Best Thread Vwap Engine

[leovirgo]

tricky spike

Got a nice short fom 89tick chart for DIA(INDU) at 82.40. It later developed into a swing trade and stop moved to breakeven + 4 ticks. Market DID crash but the spike touched the order and left with a mere +4ticks for 250+ move.

First chart is system entry and exit at lower TF 89 tick.

Second chart is INDU 10MC system trade entry and exit.

Looks very good on the chart but in reality only managed to scoop a staggering 4pts!!!

The question is ...

For a trade that goes 77% of target and hence must be trailed at breakeven stop, should it be re-entered once stopped out?

Well, market is market it will behave as it likes. I must stick to the discipline. I see no need to change anything as it was the spike. It is one harsh reality about the markets.

On stocks, had a nice run +100C on WFC. COF short is stopped out like DIA. hmmmm, run up profits vaporised although it could be a week's pay in one day.

On NQ, Vwap engine scalped some points. NQ was stronger than INDU and signal came later than INDU.

At the end of the day made some $$$ but could have been a much more profitable day. Discipline-wise a good day ....

 

[Trader333]

Market DID crash but the spike touched the order and left with a mere +4ticks for 250+ move.

Are you using any form of market volatility measure to determine how much against you a move may go before going into profit ?

Great thread though and thanks for posting


Paul

 

[leovirgo]

Are you using any form of market volatility measure to determine how much against you a move may go before going into profit ?

Great thread though and thanks for posting


Paul

Paul,
Yes I use that kind of thing but basically it is to limit the initial risk.

In this case, I mean the trade on Tuesday, it is the opposite - the trade had a run up profit of 0.77R where 1R = expected profit target, therefore the trailing stop had to be set at breakeven +a few cents to cover commission. Generally this is a thought to be a good behaviour not let a profitable trade turn into a looser. ( I couldn't close it as TA hadn't met exit conditions. If TA reached exit point I would have taken 0.7R profits.). Anyway, that trade was stopped out in the spike only to see a potential profits of 2.3R!

My risk attitude may be considered as "too conservative". Say my position sizing allows 100 shares with a stop loss of $1, I set $1 as 1R. Then I want a set up where my initial stop to be around 0.3R and expected return was at least 0.7R or as much as TA allows. I use win/loss ratio of 0.4 no matter how the strategy performed in back testing. I thought the best way to weather a string of losses would be to choose the highest R:R set up.

Thats why I choose to enter a trade with lower TF entry where initial stop is small, and yet it could be a big winner. I will present the dilemma in the next post.

 

[leovirgo]

Question for the visitors

Dear fellow traders,

I want to present my dilemma, it may also happen to you as well.

For an entry at lower TF, say 3min Overbought(OS) Short Entry with the back up of 10min and 60min all lining up behind, assuming that the trade goes in your favor, the options for profitable exit are

Cycle exits
1. take partial profits when 3min is Oversold, and let the rest go with higher TF until 10 and 60min are all OS for ultimate exit;

2. just make it freebie when 3min is OS and exit only when 10min & 60min are all OS -in this case, should it be closed if 10min cycle is turning up against the position?

3. Take profit no matter what the cycle is doing if it reached the profit target of 2x initial stoploss. It could be any thing greater than 1.

Price line exits
4. Exit at important price lines such as Pivots, vwap, yesterday's low etc. No matter what the return is.

The choice for exit could depend on the risk, so here are the options

Stoploss (R) options
1. It could be based on volatility such as 3min ATR, or 10min ATR or even 60min ATR. The higher the TF, the bigger the stoploss (R, risk) and the need for bigger return.

2. Based on Price lines - such as important resistance or pivots etc, calculate the distance between entry and that stop line and adjust position accordingly.

Depending on your choice of R, what should be the profitable exit in the long run?

 

[Trader333]

Personally, (and for this type of trade), I would go with the Grey1 approach to this and take 3/4 off the table as soon as you have reached 1 ATR in the timeframe you are trading and move your stop to BE. Then close the last 1/4 when you get the OB / OS signal.


Paul

 

[leovirgo]

Personally, (and for this type of trade), I would go with the Grey1 approach to this and take 3/4 off the table as soon as you have reached 1 ATR in the timeframe you are trading and move your stop to BE. Then close the last 1/4 when you get the OB / OS signal.


Paul

Interesting approach. Assuming that initial stop is 3min 1ATR, it can be a quick money making plan if the win/loss ratio is more than 50% . I will definitely have a look from this angle.

