Equity Curve Switches

[CanOz]

I'm quite interested in this method to manage a portfolio of strategies. This one is quite elaborate, it can be done manually, or coded as an indicator.

Very little really written about these.

I use an indicator based switch for my intra-day strats.

Cheers,


CanOz

 

[Gringotts Bank]

 

[CanOz]

Having trouble grasping the concept Gringo?

CanOz

 

[Gringotts Bank]

Having trouble seeing out of my burqa actually

I have found the link now... "method"!

 

[tech/a]

I used to customise a chart using the constituents of my portfolio.
I always bought the same $ value of shares so weighting wasn't an issue (It can be).
This gave me a chart of my continuous equity curve.
I just added and subtracted constituents when they were bought or sold.

I could then plot what every I wanted on the chart.
That was back when I was trading Techtrader.
I didn't use it as a filter--- I wasn't that bright!---arguably not a lot has altered.

I'm using just data and you can create a chart with whatever constituents you want.
Im sure you can do it with Premium data as well. I have that also but not currently using it.
Richard dale would be able to put you on track.

The idea is that if your EQUITY chart throws out a sell or EXIT signal you can do a number of things which you can test. Just reference that chart in your testing.
You keep the portfolio chart going not altering your constituents unless they throw out a sell in their own right.

While in your portfolio trading--You could
(1) Stop taking any buy signals until the curve shows a return to a buy ( Could be as simple as an M/A or whatever conditions you like.)
OR
(2) Sell the whole portfolio and re buy ---(those which didn't trigger a sale after you were switched out of the portfolio) when the switch goes back on.

Kris and I are getting closer to answering this and a whole lot more questions in the coming year.
There is a lot of work going on behind the scenes which we want to put out there to the technical community.

Hope this snippet is helpful
Thanks for the link.

 

[CanOz]

Thanks tech, I guess my interest at the moment is not so much for EOD as it is for intraday. I think the index switch or filter works well for EOD systems but for an intraday system that relies on the market to be in a certain mode, I.e.trend or counter trend, then the equity curve switch seems to be one way to moderate losses when the system is not as effective. Of course it also moderates the wins as well.
I like the fact that it can get you out of a system that becomes ineffective, as they all eventually do.

Unfortunately I have not seen a good example of the net effect on equity that the switch has had.

Cheers,


CanOz

 

[CanOz]

I found this written by Joe Krutsinger on that same site.

Cheers,


CanOz

 

[Monkeyzu]

https://www.aussiestockforums.com/forums/showthread.php?t=10805&page=1

 

[howardbandy]

Here are some of my thoughts:

http://www.blueowlpress.com/WordPress/code/equity-curve-feedback/

Thanks,
Howard

 

[CanOz]

Thanks Howard, basically the same as my research to date has shown as well regarding intra-day systems. A new article here as well...

CanOz

 

[howardbandy]

Thanks Howard, basically the same as my research to date has shown as well regarding intra-day systems. A new article here as well...

CanOz

CanOz' link points to an article by Michael Bryant. Dr Bryant's work is always worth reading.

Howard

 

[Gringotts Bank]

Howard, can that code work on portfolio testing or only individual stocks? Thanks.

 

[howardbandy]

Howard, can that code work on portfolio testing or only individual stocks? Thanks.

The thread seems to have two topics --
1. Using changes in the account balance to exit from existing trades or permit new trades. This is commonly called "trading the equity curve." It can be done using either daily equity changes or account balance based on closed traded.
2. Using changes in the account balance to change position size. This is commonly called "position sizing." It is usually done using balance of closed trades.

Trading the equity curve can add value to a system that is already pretty good. But it seldom rescues a poor system.

Either can be applied to individual issues or to portfolios. There are complications whenever there is serial correlation in the data series. If a single issue is being traded and closed trades are being analyzed to determine position sizing, if there is serial correlation, the best way to address that is to include code in the trading system to recognize the serial correlation and adjust as necessary so the resulting trade list no longer has serial correlation. If the serial correlation is minor, this is probably not worth the effort and the effect of correcting for it is minor.

When there is a portfolio, be aware if trades come in clusters, as they often do with mean reversion systems. If the system buys, say, 10 day new lows, and the portfolio consists of stocks in a 200 stock index, many issues will receive buy signals at the same time, and there is clearly serial correlation. It may be possible to work around this by using changes in daily equity. But even that might not work if the system is cherry-picking and is in cash a significant portion of the time.

