[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [amibroker] Re: My First System



PureBytes Links

Trading Reference Links

Hi Bill,

Sunday, February 4, 2007, 2:41:25 AM, you wrote:

BH> After responding to Brianzee123's email, I ran a back test on a
BH> Defined Filter of a Watchlist of the 1500 stocks that are in the
BH> S&P 3 main Indices; S&P-500, MID-Cap and Small-Cap.
BH>    
BH>   The back test produced a heafty $18K+ return, so without
BH> changing any variables, I ran the same back test on "All" Symbols
BH> (I have TeleChart's database) and came up with only a $1300
BH> return. The primary profit producer in the S&P list was DAKT, but
BH> it didn't show up in the run from "All" Symbols.
BH>    
BH>   I don't understand why that hapopened. Can anyone explain it to me?

Do you have enough initial equity to backtest "All" Symbols?  (I
didn't notice what backtest mode you were using.)  If you don't, AB
will, just as you would be forced to do in real life, buy until it
runs out of buying power. Then it will stop buying.

Moreover, if you don't specify a position size, it will use *ALL*
your initial equity on the first buy.  So you need to specify
position size, position score (which symbols would have priority),
probably maximum number of open positions, and other elements.

If you do not set your backtests up correctly, the results will be
horrendously misleading.  They might indicate a gold mine that simply
isn't there, or a complete failure that isn't quite that bad.

Last, total return in currency value doesn't mean anything, really.
"Hefty" is very subjective, and certainly would not describe an
annual return of US$ 18K on a million dollars of initial equity.  (It
would probably be the end of your hedge fund management career.)  ^_^

When you mention return, talk in percentages; then we all can relate
to it -- and it has some meaning. Also mention drawdowns. If you had
to risk 36K of drawdown to get that 18K, well that wouldn't be a very
good deal then, would it?  Talk percentages there, too.

Speaking only for myself, just put your trial system in a list
e-mail, rather than forcing everyone to take the time to visit the
files area, locate your file, and read it.  If it gets to be a
monstrously long system, maybe that's another thing, then.

Yuki


Content-Description: "AVG certification"
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.5.432 / Virus Database: 268.17.22/666 - Release Date: 2/3/2007 3:31 PM