PPoS Improvement Ideas - Brainstorming

[Reaper]

I think this recent positive period has highlighted a very pertinent question. How do we book profits? We have a trend following EA and some of us don't want to limit the potential of trades by getting out early, so we have set the TP to unlimited and only exit when the market turns against us. But is this really the right way to go?

If we trade 28 pairs we might have to get used to large floating balances, but do we really want 5000, 6000, 7000+ pips floating? We have options here...


1. TP On (e.g. 600 pips / 4x ATR)

2. TP Off (Exit only when the market turns using your JS/TS/SL)

3. TP Off + Basket Manager with % target.

4. TP On + Basket Manager with % target.

5. Another way?


If anyone has any thoughts about this please get involved.

[barry]

Good topic for discussion.

Another option that appeals to me is to have a trailing stop on the floating pips, as a basket. Something like this: once the floating pips exceed 2,000, set a trailing stop of 1,000: so, at +2,000, close all if it falls to 1,000, and so on as the float increases.

[The_Snowman]

Just an idea, how about using the last four complete weeks for the month values instead of the last calendar month. The thought behind this is to use values of a consistent age, currently the oldest values being used range from approximately 4 weeks to approximately 8 weeks.

Nobody responded yet that I can see; I know basically what you mean. Looking at EUR-JPY a lot of pips are missed because of that small UP month eliminating it from trading. And it is still eliminated this week and next 2 weeks also; possibly many many pips down to go.

But for that to be programmed into the dashboard / EA, it has to be defined exactly.

If you just replace both the 2 conditions "Previous Month and Previous Week" with only 1 condition "Last 4 Weeks" then I think it becomes a different system.

I think it needs to be the "Previous 4 Weeks Prior to Last Week" and "Last Week" which basically maintains the 2 conditions, and covers the previous 5 weeks data, e.g. the most relevant trend.

Looking at the chart posted makes it easier to understand. And you can see, the EUR-JPY now becomes tradable, today may even be an UP Day giving a entry point this week.

That's my bit of brainstorming. I'm looking for manual trades so I'll be watching this price action.

Regards, John

[The_Snowman]

Here is another example, the AUD-USD also had that short lived up month of February but as we know, the fundamental view is short this pair.

Looking at the data as per the last post, this is eligible for shorting at daily lows as per the original Post #1

[brt]

I think this recent positive period has highlighted a very pertinent question. How do we book profits? We have a trend following EA and some of us don't want to limit the potential of trades by getting out early, so we have set the TP to unlimited and only exit when the market turns against us. But is this really the right way to go?

If we trade 28 pairs we might have to get used to large floating balances, but do we really want 5000, 6000, 7000+ pips floating? We have options here...


1. TP On (e.g. 600 pips / 4x ATR)

2. TP Off (Exit only when the market turns using your JS/TS/SL)

3. TP Off + Basket Manager with % target.

4. TP On + Basket Manager with % target.

5. Another way?


If anyone has any thoughts about this please get involved.

I think #2 makes the most sense. It has worked like a charm for xfactornos settings on the original.

[JohnnyAmpere]

Not sure if this is the exact right thread for it, but because the concept behind it is so simple I was thinking on how to get more trades out of the strategy. I know some have tried it on the lower time frames 30m,15m and 5m or something, but has anyone tried it with the daily, 4h and 1h timeframes?

If you look at the chosen time periods they are quite in proportion to the monthly, weekly and daily time periods.

Monthly: ~20-25 trading days
Weekly: 5 trading days

Daily: 24 trading hours
4h: 4 trading hours

It would allow the EA to open more trades and take smaller profits, but you would be a bit more vulnerable to news events because of smaller stops. You can still disable Monday as a trading day to see how the market develops.

I think that with the amount of trades that will be opened could cover the occasional loss from a news event and the relatively higher commission costs per trade. Just my 2 cents.

[patrick93]

Just an idea, how about using the last four complete weeks for the month values instead of the last calendar month. The thought behind this is to use values of a consistent age, currently the oldest values being used range from approximately 4 weeks to approximately 8 weeks.

Sadly, Steve didn't react to platemans suggestion. I think that's a very good idea to use the 4 latest weekly bars instead of outdated data, especially when we're at the end of a month.

PP and PPoS are amazing EA's and we shouldn't optimize it to death, like the first HGB, but this is an optimization worth thinking about.

If you think that I am, like Steve would say, a dimwit or a cretin, please enlighten me why this is a bad idea.

[Reaper]

It's not a bad idea but like you said we don't want optimize it to death or change too far from the original simple logic. I understand the premise of what you and others are suggesting. The idea did cross my mind awhile back too. But personally I think we have more than enough options for the moment to test and tweak. This EA was designed as a day strategy. So to get results of what is and what isn't working already, we need more time in my opinion.

That being said, it seems this idea is in demand and it could be worth creating experimental EA for as you're now the third person to talk about it. Perhaps Steve or some other kind coder would have a go, but as always, it's up to them to come forward and dedicate their time. Don't expect it to happen. Tips of course may speed things up and I'm sure they'll appreciate it.

:hi:

[plateman]

Of course we should remember that PC had successfully traded this system manually for a long time before making the EA available to us all. When I posted the suggestion about the last four weeks I was really hoping that PC would say either that he had already tried this unsuccessfully or pass an opinion on it.
Personally I prefer the H4 timeframe and I have been running PPOS on W1, D1, H4 for the last two weeks. This means that the data is between 7 and 14 days old. So far results have been good, last week when the markets were trending + 4%, this week held its own until last night when it lost about 2% but has recovered about 0.8% since then. I'm really testing to see how it preserves my deposit in ranging and violently erratic conditions.

[The_Snowman]

Of course we should remember that PC had successfully traded this system manually for a long time before making the EA available to us all. When I posted the suggestion about the last four weeks I was really hoping that PC would say either that he had already tried this unsuccessfully or pass an opinion on it.
Personally I prefer the H4 timeframe and I have been running PPOS on W1, D1, H4 for the last two weeks. This means that the data is between 7 and 14 days old. So far results have been good, last week when the markets were trending + 4%, this week held its own until last night when it lost about 2% but has recovered about 0.8% since then. I'm really testing to see how it preserves my deposit in ranging and violently erratic conditions.

I might shift to that setup next week (W1, D1, H4) as got nothing this week as everything is counter trending, thanks FMOC

As for the 4 weeks suggestion, I don't quite get it as I drew on my earlier charts. The original idea always looked at a minimum of 5 weeks data e.g. during the 2nd week of a month you would look back and see the Month (up or down) and the previous Week (up or down)

After that, as plateman stated, it goes out to 8 weeks, maybe 9 sometimes.

Basically, now when I manually scan the charts, on the Weekly, I'm asking myself - is the close of last week below / above the open of 5 weeks ago? If yes, there is the trend direction.