Spud's Predictive MTF Stochastics

Great work! I'll give it a test.

Would that be a M5 R0 trade on the second chart even though the price line wasn't over 76.4 at the time that one would've taken the short? I was thinking a M5 R20 but please advise.

Feeding the thoughts!

WELL THIS IS A LONG POST -- PROPERLY FOR SOME DAYS MY FEW LAST ONCE !!

I thought I will share with my implementation of Spuds MTF Stochastic!

I want to warn you - this follows basically Spud's rules - but some twists by me: mainly because I like it or I do not see any other reason worth.


I share that for two reason:

1. Maybe there is something - which feeds your thoughts in implementing your design of it - or which will overall improve or simplify something.

2. Maybe I got somewhere mixed up - and you have reasonable thoughts of improvement.



OK MJ's present implementation.:

PMTF Stochastics With Time Frame Shift:

A implementation with an emergency StopLoss - BUT no Trailing - Exit is purely by Risk assessment (or Hit of EmergensyStopLoss)

1. Stoch Settings: I follow strictly Spud's Stoch settings

REASON:

a. I thought first the different settings applied to different TF-goups was a mistake - but Spud reasoned it was by design.
b. I think for the beginning it is best to use the same settings Spud proved in his own trading - if things will be successful as hoped for - I still can play with changing the settings.

# TimeFrameLevel1: TF-goup M5
FilterTimeFrame=M1 (Stoch_Filter (14,3,3))
MainTimeFrame=M5 (PriceLine (100,3,3), OriginLine (14,3,3), Stoch_Main (14,1,1))
HighTimeFrame=M15 (Stoch_HighFrame (14,3,3))

# TimeFrameLevel2: TF-goup M15
FilterTimeFrame=M5 (Stoch_Filter (14,1,1))
MainTimeFrame=M15 (PriceLine (100,3,3), OriginLine (14,3,3), Stoch_Main (14,3,3))
HighTimeFrame=M30 (Stoch_HighFrame (14,3,3))

# TimeFrameLevel3: TF-goup M30
FilterTimeFrame=M15 (Stoch_Filter (14,3,3))
MainTimeFrame=M30 (PriceLine (100,3,3), OriginLine (14,3,3), Stoch_Main (14,3,3))
HighTimeFrame=H1 (Stoch_HighFrame (14,3,3))

# TimeFrameLevel4: TF-goup H1
FilterTimeFrame=M30 (Stoch_Filter (14,3,3))
MainTimeFrame=H1 (PriceLine (100,3,3), OriginLine (14,3,3), Stoch_Main (14,3,3))
HighTimeFrame=H3 or H4 (Stoch_HighFrame (14,3,3))

* note my broker does not support H4 so I use H3 instead

2. Risk Assessment: I follow strictly Spud's Risk assessment

REASON:

makes sense to me - for the higher TF-groups OriginLine and Stoch_Main are the same and one could simplify it: I do for different reason keep it separated.

SHORT_RISK
SHORT_PriceLine_above_76_4_OR_below_23_6
SHORT_OriginLine_crossed_below_76_4
SHORT_Stoch_Main_below_80
SHORT_Stoch_Main_crossed_below_Stoch_HighFrame
SHORT_Stoch_Filter_below_23_6

LONG_RISK
LONG_PriceLine_above_76_4_OR_below_23_6
LONG_OriginLine_crossed_above_23_6
LONG_Stoch_Main_above_20
LONG_Stoch_Main_crossed_above_Stoch_HighFrame
LONG_Stoch_Filter_above_76_4

* ADDITION: 'Entry_MinDistance' at the time of entering a new Trade I do a check for a min distance of Stoch_M5 :For LONG this is the difference to the 100 for Short Distance to the 0 Stochastic number.
e.g 'Entry_MinDistance' = 30 means: For a SHORT entry the Stoch_M5 MUST be additionally above 30, For LONG entry the Stoch_M5 MUST be additionally below 70

REASON: as I do run a fully automatic system - I had it sometimes that all Risk where 0 but the Stoch_M5 had reached already the other side which I did not think I want for obvious reason.