At the moment, I am using an exit with 10min OBOS for a 3min entry. It can have small losses/wins or larger wins.

By the way this is today's results.

Trade 1: NQ short with small loss (2pts)
Trade 2: NQ short with +10.5pts profits

Happy with vwap engine's performance.

As today is non-buying day due to very bearish market internals, the engine will not generate buy signals, despite the dark green bars that are potential 'reversal long trades'.

 

[Trader333]

Looks promising, do you also take Short trades as well ?


Paul

 

[pedro01]

Dear fellow traders,

I want to present my dilemma, it may also happen to you as well.

For an entry at lower TF, say 3min Overbought(OS) Short Entry with the back up of 10min and 60min all lining up behind, assuming that the trade goes in your favor, the options for profitable exit are

Cycle exits
1. take partial profits when 3min is Oversold, and let the rest go with higher TF until 10 and 60min are all OS for ultimate exit;

2. just make it freebie when 3min is OS and exit only when 10min & 60min are all OS -in this case, should it be closed if 10min cycle is turning up against the position?

3. Take profit no matter what the cycle is doing if it reached the profit target of 2x initial stoploss. It could be any thing greater than 1.

Price line exits
4. Exit at important price lines such as Pivots, vwap, yesterday's low etc. No matter what the return is.

The choice for exit could depend on the risk, so here are the options

Stoploss (R) options
1. It could be based on volatility such as 3min ATR, or 10min ATR or even 60min ATR. The higher the TF, the bigger the stoploss (R, risk) and the need for bigger return.

2. Based on Price lines - such as important resistance or pivots etc, calculate the distance between entry and that stop line and adjust position accordingly.

Depending on your choice of R, what should be the profitable exit in the long run?

One thing I would say is in terms of cycle exits - it depends how you are using cycles.

If you are using MACCi, then you are using a lagging indicator. By the time the 10 minute turns on the MACCi, the 10 minute has already turned down.

MACCi is not cycle analysis - it is an oscillator and not much different from RSI, stochastics etc. There is nothing inherently wrong with this but it is not cycle analysis.

Cycle analysis is the interpretation of past market cycle data to predict the next cycle. Hurst does this by averaging the cycles and Ehler does it by sampling a small portion of a cycle and predicting the rest of the cycle. I haven't seen many indicators yet that even come close to making such predictions.

On any oscillator - you have a choice between a longer lookback period for a smoother but more lagging signal and a shorter lookback period for a more reactive but choppy indicator. What you will find is that the lookback period will need to be tweaked on occasion to fit the current market conditions - as things calm down, any indicator with a lookback period will need to be tweaked to continue producing good results.

Take the MACCi on the 60 minutes - even a short average of the CCi such as 4 periods will give you signals hours after the 60 minutes has turned. When you look at how much the 1 min will have moved in that time, it will be a large amount.

So - the question is - how long after the price moves at the 10 min does your indicator give you a signal ? If it's too late, then there's a danger in using the 10 min as an exit point.

 

[dcraig1]

Cycle analysis is the interpretation of past market cycle data to predict the next cycle. Hurst does this by averaging the cycles and Ehler does it by sampling a small portion of a cycle and predicting the rest of the cycle. I haven't seen many indicators yet that even come close to making such predictions.

But does Ehlers stuff for example actually work and have any real predictive value. I've looked at fair bit of his free stuff (not MESA) and I'm yet to be convinced. For example I couldn't find anything really useful in his bank of bandpass filters to determine dominant cycle. Very nice eye candy if you replicate his heat map charts from the eminiz web site , but thats about it. No better than MACCI over several TF's as far as I could see. Maybe not as good. I would like to hear otherwise. Just for laughs I tried applying it to market delta and order book stuff too and couldn't see anything.

I've also tried in the past looking at Hurst's channels based on centered MA's and forcasting them ahead, looking at their intersection and didn't find anything conclusive.

A bit off toipic, but the best predictor I've been able to find for market turning points in current conditions is to eyeball a whole bunch of stock charts. When a lot of stocks are sitting on clearly defined support, the market bounces and the inverse for a top. By support, I mean a horizontal line drawn in the conventional manner. No fib flim flam or other such nonsense. It's almost laughably simple, though time consuming and not easily automated. I'm referring to tops and bottoms in a swing trading sense. Something like 30 or 60 minute charts.

