|
PureBytes Links
Trading Reference Links
|
Michael E. Strupp wrote:
>
> Dear RealTraders -
>
> As a true trading systems nerd, I've been spending my hours tooling
> around with the QuickEditor on SuperCharts v4.0 with some end-of-data on
> CBOT Soybeans from Pinnacle Data. I've been playing with some very
> simple patterns, and emerged with some interesting results. This quick
> message is to share those results and invite any and all traders who
> have some ideas on how to improve this system to share it with me and
> the rest of the traders here.
>
> Again, this is a short-term system. Sell signals are the opposite of buy
> signals. All orders are to BUY at the market opening (I know, not always
> a lot of fun) and usually close out the day with a profit or hold on one
> day for a profit. Here is the system:
>
> TO BUY SOYBEANS TOMORROW ON THE OPENING:
>
> * Today's close has to be less than its open
> * Yesterday's close has to be higher than yesterday's open
> * The close two days ago had to be higher than the open two day's ago
> * Today's close has to be the lowest close in the last nine days
>
> In EasyLanguage, these criteria to buy at the market are:
>
> Close < Open and C[1] > O[1] AND C[2] > O[2] AND C > Lowest(Close,9)[1]
>
> TO EXIT POSITION AT TODAY'S CLOSE:
>
> * Today's close must be higher than today's open
> * The position is one day old and we close below today's open
>
> In EasyLanguage, these criteria to sell at the market close are:
>
> (BarsSinceEntry(0) > 0 AND C < O) OR C > O
>
> Again, the sell signals are the exact reverse of the buy signals. I
> have attached an .ela file of the system for those people out there with
> SuperCharts v4.0 or TradeStation 4.0 to tool around with.
>
> Now, I may be extremely naive of trying to test a soybean trading system
> based on entry and exit rules that rely on the volatile open and close,
> but this is what I ended up with as results:
>
> Soybeans - CBOT-Daily 01/02/69 - 04/24/97
>
> Performance Summary: All Trades
>
> Total net profit $ 94881.50 Open position P/L $ 0.00
> Gross profit $ 263680.50 Gross loss $-168799.00
>
> Total # of trades 1421 Percent profitable 73%
> Number winning trades 1040 Number losing trades 381
>
> Largest winning trade $ 2000.00 Largest losing trade $ -2250.00
> Average winning trade $ 253.54 Average losing trade $ -443.04
> Ratio avg win/avg loss 0.57 Avg trade(win & loss) $ 66.77
>
> Max consec. winners 19 Max consec. losers 6
> Avg # bars in winners 0 Avg # bars in losers 1
>
> Max intraday drawdown $ -5987.50
> Profit factor 1.56 Max # contracts held 1
> Account size required $ 5987.50 Return on account 1585%
>
>
> These results were obtained by using $1000 profit, $1000 protective stop
> and $200 breakeven point, trading 1 contract.
>
> Now, I realize that 1585% return over 26 years is not stunning given the
> recent performance of the stock market, and that $235 per winning trade
> for beans is about 4.5 cents per trade per contract (skippity doo) BUT I
> was somewhat encouraged by the percentage of winning trades (73%)
> overall, both long and short, given the high number (almost 1500 trades
> over 6250 trading days, or one trade about every four days).
>
> My question for everyone out there is -- given these results and ANY
> familiarity you may have with trading soybean futures, HOW WOULD YOU TRY
> TO IMPROVE THIS SYSTEM, IF IT CAN BE IMPROVED AT ALL (i.e., if there is
> any hope)?
>
> Or, since this system relies on trading buys or sells at the market open
> and close, are these results, in your opinion, a fiction, given
> slippage, etc.?
>
> Any feedback would be greatly appreciated!!!
>
> Michael Strupp
> Chicago, IL
>
> ---------------------------------------------------------------
>
Michael,
First, I commend you on your willingness to share your complete system
and ideas with the group. Hopefully, that willingness to learn will pay
off.
I system trade Beans also so I'll just give you my opinion on a few
things.
First, I see what your basic trading idea is and there is nothing
inherently wrong with it. In other words, it does make some logical
sense. That is one of the first basic rules when you develop a system.
Next, beware of your data. Call Pinnacle and find out what they use for
their opening price. Some vendors use the first price traded, others
use an average of the opening range. It makes a difference. Especially
beware of the old data (60's and 70's). They used different methods of
price reporting then and because of rule changes and technology changes,
those prices may not be correct. But they will be close, so a good
system will still be a good system.
That brings me to the next point. IMHO, your 28 year test has little
relevancy (this is not necessarily the common view). Because of many
reasons, today's trading IS fundamentally different from trading back
then, even in the Beans(ie government price supports, advent of commod
funds, quicker dissemination of info and prices, etc)
Because of this you need to test shorter periods in a rolling
technique. i.e test 69-74 then71-76 then 73-78 then 75-80, etc. You
should even do this in increments smaller than 5 years. What you are
looking for is similarity of performance in each period. If they are
similar, you may have something that is good. If not, look closely at
the price action in that period to honestly assess what happened. Again,
this is my opimion, but I put more weight on more current results. No
matter what you read, it is a rare system trader who leaves his system
unchanged for more than a couple of years. They change them because
markets do change.
Finally, on your numbers. You seem to be focusing on the average win.
I believe you should focus on the average trade. I don't know what you
have entered for comm and slip; but if it is anything less than $100, I
fear your system would fail. I get the impression that you have entered
no slippage, if that is true, this system is not good at $66 Average
Trade.
Yes, 73% winners is not bad, but as you test you will find that it is
not that difficult to find (in testing only) short term systems with
80%+ winners. Profit Factor is also important but 1.5 is not great. And
you would like to have a larger biggest winner than the biggest loser
but it is not necessary.
Again, if the more recent numbers are better, definitely try to work
with this idea.
My last comment is unfortunately this, take it how ever you want to. I
have NEVER met another CTA or system trader who believes that systems
that use profit targets work.
Good Luck,
Eric
|