sinecure
324
sinecure  

More on exits

Thanks Derk,

It seems to me that both the entries and exits are coming late, but one may be a result of the other. Anyway, here are the settings I was using. Only the trail is following the opposite PAC, so I don't see why exits should come so late.

Exits are probably the hardest thing to deal with in EAs, and unfortunately Malone does not help much here. He always says "consider an exit when..." without giving a firm and hard rule, and it may be difficult to formulate such a thing.

I will give it a bit more thought and see what I can come up with.

autumn

Files:
sinecure
324
sinecure  

Log File

Derk,

Perhaps the attached log file will help you identify some of the problems

autumn

Files:
Derk Wehler
345
Derk Wehler  

Autumn:

Yes, I suppose one could say the entries are a bit late, but that is part of the nature of the system: it waits until several things line up to signal (notably the TDI), and that helps to avoid false signals, but of course comes with the price of being delayed.

That is why I focus on the GBPJPY; there are so many pips in the moves that even a "late" signal can pretty consistently get 50-100 pips.

Part of me has wanted to give up on this whole EA/system for a while but, perhaps because I have put a lot of work into it, I am hesitant to do so. Some part of me feels there is a way to get pretty regular profits out if it --maybe even just with the present feature set, but I have to find the right settings. (Of course I also have to get some of the irritating bugs out, but I assume I will find them all eventually).

There are lots of runs that are good for three or four hundred pips, but by the time it reverses to the opp PAC, it has given 200 back, and that is not acceptable in light of the fact that some trades will just be straight-up losers. So I figure either we need to find a much better exit that improves it a great deal, or perhaps go back to what I am trying now (take 50-100 and just quit, hoping that it will make the W/L percentage much higher).

Thank you for your continued consideration.

-Derk

autumnleaves:
Thanks Derk,

It seems to me that both the entries and exits are coming late, but one may be a result of the other. Anyway, here are the settings I was using. Only the trail is following the opposite PAC, so I don't see why exits should come so late.

Exits are probably the hardest thing to deal with in EAs, and unfortunately Malone does not help much here. He always says "consider an exit when..." without giving a firm and hard rule, and it may be difficult to formulate such a thing.

I will give it a bit more thought and see what I can come up with.

autumn
sinecure
324
sinecure  

Exits and entries

I know the feeling, as I have stared long and hard at charts and often wondered if anything solid would ever come of it. Here are a few thoughts on possible modifications to the EA.

Synergy at its core is an RSI on which 3 moving averages are applied, and a the HA (really a 1 period moving average of typical price) plus PAC. Trades occur on an HA breakout from the PAC, confirmed by alignment of the 3 moving averages. Nothing theoretically wrong with it, and it stands up to visual scrutiny.

Attached is a chart showing HA, PAC, plus an Alligator. At the bottom, a 13 period RSI and 2 moving averages (red is slow-2, blue is fast-7) on it, the same as Synergy. Also on the chart are 13 period Force (with 3 and 13 period EMAs), Stochastics (13 period+5period MA, 12 period smoothing), Standard Deviation 8 period, and DeMarker (13 period, plus 8 and 3 period SMMAs). I have used SMMA on the RSI because I like to cut out some of the chop whenever possible; that is the idea behind the MAs.

Note that the period and type of the MAs can be changed at will. For optimization you would want to be able to vary these parameters, especially for the RSI, as the 2-7 combo is not necessarily ideal for all conditions or pairs. Note also that Stochastic provides an added smoothing that puts the crossovers and changes of direction in close relation to what is happening with the Alligator or HA-PAC indicator.

Standard Deviation can be used to confirm an entry (when STDev rising and above a certain value or in a certain range of values). Force is also a potentially good confirmation because it incorporates volume. DeMarker behaves similarly to the Stochastics, and because it is not collinear could also provide a good confirmation. Any of the indicators could be used to restrict entries to overbought/oversold conditions by setting a max or min level, e.g., only enter short when Stochastic above 60.

If further confirmation is required, a 3rd moving average can be added to any of the oscillator MAs, as Malone has done with the RSI. I feel that this is not necessary, however. I would also remove the 50 condition, as it may eliminate entries in overbought/oversold conditions.

Exits can be controlled by noting change of direction of the blue line in one or more oscillators, and confirming with a downturn in the Standard Deviation.

Although I know that people like HA for its visual appeal, I tend to prefer Alligator using EMAs because the fast/slow crossover is very responsive. There is a danger that wiggle could trigger trades that are not wanted, however.

All moving average periods are built on Fibonacci numbers, i.e., on the sequence 2,3,5,8,13,21,34,55. This provides a natural order of magnitude in intervals between MA periods.

A system built on, say 3 or 4 entry signals and 2 or 3 exit signals, with parameter optimization (not just ON/OFF settings) could be both backtested and forward tested quickly and efficiently.

autumn

Files:
sinecure
324
sinecure  

One more thought

Ideally, you would not want to use either the Alligator or the HA-PAC system, and trade only on the oscillators.

autumn

Derk Wehler
345
Derk Wehler  
autumnleaves:
Derk,

Perhaps the attached log file will help you identify some of the problems

autumn

Thanks. Mostly "Invalid Price" errors, which seems to be what a lot are getting. I'll track it down....

-Derk

cochran1
97
cochran1  

Another Observation:

One thing I have noticed is that if the green RSI crosses the red MA and there is a 'large' distance for the green to cover before it crosses the yellow, a bad trade may occur. Large would be 20 units or so. Also seems like if the red and yellow are very close (almost overlayed) and the green crosses through both at nearly the same time, a good trade is likely. Not sure how consistant this is, just something to watch for and possibly make it a filter. thx mike

sinecure
324
sinecure  

Another option

An effective way to control exits is with Parabolic SAR, which is also a good entry signal and can provide a trailing stop. Exits occur when PSAR reverses on the price line, and longs/short enter when the PSAR is below/above the price line. A PSAR trailing stop by definition ratchets closer to the line as a trade goes further into profit.

autumn

Files:
sinecure
324
sinecure  

Green Red and Yellow

Hi Mike,

Very astute observation. It would be interesting to see if statistics could be developed to support the case. I expect that the situation corresponds to a steep movement of price line, but not of the whipsaw sort.

autumn

cochran1
97
cochran1  

Yes I agree. When this happens the price action is steep and by the time the entry signal is reached the currency has already moved too far and lost momentum. Another thing I look for is if I have an entry signal say on the hourly chart at the US open and the currency moved 100+ pips (depending on the currency) just prior during the London session, I don't get in the market. After this much movement the currency has usually lost its momentum. Each currency will only move so much a day under normal conditions. Looking at flat or overlayed yellow and red says that the market has stalled and is then ready for a breakout. thx mike