If you believe that there is security in having a portfolio (which I do not believe), and that the distribution of trading results is stationary (which it never is), then some of Ralph Vince's work might be helpful. "The Leverage Space Trading Model" is his latest and summarizes much of his earlier work.

But beware of relying too heavily on portfolio construction. Safety through diversification is an illusion. In times of crisis people sell whatever they can sell, not necessarily what they want to sell, and everything drops together.

No set of trading results has a stationary distribution. The position sizing method used must be able to adapt dynamically.

My recommendation is to develop one good model that trades one highly liquid index long / flat / short. To both maximize account growth and minimize drawdown, trade frequently with short holding periods. Trade that index or some derivative of it. If necessary (primarily to increase trading frequency which is one of the keys to account growth), add another system.

I hope this helps,
Howard

 

[tech/a]

My recommendation is to develop one good model that trades one highly liquid index long / flat / short. To both maximize account growth and minimize drawdown, trade frequently with short holding periods. Trade that index or some derivative of it. If necessary (primarily to increase trading frequency which is one of the keys to account growth), add another system.

Every now and then a snippet of great value is placed on these boards.

If your having difficulty profiting in these non trending markets here is your answer.

Thanks Howard.

 

[Gringotts Bank]

Thanks Howard. I've never had any luck developing a profitable strategy for an index ETF. What would you consider a average/good/excellent % annual return? Do you know of anyone selling fully disclosed systems for say SPY with solid returns and low DD?

 

[howardbandy]

Thanks Howard. I've never had any luck developing a profitable strategy for an index ETF. What would you consider a average/good/excellent % annual return? Do you know of anyone selling fully disclosed systems for say SPY with solid returns and low DD?

Hi Gringotts --

Look into the systems published in Connors and Alvarez' book, "High Probability ETF Trading." You can get the book from Amazon or at the Connors website: www.tradingmarkets.com/products/books.

They are mean reversion, short holding periods, high winning percentage, easily programmed into almost any platform. Or use Chris White's ETF Trading Bandit from his EdgeRater website:
http://www.edgerater.com/products/etftradingbandit/
ETF Bandit has the seven strategies implemented, has an easy download of free data, gives backtest results, and gives live signals.

I know some of the people at Connors and some of the people at EdgeRater. Their work is solid. I have posted analysis of one of the Connors strategies on my blog site:
http://www.blueowlpress.com/WordPress/trading-systems/mean-reversion-based-on-rsi/
Also read the comments related to that article.

A question that often arises related to mean reversion systems is exits. Trend following systems have a self correction mechanism built in. Mean reversion systems do not. In itself, that does not make one better or worse than the other. Read my article on "testing stops" and the comments that follow it:
http://www.blueowlpress.com/WordPress/articles/testing-stops/#comment-116

I hope this helps,
Howard

 

[bomberone1]

Folks by multicharts how can we work on equty curve?
Have you got any sample of trading system that show how to do that?
CanOz, for example form a sample of moving average traDING SYSTEM HOW CAN YOU GET the stat in real time and import them in the strategy?

A complete example with comment it's very appreciate.

Many thanks.

 

[Gringotts Bank]

I found a displaced DEMA applied to the equity curve works much better than even a low lag Jurik MA.

Also I found an equity curve and its switch is easier to read if I backtest over a shorter time frame, say 6 months, taking all trades. Even though I wouldn't normally take all trades, the main purpose is to have a eq curve that is smooth and a switch that doesn't whipsaw. This seems to achieve what I want.

 

[tech/a]

I found a displaced DEMA applied to the equity curve works much better than even a low lag Jurik MA.

Also I found an equity curve and its switch is easier to read if I backtest over a shorter time frame, say 6 months, taking all trades. Even though I wouldn't normally take all trades, the main purpose is to have a eq curve that is smooth and a switch that doesn't whipsaw. This seems to achieve what I want.

GG

Unless you have enough capital to take all trades then your defeating the purpose. In other words its useless.

I simply make up an un weighted composite index of my portfolio (when Im trading portfolios) {You do this through your data manager Just data has this facility Im sure Premium Data would as well} and use that chart.
Of course when the equity curve switches you off the chart continues week in week out.
The constituents either drop out and new ones are added or they stay open until the switch is on again (Whatever criteria you have to switch off the portfolio). Ive used ATR and a simple long term trend line Ive also used the same exit criteria as the system constituents.

You buy or sell the entire portfolio.