As a more philosophical point, I find this thread interesting because of VWAP. We know that VWAP is one of the things that a lot of market participants "take note of". I would also contend that things like floor pivots, market profile levels and simple S/R levels also fall into that category. In the case of floor pivots, I'd bet this assertion is provable by looking at the volume distribution around the pivot levels.

On the other hand, ascribing certain properties to price time series by what are basically data mining techniques without plausable explanation as to why the TS should exhibit such behavior does seem somewhat more problematic. I'm not saying it's fundamentally flawed, but it is very easy to get it wrong.

One thing I would like to know is how many real quants use DSP techniques (such as Ehler). You don't hear anything about it, but that doesn't mean it's not going on.

 

[pedro01]

But does Ehlers stuff for example actually work and have any real predictive value. I've looked at fair bit of his free stuff (not MESA) and I'm yet to be convinced. For example I couldn't find anything really useful in his bank of bandpass filters to determine dominant cycle. Very nice eye candy if you replicate his heat map charts from the eminiz web site , but thats about it. No better than MACCI over several TF's as far as I could see. Maybe not as good. I would like to hear otherwise. Just for laughs I tried applying it to market delta and order book stuff too and couldn't see anything.

I've also tried in the past looking at Hurst's channels based on centered MA's and forcasting them ahead, looking at their intersection and didn't find anything conclusive.

A bit off toipic, but the best predictor I've been able to find for market turning points in current conditions is to eyeball a whole bunch of stock charts. When a lot of stocks are sitting on clearly defined support, the market bounces and the inverse for a top. By support, I mean a horizontal line drawn in the conventional manner. No fib flim flam or other such nonsense. It's almost laughably simple, though time consuming and not easily automated. I'm referring to tops and bottoms in a swing trading sense. Something like 30 or 60 minute charts.

From what I have seen of Ehlers indicators, it only plots on the current bar & to the left. If they are predictive, I'd expect some predictions to the right of the chart or some outputs that I could use in my scripts.

As of yet - I have seen none of that from Ehlers indicators - so I don't see in what way they really have any predictive value in terms of predicting the next cycle.

As momentum indicators they are fine.

This picture shows 1 min cycles within a 10 min cycle.

I think it explains itself & how you can perhaps look at the lower timeframes higher cycle low to give an idea of when the higher timeframe cycle low is approaching with less lag.

 

[leovirgo]

One thing I would say is in terms of cycle exits - it depends how you are using cycles.

If you are using MACCi, then you are using a lagging indicator. By the time the 10 minute turns on the MACCi, the 10 minute has already turned down.

MACCi is not cycle analysis - it is an oscillator and not much different from RSI, stochastics etc. There is nothing inherently wrong with this but it is not cycle analysis.

Cycle analysis is the interpretation of past market cycle data to predict the next cycle. Hurst does this by averaging the cycles and Ehler does it by sampling a small portion of a cycle and predicting the rest of the cycle. I haven't seen many indicators yet that even come close to making such predictions.

On any oscillator - you have a choice between a longer lookback period for a smoother but more lagging signal and a shorter lookback period for a more reactive but choppy indicator. What you will find is that the lookback period will need to be tweaked on occasion to fit the current market conditions - as things calm down, any indicator with a lookback period will need to be tweaked to continue producing good results.

Take the MACCi on the 60 minutes - even a short average of the CCi such as 4 periods will give you signals hours after the 60 minutes has turned. When you look at how much the 1 min will have moved in that time, it will be a large amount.

So - the question is - how long after the price moves at the 10 min does your indicator give you a signal ? If it's too late, then there's a danger in using the 10 min as an exit point.

Hi Pete
Thanks for the cycle explanation. By cycle I mean MACCI OB to OS or vice versa, just a simple cycle with OB node and OS node.

Yes, MACCI is a lagging indicator. But it doesn't matter. Because the ultimate non-lagging indicator is the price. With the use of "price action", the cycle becomes easier to trade. Stochastics or RSI or any oscillator will do. Thats where the use of vwap becomes handy.

To me as a trader is not to predict the exact cycle turning point but to ride a cycle with a certain degree of confidence that it will reach the desired target.

For exits, I find that exhaustion exits are better than reversals. The use of 10min as cycle exit is perfectly alright for a lower Tf entry ,say 3min, provided that the position is closed on 10min OB or OS i.e, when MACCI is >100 or <-100 and not when it starts to turn back. It is also better if 3min is also OBS at the same time. As long as R:R works out, it is just another trade. If a trade goes from A to B then it is profit exit and the job is done. If the cycle turns back before reaching B, then it is TA exit.