I count Risk Levels like that: e,g: before risk assessment I set
SHORT_RISK = 5 - I assume at start of assessment always Highest Risk- until I proved it to be lower.

for each failed check I subtract 1.

So my Risk Levels go from 0 - 5

Acceptable Risk Level is 0,1 and 2.

The additional 'Entry_MinDistance' is not counted in by the Risk Levels (to be consistent over TF_groups) but only applied separately for new Trades.



3. NEW TRADES ENTRY:

My broker does not allow SHORT and LONG Trades in one account: So I use 2 Accounts - one for only Short trades the other for only Long trades.

a. I only enter new trades on TF_group M5 -> and only if on this particular TF_group and Account there is not already a open trade: (you will understand better what I mean when I come to Switching TF-groups.

b. I enter new trades only by 3 assesments:
TF-group M5 must have Risk level: 0
additional 'Entry_MinDistance' : must be met
only if there is NO open trade on TF_group M5 at the moment (for that Account)


REASON: only if there is NO open trade on TF_group M5 at the moment (for that Account)

I re asses the Risk Levels every Minute (actually 5 seconds after a full min - so to have just a closed Candle)

If TF_group M5 had just Risk 0 and I enter a New trade - it properly will have Risk 0 in the next minutes check again - I would get too many multiply Trades for actually one clear Entry. (There are many ways to solve this - I did it as described above)


you will see that I fully allow multiply trades. Theoretically there could be hundreds of Long and hundreds of Short trades open at the same time.


If all Trades are on higher TF_groups I enter still New trades: Short and Long whatever comes



4. Time Frame Group Switch and EXIT: I differ here quite a bit from Spud.

REASON:

For me it makes more sense to switch Time Frame Levels (TF-groups) when the acceptable Risk is broken.

my Acceptable risk is: Risk 0,1, or 2 (by my Risk assesment as defined above)


EXIT: I exit never by moving Trailing StopLoss or TakeProfit Trades (StopLoss is purely emergency here)

I exit only by Risk assesment.


Basically I implemented 3 version (2 are quite similar) of Time Frame Group Switch to choose from:

I think they advance go from safest to most risky TF_group switch (remember I don't trail StopLoss)

1 TimeFrameLevel_ChangeType 1 Rule: Stay on current TF-group until Risk reaches 60% or more (by my definition: 3 or more) --> re-asses immediately next higher TF-group - if Risk below 60 % move Trade up otherwise close --> when Last TF-group Risk reaches 60% or more always exit.


2 TimeFrameLevel_ChangeType 2: Rule: Stay on current TF-group until Risk reaches 60% or more --> re-asses immediately next higher TF-group - if Risk below 60 % move Trade up otherwise close --> when Last TF-group Risk reaches 60% or more - re-asses immediately the very first TF-group - if Risk below 60 % move Trade to it otherwise close --> and we start the Circle from the beginning

REASON 2: just a small twist to number 1 --> basically I just liked the though of possible theroretically entering an ever ongoing profitable trade ;-)


3 TimeFrameLevel_ChangeType 3: Rule: Stay on current TF-group until Risk reaches 60% or more --> re-asses immediately next higher TF-group - if Risk below 60 % move Trade up otherwise -- check in an order (always to the next lower TF-group) if any of the lower TF-groups have a supportive Risk Level if so move the Trade to it (and continue as before) - if the next Higher TF-group and NO Lower TF group have a supportive Risk level close the trade (never over jump a TF-group higher - but you can overjump lower TF-roups.

REASON 3: I think overjumping higher TF_groups add to my risk as there are substantially higher reversal within the still acceptable Risk assessment. - whereas going to Lower TF_groups does not have such a problem.

Side Note: only with TimeFrameLevel_ChangeType 3 I can have multiple trades on TF_group M5 (in case I move Trades down).

There are other things as possible money management ect - but I think I shared most of the trading rules

I should say: That I'm still not 100% convinced if Switching Time Frames makes really sense and is necessary to get the optimum of profit. Because of that I also implemented a second version.




