONT where there have been 3 lower lows in price after the trigger bar. About the last chart I posted, I was just thinking aloud. I am betting Linda Raschke would say the ONT chart is not a grail trade. It definitely started as one when it touched the 20 EMA Intraday, retraced, but later penetrated the 20 EMA at the close. Subsequent price action has excluded it, in my opinion. I was just looking for reasons price might retest the recent high. In my defense, I did say: ...price is not making its usual grail pattern where we start to see some support around the 20 EMA. In summary: 1. How do we determine the direction of The Grail setup? ADX shows us the strength of a trend irrespective of the direction. Long: We are looking for a retracement of price from the most recent Swing High back down to the 20 EMA, then a retest of that most recent High. Short: We are looking for a retracement of price from the most recent Swing Low back up to the 20 EMA, then a retest of that most recent Low. 2. Should we add a filter EMA(8) > EMA(21) for LONG trades, and EMA(8) < EMA(21) for short trades? You mentioned it in your recent note. I a not sure if it is discretional confirmation or a hard rule. The 8 & 21 EMA's are not a part of Linda's Grail system. Again, in my comment in the above post, I was looking for reasons for Price to start a retest of the recent high. 3. Should I remove the LookBack from the system? It is designed to tell us that the market was above EMA(20), briefly dipped below the EMA(20) (for no more than 10-bars, or whatever you set the parameter to) and the market is above the EMA(20) again. EOD a 5 or 10 period lookback from the Swing High or Swing Low to retracement to the 20 EMA will miss many Grail trades. However, EOD, for the testing of how long price hovers around the 20 EMA after touching or penetrating it, a 5 day lookback should be adequate for the classic Grail pattern. Intraday, a second loop might be considered for this time Price hovers around the 20 EMA after touching/penetrating it but before moving toward a retest. If I have not been clear , let me know. I have tried to clarify it a little in answering your next question. 4. If we remove LookBack parameter, should we have another filter, e.g. go long only if EMA(20) is trending up? on your recent picture/post EMA(20) has turned down, but you still consider it as a LONG Grail candidate. Hey, that's more than one question! -:) Above I have talked about 2 uses of the Lookback function: Use 1: The number of periods from the most recent Swing High [Swing Low] to retrace to the 20 EMA. Use 2: The number of periods Price will hover around the 20 EMA before starting a retest of a recent Swing High or Swing Low. RE: Use 1, we should find an alternative method, maybe a loop, for both EOD and Intraday. RE: Use 2, a lookback is ok EOD but we should have an alternate method Intraday. The main problem, as I see it, with the lookback approach for the phase from a Swing High/Swing Low to the 20 EMA is this: if the lookback period is too short, it will miss trades. If the lookback is too long it could pick up an additional swing High or Low, which might cause a problem trying to set up a loop. When we come to the most recent Swing High or Low, we want to check the ADX > 30 and that it is rising ~ a linear regression of the ADX at the bar of the Swing High or Low. Then we start the loop looking for a retracement to the 20 EMA. All the examples in Linda's book show that the most recent Swing High was a recent Highest High even though her rules refer to it as a Swing High. Also, "after a successful trade, The ADX must once again turn up above 30 before another retracement to the moving average can be traded." A lookback may have difficulty with this rule as well? It's ok if the 20 EMA is sloping down a little when price touches or penetrates it. Because Price has just made a recent Swing High or Swing Low, the trend is strong, so the 20 EMA will NOT be sloping down very much. You have probably already guessed by now that I would have to say the ONT chart no longer qualifies as a Grail Trade even though it may retest the recent high. What I did not say before about ONT is that this stock has some strong published fundamentals. But it is still highly volatile! Therefore, we will just have to see what next week brings. Kris, hope this helps! [Edited by Bill Graves on 6/10/2007 5:24 PM] | |
