Jump to content

Welcome to the new Traders Laboratory! Please bear with us as we finish the migration over the next few days. If you find any issues, want to leave feedback, get in touch with us, or offer suggestions please post to the Support forum here.

  • Welcome Guests

    Welcome. You are currently viewing the forum as a guest which does not give you access to all the great features at Traders Laboratory such as interacting with members, access to all forums, downloading attachments, and eligibility to win free giveaways. Registration is fast, simple and absolutely free. Create a FREE Traders Laboratory account here.

suby

Regime Changing

Recommended Posts

Some say the success to succesful trading is money management (i.e. as long as your winners beat your losers thats great).

 

Every trader says you need to find a strategy or system that jives with your personality. Sure, also great

 

Real money is made from being able to identify regime changes. Theres a reason why the majority of "daytraders" go broke. Why beacause there trading noise.

 

I want to open up this thread to regime changing.Its in technical analysis since that probably what the majority of people use here to help them identify their trends; however, indicators are secondary variables.

 

I look forward to hearing back to the community on their inputs about regime changing.

 

I'm trying to learn more about this. If anyone has any recommended books/resources, i'd love to learn

 

Suby

Share this post


Link to post
Share on other sites

Hi Suby,

 

It might be worth searching through some of Jeff Swanson's articles on this site - he has posted useful and well tested ideas for regime switching models incorporating concepts like hysteresis.

 

If you just want to know whether to be long or short only, then you can do a hell of a lot worse than a simple moving average - the problem is always arriving at a solution that works and is not curve-fitted.

 

Regards,

 

BlueHorseshoe

Share this post


Link to post
Share on other sites
Hi Suby,

 

It might be worth searching through some of Jeff Swanson's articles on this site - he has posted useful and well tested ideas for regime switching models incorporating concepts like hysteresis.

 

If you just want to know whether to be long or short only, then you can do a hell of a lot worse than a simple moving average - the problem is always arriving at a solution that works and is not curve-fitted.

 

Regards,

 

BlueHorseshoe

 

Bluehorseshoe,

 

I appreciate the advice, i looked through Jeff Swansos articles on the site, smart man. He mainly focuses on individualistic systems towards the indices. Lately i've been throwing in a lot of secondary variables into my trading. I.e. Stocks vs Bonds or Stocks vs the Vix.

 

Do you ever look at intermarket relations in your trading?

Share this post


Link to post
Share on other sites

 

Do you ever look at intermarket relations in your trading?

 

I have tried doing so in the past with intraday stuff, but without any success. I also went through all the $tick, $trin, $vix "market internals" stuff years back (before I was backtesting - so I lost real money trying that stuff out), and then the Larry Williams indicators based around bond/stock index relationships . . . There was nothing there for me.

 

One thing I have noted, but have yet to properly investigate, is that when a group of correlated markets are trending (and they need only be loosely correlated), and one market pulls back where the others don't, then an entry in this market has better probability of a successful outcome than when multiple markets undergo a correction together. Not only is such an instrument reacting against its own long term trend, but against the trend of the broader market. A reaction that will become a reversal is more likely to unify behaviour ("when the shit hits the fan, all correlations go to 1 . . . !").

 

What I am describing is possibly an aspect of what OneSmith has in mind with the 'M' of CANSLIM.

 

BlueHorseshoe

Share this post


Link to post
Share on other sites
I have tried doing so in the past with intraday stuff, but without any success. I also went through all the $tick, $trin, $vix "market internals" stuff years back (before I was backtesting - so I lost real money trying that stuff out), and then the Larry Williams indicators based around bond/stock index relationships . . . There was nothing there for me.

 

One thing I have noted, but have yet to properly investigate, is that when a group of correlated markets are trending (and they need only be loosely correlated), and one market pulls back where the others don't, then an entry in this market has better probability of a successful outcome than when multiple markets undergo a correction together. Not only is such an instrument reacting against its own long term trend, but against the trend of the broader market. A reaction that will become a reversal is more likely to unify behaviour ("when the shit hits the fan, all correlations go to 1 . . . !").