PMTF Stochastics Without Time Frame Shift_USING Trailing StopLoss:

REASON:

Thought is like that: A trade itself has no idea about TF http://www.forexfactory.com/images/icons/icon12.gif

So if a trade follows certain rules (like the one presented by Spud MTF Stochastic) it will do so if we switch or if we do not switch TF.

So it would seem to me reasonable to apply the same Entry rules on TF_group M5 and than just forget all thoughts about TF and use a trailing StopLoss.

Tricky thing is to get the optional StopLoss so to not get Stopped out too early -- but also not too far away to loose halve of our profit if the Trade turns.


SO: for my second VERION: PMTF Stochastics Without Time Frame Shift_USING Trailing StopLoss

1. I use ONLY TF_group M5 and only to define New Trade entries.

2. I apply the exact same rules for Trade Entry as in the other system.

Note: (this means I can not have multiply trades in one account) max is one Long and one Short trade open at the same time.

3. I still set the exact emergency StopLoss in the beginning.

4. I have added an optional parameter to set a fixed Take Profit by Trade Entry. just to use it if wanted - if used it changes to some extend the reason for the trailing stop loss.

5. most important in this approach: TRAILING STOP LOSS SYSTEM:

I can think of different possiblilities to do a flexible system:

The one I implemented: conficuration is like this:

"TrailingStopLoss" : [
{ 'MinProfitInPips' : 12, 'TrailingStopLossPercentage' : 17},
{ 'MinProfitInPips' : 20, 'TrailingStopLossPercentage' : 30},
{ 'MinProfitInPips' : 30, 'TrailingStopLossPercentage' : 45},
{ 'MinProfitInPips' : 40, 'TrailingStopLossPercentage' : 50},
{ 'MinProfitInPips' : 60, 'TrailingStopLossPercentage' : 60},
{ 'MinProfitInPips' : 80, 'TrailingStopLossPercentage' : 60},
{ 'MinProfitInPips' : 100, 'TrailingStopLossPercentage' : 70},
{ 'MinProfitInPips' : 120, 'TrailingStopLossPercentage' : 80},
{ 'MinProfitInPips' : 150, 'TrailingStopLossPercentage' : 85},
{ 'MinProfitInPips' : 170, 'TrailingStopLossPercentage' : 90},
],

One can add as many settings or as few as one wants - completly adjustable.

REASON:

I one uses a fixed let's say 50% Trailing Stop Loss you get this problem:

a. In the beginning where one is just 10 or 20 PIPS up it is too close and get's easily hit.

b. later on if one was lucky to nmake it throught the first stage and say we are up 150 PIPs it is too far aways and if the trade turns against you would have in the end just (50% from 150 = 75) 75 PIPs.


With adjusting the Trailing StopLoss by percentage of profit this is somewhat taken care for.

There would be a even more delicate way to do it and adjust it all purelly by percentage like:

{ 'MinProfitInPips' : 12, 'MinTrailingStopLossPercentage' : 17, MaxProfitInPips: 200, MaxTrailingStopLossPercentage' : 95, }

And than calulate dynamicaly at any point the correct Percentage When we Reach our MaxProfitInPips: 200 - we reached also our MaxTrailingStopLossPercentage -- after that we just keep it by MaxTrailingStopLossPercentage.

If you get what I mean.


I implemented my version of Trailing Stop Loss because I wanted to set key points. and not only be linear.

So: I will still need a couple of days to run both implementations parallel.



HOPE for someone - something I shared might be a idea to take and stimulates or thinking too. Or you spot a problem and point it out.


MJ




Maybe I forgot to mention: that both system implementation are designed to trade multiple Pairs at the same time - basically as many as you define.


(Note for someone who thinks manual trading is always better:
if one trades 8 Pairs at the same time you have to assess 16 Risk potions(SHORT and LONG) for each TF_group having 4 TF_groups one would have to keep track of 64 Risk assessments not to mention all the different Stochastic lines per TF_group. AND additionally keep track at about each Trade on which TF_group it is actually trading just to mention a few reason -- of course one can always just stick to 1 or 2 pairs ... but there are some god points in automatizing....)

