|
PureBytes Links
Trading Reference Links
|
Delengo,
Believe me, I have lots to learn, too. To run that in the old backtest, I found out that you really need to comment out all the portfolio trading code, then you get CAR that are more reasonable 13% with 7% DD. I didn't do that when I originally posted, and consequently the IWM test numbers werw wrong. Fortunately, you did the due dilligence.
You are right on, RUTVOL is a trend-based system, and has gotten beat to death these past three months. The trick is to find an indicator that will tell you when things are beginning to chop rather than trend. Unfortunately, ADX is lacking.
I'll tell you the best thing you have going for you is the computer between your ears. If you notice there was a trade in August that was short-lived. Just a short thrust with a retracement that got us into cash. This should've been a warning since we just finished the mother of all moves from mar to july. Maybe pear back position sizes a bit. I still take the trades, to keep me honest, but I trade will smaller size.
Fred has made some great posts about incorporating discretion into system trading. They are worth your time to read through.
Hope this helps, and let us know what more you find.
Regards,
Garydalengo <dalengo@xxxxxxxxx> wrote:
Gary,Many thanks for your comments. I've now changed onlyMaxPos = Optimize("Max Positions",1,1,15,1);//MaxPos = Optimize("Max Positions",5,1,15,1);(and commented out the Indicator) to have 100k available for the code (I am using AB 4.40.4).My results seem to be much more optimistic: [52% total return/account], 16% CAR, MDD=-12%. Actual equity ended up at $137,634 (# of lots?), so the real 3 year return (with 0.5% slippage) is 38%.I would like to know if we are getting similar, or different results, and what might cause the difference.________________________System: I need to play with the modifications you mentioned but a couple of things seems obvious: (i) the system is lagging quite a bit, and missed the mother of all bottoms in Oct'02, (ii) very recently, it tended to buy high and sell low, giving back 14k out of
36k hard-earned on a bull run Apr-Jul'03. The point (ii) is really frustrating, and needs a correction for market choppiness, at least in terms of %portfolio management, I guess.Thanks for invitation to come to AB conf - I enjoy learning from people's talks, it's fastest for me, - and I am grateful for your comments. Not sure I can make it, but still. -D. 8-D.________________________Report:Total net profit: 52651.02 Total commissions paid: 0.00 Return on account: 52.65 % Open position gain/loss 0.00 Buy&Hold profit: 12880.00 Bars (avg. days) in test: 726 (1054) Buy&Hold % return: 12.88% System to Buy&Hold index: 308.78% Annual system % return: 15.78% Annual B&H % return: 4.28% System drawdown: -1720.00 B&H drawdown: -27810.00 Max. system drawdown: -20187.00 B&H max.
drawdown: -40150.00 Max. system % drawdown: -11.80% B&H max. % drawdown: -35.42% Max. trade drawdown: -13024.00 Max. trade % drawdown: -7.61% Trade drawdown: -9184.00 Total number of trades: 11 Percent profitable: 63.6% Number winning trades: 7 Number losing trades: 4 Profit of winners: 70326.01 Loss of losers: -17674.99 Total # of bars in winners: 215 Total # of bars in losers: 50 Commissions paid in winners: 0.00 Commissions paid in losers: 0.00 Largest winning trade: 36244.00 Largest losing trade: -5114.99 # of bars in largest winner: 90 # bars in largest loser: 10 Commission paid in largest winner: 0.00 Commission paid in largest loser: 0.00
___________________________________________________Scan from 01-01-01 to 11-21-03 :IWM Sell 13.02.2001 99.85IWM Buy 27.04.2001 96.90IWM Sell 17.05.2001 100.76IWM Buy 15.10.2001 85.40IWM Sell 28.11.2001 90.05IWM Buy 10.12.2001 94.11IWM Sell 26.12.2001 97.50IWM Buy
12.03.2002 99.24IWM Sell 13.03.2002 98.40IWM Buy 30.04.2002 101.30IWM Sell 10.05.2002 98.13IWM Buy 25.10.2002 74.44IWM Sell 11.12.2002 78.65IWM Buy 21.03.2003 74.77IWM Sell 29.07.2003 94.85IWM Buy 29.08.2003
99.21IWM Sell 26.09.2003 96.60IWM Buy 13.10.2003 105.45IWM Sell 24.10.2003 101.25IWM Buy 04.11.2003 107.44IWM Sell 18.11.2003 103.76--- In amibroker@xxxxxxxxxxxxxxx, "Gary A. Serkhoshian" <serkhoshian777@xxxx> wrote:> Dalengo,> > I hope Tomasz reads this as my response below will prove that I actually paid attention and read his posts on the AmiBroker Beta board : )))))))))))))> > I'm glad that you took the time to ask these questions as they
are critical to understand, and as in prior posts those not familiar with portfolio tester open up your ears and listen in as this is basic stuff elaborated in the read-me and if you ask Tomasz he will find you and club you over the head for not reading the read-me : ) Lemme save you the headache, and hopefully save Tomasz the time.> > Let's tackle your e-mail in three parts:> > 1. Why you are not getting results when using old backtester and RUTVOL on cash Russell 2K index.> > 2. Why are arrows showing up in the equity graph for the old backtester.> > 3. Your basket testing returns compared to mine.> > Okay #1:> > The reason you are not getting results is driven by settings in the code governing initial equity, position size, min size, and round lot size.> > Let's go through each of