 

What I am describing is possibly an aspect of what OneSmith has in mind with the 'M' of CANSLIM.

 

BlueHorseshoe

 

BlueHorseshoe,

 

Can you give an example of something you've seen with this? What your describing is more or less how paul tudor jones trades from my understanding and even victor niederhoffer, only he quantifies everything and uses some next level voodoo that only him and his team understand.

 

In regards to correlations reverting back to 1 when shit hits the fan and the M of canslim its interesting you mention that. I havn't had much success (yet) in determining leads/lags or arbitraging intermarket relations; however, I have noticed 2 things of interest. 1... I find the eurodollars will almost lead the american opening or at the very least give a lot of insight into where the price will be heading for the next hour on the american indices and 2... I've noticed a lot of arbitrage opportunities specifically midday/late day when the 3 indexs are out of line

 

Example Nasdaq and S&P are both down and so is the dow but in relation to the two its still higher. More times than not, if the downward trend or upward (whatever the trend is for that moment/session) will allow the trader to take advantage of this anomalie. Yesterday was a perfect example of this. After 2 oclock both NQ and ES were down substantially but the Dow was lagging. Anyone who caught onto this going into the close made a boatload of money

Share this post


Link to post
Share on other sites
Some say the success to succesful trading is money management (i.e. as long as your winners beat your losers thats great).

Good money management can compensate for negative expectancy, bad money management can kill positive expectancy.

 

Every trader says you need to find a strategy or system that jives with your personality. Sure, also great

 

I think this is true, it is imperative as a trader to completely understand your trading concepts and exposure in the market.

 

Real money is made from being able to identify regime changes. Theres a reason why the majority of "daytraders" go broke. Why beacause there trading noise.

 

Most traders go broke, day traders just do it in a surprisingly efficient manner. "Regime changing" exists on every frame of reference, not just large time frames.

All price data is noise, Time frames are just different ways of parsing the same data, and if you asked me, I would much rather start with the most granular data and construct the buckets myself.

 

One example of a major regime change is time-series evaluation. Time-series are an example of an archaic evaluation method. Your opponents aren't restricted by time any more, so why are you?

 

I want to open up this thread to regime changing.Its in technical analysis since that probably what the majority of people use here to help them identify their trends; however, indicators are secondary variables.

 

At the end of every period of trading you know who won. We either close down or up, If you start with that assumption and work backwards, you can start building an idea of determining who WAS in charge and all the variable that landed in their favor. Price variance would be a fairly good place to start.

 

I look forward to hearing back to the community on their inputs about regime changing.

 

I'm trying to learn more about this. If anyone has any recommended books/resources, i'd love to learn

 

Suby

 

Suby,

Sorry to see your having trouble carving out your niche. So I'll offer up some of my thoughts on your statements above,

Share this post


Link to post
Share on other sites

 

Can you give an example of something you've seen with this?

 

Sure - I'll post a chart example for you over the weekend.

 

What your describing is more or less how paul tudor jones trades from my understanding and even victor niederhoffer, only he quantifies everything and uses some next level voodoo that only him and his team understand.

 

I don't know anything about Jones (in fact, I would have guessed he was global-macro!), and Niederhoffer is an options seller, and I don't know anything about options either, but I would imagine he (can he afford a "team" nowadays?) probably goes through a very complicated mathematical process to arrive at trading decisions.

 

I havn't had much success (yet) in determining leads/lags or arbitraging intermarket relations;

 

I'm not sure I would think of this in terms of lead/lag.

 

What I am talking about would be something more like . . . erm . . . one-sided pairs trading . . . Think classic pairs trade, replace the mean you expect reversion towards with a broad market index, lose one half of the pair . . . You now have a straightforward directional position in one market, and you're not beta neutral

Yesterday was a perfect example of this. After 2 oclock both NQ and ES were down substantially but the Dow was lagging. Anyone who caught onto this going into the close made a boatload of money

 

Or . . . if you start with a directional approach in the DOW that works, and then bring in directional diversion from a broader index as a filter, then that would equate more with what I was trying to describe. The number of trades will fall off sharply, and the performance will (if what I am claiming is true) skyrocket. Then you need to find enough opportunities.