To help me thinking at what candle I'm looking at in a multi time frame system I made a simple chart. I thought why not put it up - maybe it is for some others useful too.

I hope it works in excel (I actually don't use Excel so I exported it).


MJ

Well,
I noticed that my indicator does not work so well.
It isn't very easy to put together all the combinations, and certainly everything that we see with our eyes in movement and position of stochastic can't put it in the indicator.
However I am working to improve it.

Good job to all.

P.S. in the last pages of thread, there are a lot of new information, I'm not yet able to read them all, working to the indi.

hi LuboLabo,

The only real thing to think about and which greatly complicated it for me was the 2 crosses (actually 4 if one counts SHORT and LONG)

All other Risk assessment would be easy - but how do you know that a cross has happened.

And this has to be done for all TF-groups all the time.



I solved it that I keep track of it (I actually keep track of all conditions as it was not much difference). And if I start fresh or restart for any reason I go back for each time frame group about 481 Candles(Filter time Frame.)

To align there all other candles rightly was quite some thinking work too.

Don't know how one would solve this differently - except to alter the rule and use only the current position. (as I tried and Spud said the cross is important and I agreed with the Master;-)

Happy programming

MJ

Strictly speaking yes it is an R20. My point was..grab your pips before you get there.

Perhaps an indicator without the crossover? That would at least simplify it(?) Also, wondering if it would be better just to update the bars on the last candle close, rather than live. Would that not be easier to calculate a crossover (on candle close)?

Hi Spud today i seem to be starved of any good entries with the price line not doing its thing so maybe I'm a bt bored for my own good and trying to find pips where there are none... But

I was just wondering though - Going by your risk% if we take a trade with the price line out of position then technically it if everything else is in the right position it would give us a 20% risk (calculation)

looking back at today there seem to be many entries like this that worked out and probebly roughly about 4 out of 5 success rate...

If we can justify the 20% risk with the position of the higher timeframes is it still a safe entry instead of a gamble?

Did not skyline do such indicator already?? As I do not use MT4 do not know.

I think without the crossover it would be much simpler as one has just to verify the current positions at any time in history.

Simply: is it now: PriceLine_above_76_4_OR_below_ 23_6

one does not need to keep track of anything: just ask the question at the any time.

the crossover has the real problem (I don't think skylines version had solved that even though if I remember correctly he checkt for crosses): How do I know when I start my computer - if a crossover had already occurred before? - one has to look back a certain number of frames. But who knows how many ;-) you get my point. So thats why I think it is better to keep track - also less calculation when one has it up and running.

I choose 481 Filter Candles only because I do not get much more History Candles from my broker. So I do the best as I can and afterwards keep track.

Not sure if I understand what you mean: Also, wondering if it would be better just to update the bars on the last candle close, rather than live. Would that not be easier to calculate a crossover (on candle close)?

skyline and I anyway followed your suggestion to use only Last Closed candle as Risk assessment point. In case of a crossover there is also involved the closed candle previous to our Last One.


Anyway - I'm do not know anything about MT4 EA's and do not know if it is suited to address all this question -

But I could imagine that properly a programmer could implement such thing - BUT because of the complexity it took me about 10 long days - and I did not completely finish it yet - neighter look for all posible bugs ;-)

MJ

Hi CindyXXXX - maybe an other alternative could be looking at other pairs if one does not do anything.

Just a thought - not sure if it really works as Spud always mentions just EUR/USD - but how he presented the system would suggest that other pairs will follow similar rules.

Happy trading to you MJ

Hi All,
I have improved the indicator.
Now all bars are indipendent and i have added a new feauture, we can choise no last bar.
Thrown in the recycling bin the previous indicator and try this.

Lubo

P.s i don't understand well the problem with corss.
help me to improve the indicator.


Thanks.

You can certainly trade all pairs with this....I find EUR/USD enough for me and prefer the low spread...but the same fundamentals hold true for all currencies. You will want to find currencies that tend to break out of the overmarkets more quickly rather than those that doddle forever in them.