these in the order they occur in the code.> > > SetOption("InitialEquity", 100000);> > Straightforward. This is cash we start with. Why code it versus using the GUI? Because different systems require different captial requirements. This allows you to customize.> > RoundLotSize = 100;> > All trades must occur in blocks of 100 shares. I think this is fair. It you wanna try odd lots, good for you. I don't.> > SetOption("MinShares",100);> > Minimum size must be 100 shares. You'll see how this is important in a sec.> > > > MaxPos = Optimize("Max Positions",5,1,15,1);> > SetOption("MaxOpenPositions",MaxPos);> > Allow 5 open positions at one time if cash is avail. I've given you the ability to optimize this via MaxPos variable.> > >
> PositionSize = -100/MaxPos;> > If there is a " -" in from of the number that means we are designating percent. In this case positon size for one position can not exceed 20% of total portfolio equity.> > > > Okay, here's the punchline: Trading the RUT cash index around 400 or so> > 400 x 100 min share size = $40,000 position which is bigger than $20,000 position size limit. So the position never gets put on.> > If you want to run this in the old backtest, comment out the positionsize line, and it will run. You'll want to increase initial equity, too. Run it on the IWM to be realistic.> > > > 2. Regarding why you are getting arrows in the old backtester equity curve is that the equity curve for the old backtester is taking all the code from the aa window and including in the equity line
code. This means the indicator code at the bottom of the system code actually plots buy sell arrows which makes the equityline indicator hard to read. If you are going to run this in the old backteser, comment out the indicator code. Be careful not to comment out anything important : )> > 3. Okay now the meat and potatoes. The returns. Don't get worried about the returns, because I'm going to show you how to make them better.> > 1/01 - 11/03 trading IWM (no portfolio trading):> > CAR: 7.27> > MaxDD: -8.23%> > > > Same Period basket trading with BB selection:> > CAR: 5.25%> > MaxDD = -28.9%> > Now for some detail. We must keep in mind that this is a long only system over what was primarily a bearish period. The basket trading return was horrible. The
important message is the Bollinger Band selection method is not the best. It was an example to get you going, but in the spirit of more code and less talk (an idea which I hope all of our more verbose contributors will espouse...you know who you are) let's show you the ncAlpha selection method:> > Same Period basket trading with ncAlpha selection:> > CAR: 49%> > MaxDD = -22.79> > Woah. Not bad, eh? CAR went way up and we reduced maxdd. At least we are in >2 MAR country which Fred has taught me is a good place to be. Again this was an all long system over a lot of bear market moves. There are better signals out there, but RUTVOL isn't a bad place to start.> > Here's the code for ncAlpha to play with (early X-Mas present so be nice till Jan 1 ; )> > //GLOBAL SYMBOLS> > INDEXSYM = ParamStr("INDEX
SYM","!RUT");> > RAWINDEX = Foreign(INDEXSYM,"C"); //CLOSING PRICE OF INDEX> > RAWFUND = C; //CLOSING PRICE OF FUND> > LB = Param("LOOKBACK PERIOD",126,0,1008,10);> > //NC ALPHA> > //COMPUTE DAILY STDEV & APPLY MONTHLY CONVERSION FACTOR> > SDF = StDev(ROC(RAWFUND,1),LB) * sqrt(22);> > SDI = StDev(ROC(RAWINDEX,1),LB) * sqrt(22);> > RELSD = SDF / SDI; //RATIO IS SAME REGARDLESS OF WHETHER WE USE DAILY OR MONTHLY SD VALUES> > NCALPHA = MA(ROC(RAWFUND,1),LB) - RELSD * MA(ROC(RAWINDEX,1),LB); //DAILY ncALPHA> > I need to get back to work so I'll spare you a definition of ncAlpha, but go to Werner's site, and you'll get a better definition than I can ever give.> > Also, I'd strongly encourage you to get out to Clearwater because there will be a good presentation by Bruce Robinson on blending a signals like
RUTVOL with simple to complex filters to create better long signals, and tradeable short signals. I'll also be running through a lot of these signals in terms of analyzing them through the portfolio backtester with various baskets of stocks as well as optimization analysis. www.ftmonitor.com has details.> > Hopefully I've left enough for you to play with.> > Hope this helps,> > Gary> > Send BUG REPORTS to bugs@xxxxxxxxxxxxxSend SUGGESTIONS to suggest@xxxxxxxxxxxxx-----------------------------------------Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx (Web page: http://groups.yahoo.com/group/amiquote/messages/)--------------------------------------------Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
Do you Yahoo!?
Free Pop-Up Blocker - Get it now
Yahoo! Groups Sponsor
ADVERTISEMENT
Send BUG REPORTS to bugs@xxxxxxxxxxxxx
Send SUGGESTIONS to suggest@xxxxxxxxxxxxx
-----------------------------------------
Post AmiQuote-related messages ONLY to: amiquote@xxxxxxxxxxxxxxx
(Web page: http://groups.yahoo.com/group/amiquote/messages/)
--------------------------------------------
Check group FAQ at: http://groups.yahoo.com/group/amibroker/files/groupfaq.html
Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
|