 

Or, you trade the original strategy (which is profitable in its own right, remember), but increase your position size for these higher-probability divergent trades.

 

A few thoughts above. I also thought AddChild's comments were extremely helpful.

 

BlueHorseshoe

Edited by BlueHorseshoe

Share this post


Link to post
Share on other sites
Suby,

Sorry to see your having trouble carving out your niche. So I'll offer up some of my thoughts on your statements above,

 

ADDchild,

 

Thank you for your insightful reply. I took a lot of of that.

 

In regards to time series analysis being archaic, what methods do you recommend one to use in the modern world? The only thing I can think of would be econometric tools or data mining (specifically data mining), is that what you were referring to ?

Share this post


Link to post
Share on other sites
A few thoughts above. I also thought AddChild's comments were extremely helpful.

 

BlueHorseshoe

 

Bluehorseshoe,

 

Thats more or less what I try to structure in my trades - a one sided ERM pair trade. It's one thing to look at charts or prices and notice that things are out of wack but its another to know with certainty through testing that a one sided pair trade under that hypothesis has statistically significance. I've been doing all my work with EOD data because its easiest right now so I have no idea what to look for using EOD data to structure these kinds of trades but I would imagine that one should use intraday data to structure these kind of trades

Share this post


Link to post
Share on other sites
I have no idea what to look for using EOD data to structure these kinds of trades but I would imagine that one should use intraday data to structure these kind of trades

 

Hi Suby,

 

As promised, an example of what I was trying to explain.

 

The first image shows strat applied with signals from primary market only (buy dips in uptrend; sell rallies in downtrend). Over a ten year test period . . .

 

Profit Factor is 2.50 (Longs: 2.3, Shorts: 2.93)

 

The second image shows strat applied with signals from primary market filtered with data from three additional related markets (only buy dip in uptrend when related markets either have not dipped or are not in uptrend etc).

 

Notice how five trades drop out over the period shown? Over the course of the test period this is sufficient to have the following effect:

 

Profit Factor is 5.97 (Longs: 10.80, Shorts: 4.32)

 

This is just an example, so I put no effort into selecting the markets used as filters.

 

I hope that's enough to get you thinking and researching if this is something of interest.

 

Regards,

 

BlueHorseshoe

PrimaryMktSignals.thumb.png.cf4606c24e659ae755d10c40653879a4.png

SecondaryMktFiltered.thumb.png.309c3d480f8df2d3c878e95ce6907edc.png

Share this post


Link to post
Share on other sites
ADDchild,

 

Thank you for your insightful reply. I took a lot of of that.

 

In regards to time series analysis being archaic, what methods do you recommend one to use in the modern world? The only thing I can think of would be econometric tools or data mining (specifically data mining), is that what you were referring to ?

 

Well, its not so much the "analysis" that I feel is archaic, it's the raw data being analyzed, specifically tick data aggregated by time.

 

Econometric tools can be very useful, but in the same light, there are quite a few technical indicators that are actually very useful when you use them in a more realistic, and controlled manner.

 

Data snooping is something I know very little about. It's not an avenue I choose to explore, mainly because for me, I like to build the trading process up from the concept, while data snooping is more like building the process and trying to work backwards to determine a concept.

Share this post


Link to post
Share on other sites
Hi Suby,

 

As promised, an example of what I was trying to explain.

 

The first image shows strat applied with signals from primary market only (buy dips in uptrend; sell rallies in downtrend). Over a ten year test period . . .

 

Profit Factor is 2.50 (Longs: 2.3, Shorts: 2.93)

 

The second image shows strat applied with signals from primary market filtered with data from three additional related markets (only buy dip in uptrend when related markets either have not dipped or are not in uptrend etc).