You can also trade the overmarkets, you need to change the rules a bit but the same rules and combinations that take you out can keep you in an overmarket trade (think backwards)...take the entry on the entry of the overmarket for M5 and go from there.

Spud, Cindy and I were talking and she is only entering trades once the priceline has peaked at 90 or 10. This filters out a lot of trades but pretty much ensures that price is changing direction. What I don't understand though is do we WANT the priceline to be separated a good bit from the Origin line? and if so, why? Is it not good to have them both moving up over 23.6 or down over 76.4 together?

Lubo good to see you working on the indicator.

not sure what you do not understand about the cross.


I try to explain simply what I mean:

Let's just use a SHORT Risk assessment as example:

You switch on your PC and start your indicator:

you check 5 things:

EASY: SHORT_PriceLine_above_76_4_OR_below _23_6
EASY: SHORT_Stoch_Main_below_80
EASY: SHORT_Stoch_Filter_below_23_6

PROBLEM:
SHORT_Stoch_Main_crossed_below_Stoch_HighFrame
SHORT_OriginLine_crossed_below_76_4


Tell me how do you know if the OriginLine has crossed below 76.4% ?

You may say it the easy way: well if the Previous_OriginLine was above or at 76.4% and the next (LastClosed Candle) is below 76.4% it has crossed.

This is True (if I remember correctly skyline did something like that) and I do it also.


But it is just the half truth: look at Spuds picture below about the much later entry - this will happen (maybe in a slightly faster way) with using the M1 filter most of the time.

The cross could have happened at any time before the entry ... It could have happened already before you switched on your PC. How do you know it has happened ???

Hope this clears a bit about the Problem of the crossing - it is mainly a problem because it must not happen simultaneously with our Trade enter (in fact it will properly not happen often simultaneously because of the other assessments)

Hope this helps a bit.

Think of the price line and 14,3,3 much like a %K and %D..or almost any oscillator or moving avaerage with 2 lines. Strong moves come when the lines split apart and then they eventually snap back together.

The problem with a 100,3,3 line that is responding quickly with the 14,3,3 line is that it usually is a weak situation. The ideal situation, is that the 14,3,3 pulls away from the price line and heads for say, oversold. Then the price line moves and follows it down. That means we have travelled in a very strong down move.

If you look at good escalator moves in a down trend, you'll see this move. In the strict system, I'm trying to use this split as a good down move. We simply repeat this split through the time frames...in otherwords we are trying to avoid the snap back and take advantage of the stochastic placement that gives us the strongest move (reliably). However, if you look back at M5 say from M30 and we are in a down trend you will likely see the 14,3,3 sticking in an oversold position and the M5 price line falling.

I don't want to confuse issues too much, so I'll leave it at that as to why we are using the split of the Origin and Price Line to enter a trade.

Well, i understand what u mean.
In my indi there is not this problem, all bar are indipendent:
Price:
- price above 76.4 & price < price prev : color red
- price below 23.6 & price < price prev : color green
Origin:
- origin below 76.4 & < origin prev : color red
- origin above 23.6 & > origin prev : color green
Stoch Cross:
- stoch cross1 < 20 & stoch cross1 < stoch cross2 & stoch cross1 < stoch cross1 prev : color red
- stoch cross1 > 20 & stoch cross1 > stoch cross2 & stoch cross1 > stoch cross1 prev : color green
Filter
filter < 23.6 : color red
filter > 76.4 : color green

When this combinations aren't meet : color neutral.

that's all!

Lubo.

hi Lubo,

I suppose:
color red means: SHORT OK
color green means: LONG OK
color neutral. means not OK

if that is so I can see a lot of differences to what I understood -- who knows I might have misunderstood something.

MJ

CAN a gifted programmer convert this indicator to an all stochastic no last bar indicator
thanks

I have updated preview post because i missed the underline specification

Price:
- price above 76.4 & price < price prev : color red
- price below 23.6 & price < price prev : color green


I' don't know if you have misunderstood something, but red & green does not mean necessarily short and long, but only the condition that I mentioned in prev post are reached.

Lubo