 

[dcraig1]

pedro01,

Could you repost that chart with price and time axes (and date and instrument if not shown). It would be interesting to look at that slice of the market from another perspective, and compare.

 

[pedro01]

Hi Pete
Thanks for the cycle explanation. By cycle I mean MACCI OB to OS or vice versa, just a simple cycle with OB node and OS node.

Yes, MACCI is a lagging indicator. But it doesn't matter. Because the ultimate non-lagging indicator is the price. With the use of "price action", the cycle becomes easier to trade. Stochastics or RSI or any oscillator will do. Thats where the use of vwap becomes handy.

To me as a trader is not to predict the exact cycle turning point but to ride a cycle with a certain degree of confidence that it will reach the desired target.

For exits, I find that exhaustion exits are better than reversals. The use of 10min as cycle exit is perfectly alright for a lower Tf entry ,say 3min, provided that the position is closed on 10min OB or OS i.e, when MACCI is >100 or <-100 and not when it starts to turn back. It is also better if 3min is also OBS at the same time. As long as R:R works out, it is just another trade. If a trade goes from A to B then it is profit exit and the job is done. If the cycle turns back before reaching B, then it is TA exit.

OK - gotcha - makes more sense to look for OB/OS than a turn, especially as the turns can be a bit lazy unless you have a small lookback.

 

[pedro01]

pedro01,

Could you repost that chart with price and time axes (and date and instrument if not shown). It would be interesting to look at that slice of the market from another perspective, and compare.

Sure. That was a slice from the following chart:

Blues are 10 min, whites are 1 min cycles.

 

[leovirgo]

A bit off toipic, but the best predictor I've been able to find for market turning points in current conditions is to eyeball a whole bunch of stock charts. When a lot of stocks are sitting on clearly defined support, the market bounces and the inverse for a top.

You're right. Just scan SPX100 and NDQ100 daily intervals in RS and apply any oscillator, if 80% are OS then market will go up for sure.

As a more philosophical point, I find this thread interesting because of VWAP. We know that VWAP is one of the things that a lot of market participants "take note of". I would also contend that things like floor pivots, market profile levels and simple S/R levels also fall into that category. In the case of floor pivots, I'd bet this assertion is provable by looking at the volume distribution around the pivot levels.

Turning points from these levels are tradable regardless of the TF. And you know the potential target straight away.

 

[leovirgo]

Looks promising, do you also take Short trades as well ?


Paul

thanks! I don't get what you mean by Short trades? If you were refering to long trades then .. no I did not take any long trades yesterday although it was a very good move. Alot of INDU traders got it.

 

[leovirgo]

The set up for today was STRONG set up with alot of resistance levels to breakthrough.

3 trades today- 1 win, 2 b/e. Net positive day but could have been twice as profitable. Second trade was profit target exit rather than cycle exit. Set third trade as breakeven too soon (manually). Should have waited 1min and it could be another profit exit at target 1247.

Looks like it's become a part-system part-discretionary approach. Probably due to two things-

1. need better program coding skills AND/OR
2. the engine has too many variables.

All in all, a good week for the vEngine.

 

[leovirgo]

No trades today due to two things - 1. internet connection is not good 2. system did not produce valid signals. Must be a very good day for swing traders or higher TF traders as there were shorts signals at 1hr charts since Friday. will write my reflections about today if I have time later on...

 

[leovirgo]

Vix in action

I posted this Vix chart in my dow thread, it was as of 11Feb. Vix indicated it could break out i.e, markets to go crazy downside. Longer term vwap engine must include vix factor. food for thought!!

 

[leovirgo]

how to remove whipsaws in cycles

System generateed three trades

1) +0.25 pts ( after reaching +10pts run-up), 2) -4.25pts, and 3) -1pt

Funny thing is I missed to take the first profitable trade. I was parallel running an improved version (based on ticks) and hence I was confused. Enhanced version signals were

1) +8.5pts, 2) -2pts.

This tick version has been tested since the early days of journal but I was taking time to see if it is really necessary. I have assessed theoretical frameworks for both of them and need to see them in live environment. They both work in the sameway. Current version has 10min and 3min as the decision TF whereas the improved version makes use of ticks data. Cycles are far smoother in the latter as it produces more bars when things are moving quickly and produces lesser number of bars when things are idle such as during lunch time. The result is that it is a far better cycle to trade than fixed-interval (fixed number of bars) cycles. This is very important for cycle trading especially for 24hour rolling futures where some out-of-market cycles do/don't matter for cycle calculation.