 

Notice how five trades drop out over the period shown? Over the course of the test period this is sufficient to have the following effect:

 

Profit Factor is 5.97 (Longs: 10.80, Shorts: 4.32)

 

This is just an example, so I put no effort into selecting the markets used as filters.

 

I hope that's enough to get you thinking and researching if this is something of interest.

 

Regards,

 

BlueHorseshoe

 

BlueHorseshoe,

 

Sorry for the delay in reply but thanks for this, defiantly got my thinking cap going

 

Hope you are getting ready for apple earnings tomorrow, PUTs on the indices are looking so cheap right now

Share this post


Link to post
Share on other sites
Well, its not so much the "analysis" that I feel is archaic, it's the raw data being analyzed, specifically tick data aggregated by time.

 

Econometric tools can be very useful, but in the same light, there are quite a few technical indicators that are actually very useful when you use them in a more realistic, and controlled manner.

 

Data snooping is something I know very little about. It's not an avenue I choose to explore, mainly because for me, I like to build the trading process up from the concept, while data snooping is more like building the process and trying to work backwards to determine a concept.

 

Add,

 

Thank you for clarifying that for me. It's a common saying in the quant community "I'll sell my kids before i'll sell my data and my kids are not for sale" - The type of data used for research an analysis plays a key role in any strategy.

 

I havn't used econometric methods yet in a trading model; however, I have been able to develop some robust models using simple MA's and RSI.

 

In regards to data snooping, i'm not sure if you have heard of Jaffray Woodriff but read up on him if you have not...

Share this post


Link to post
Share on other sites
Add,

 

Thank you for clarifying that for me. It's a common saying in the quant community "I'll sell my kids before i'll sell my data and my kids are not for sale" - The type of data used for research an analysis plays a key role in any strategy.

 

I havn't used econometric methods yet in a trading model; however, I have been able to develop some robust models using simple MA's and RSI.

 

In regards to data snooping, i'm not sure if you have heard of Jaffray Woodriff but read up on him if you have not...

 

I have heard of jaffray woodriff, but only to the extent which he was covered in Hedge fund market wizards.

Share this post


Link to post
Share on other sites

what they now call regime changing - we have always known as switches from trending to non trending. market profile is a good example of this as well. it is very important can't stress it enough, probably one of the leading causes of immediate failure.

 

mark

Share this post


Link to post
Share on other sites
Hi Suby,

 

Notice how five trades drop out over the period shown? Over the course of the test period this is sufficient to have the following effect:

 

BlueHorseshoe

 

 

very nice - would it be possible to put both those screen shots on one screen so i could better line them up and tell the differences? it really looks good.

 

mark

Share this post


Link to post
Share on other sites
very nice - would it be possible to put both those screen shots on one screen so i could better line them up and tell the differences? it really looks good.

 

mark

 

Hi Mark,

 

It's several months since I took these shots, but I think the differences should be obvious to pick out - trades 8,10,11,16, and 17 for the basic strategy are removed by the filter. The filter is not affecting where the entry is made, remember, but whether an entry is made at all.

 

Remember that I have cherry-picked a screenshot that looks good there - what is more important is the change to the performance metrics I quoted. Without knowing a bit more about the strategy it is difficult to know whether the example I give is meaningful - if no stop-loss is used, for example, then the filter need only remove a single losing outlier during the ten year test period to have a significant impact on performance. But this would be luck, and nothing else.

 

The point of the example was to give anyone interested a starting point to begin doing thorough research of their own.

 

Hope that's helpful.

 

BlueHorseshoe

Share this post


Link to post
Share on other sites

I am especially interested in your screen shots because they are similar in style to something I have been working on as well. Multi data source hidden, variable contract sizing, regime dictated, finessed entry trades.

 

I have also obtained "like you" an honest 5 to 1 profit factor from a model that was originally about a 1 to 1.75 or so (all on 20yrs daily). Just kept at it and logically refining it just as I am sure you have. Congrats!

 

Mark

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.