Just for the record, and for the sake of clarity. I've found 3 sources so far for gap fading techniques that I'd consider legit. (1 ebook and 2 articles) 1 suggests a 6 point profit target and a 17 point stoploss. One suggests entering the trade at 9:35 AM and using a time stop of 9:55 AM with a profit target of gap closure. The third, an article by John Carter (google him or google YM gap strategies) suggests gap clossure as a profit target with a stoploss of 1.5:1. ie if the gap is 20 the stoploss is 30 pts. And he leaves the trade all day and sometimes overnight. These are the varriables I'm trying to make sence of.
If we could test on longer than six months of data, it would help. If anyone has any more 5-min data on the minidow, please let me know. Ewile, I tried all you mentioned (closing the gap, etc.), but on the small data set I have it doesn't work as well as what I have now. Here's the system as it is now: {*** entries ***} If time=1530 and c > c[1] and c < average(c,100) Then Begin sell this Bar at c; value2=c; End; {*** exits ***} exitshort at value2-14 limit; exitshort value2+12 stop; ------------------- Translation: If the close of the 9.30 bar (which happens at 9.35) is higher than yesterday's close and we are in a downtrend (the close is less than the average of the last 100 5-min periods), then go short and stay short until you either have a profit of 14 ticks, or a loss of 12 ticks. --- I think the system should stay this simple, but I wish I had more data. At least 2 years of 5-min data. For being this simple I think the results are good and it's not overoptimised: Total Net Profit $1.064,00 Gross Profit $1.320,00 Total # of trades 24 Number winning trades 20 Largest winning trade $66,00 Average winning trade $66,00 Ratio avg win/avg loss 1,03 Max consec. Winners 9 Avg # bars in winners 5 Max intraday drawdown ($114,00) Profit Factor 5,16 Account size required $114,00 --- Gross Loss ($256,00) Percent profitable 83,33% Number losing trades 4 Largest losing trade ($64,00) Average losing trade ($64,00) Avg trade (win & loss) $44,33 Max consec. losers 1 Avg # bars in losers 6 Return on account 933,33% --------------- Only problem - in 6 months it only makes 24 trades, so it only trades once a week. But if we could put together another 9 systems like this one, it would be very good I think.
I am done working on this system, because if we raise the stoploss to: 5*gap, it gets 100% of the 24 trades right. (Thanks for your mentioning those theories, it finally set in). Mission accomplished - system curve fitted and 100% profitable. Basically, I'd have to test it on a longer data set, because as I have it right now it gets 100% of trades right and makes 1500 dollars in six months with 1 minidow, with 24 trades: {*** entries ***} If time=1530 and c > c[1] and c < average(c,100) Then Begin sell this Bar at c; value2=c; value3=c[1]; value4=c-c[1]; End; {*** exits ***} exitshort at value3+value4/3 limit; exitshort value2+value4*5 stop; ------- I will certainly use it from now on. By the way, it works best if you sell as soon as it fills 2/3 of the opening gap. Also, important to remember, that we are not going short at 9.30, but at 9.35. I checked if it would have been better to go short at the 9.30 open, but it's the opposite. It's best to go short at 9.35 sharp.
I worked on it for a little longer and here it is: {*** long entries ***} If time=1530 and c < c[1]-c/333 Then Begin buy this Bar at c; value1=c; value2=c[1]-c; End; {*** long exits ***} exitlong at value1+value2*0.66 limit; exitlong value1-value2*1.2 stop; if time=1720 then exitlong; {*** short entries ***} If time=1530 and c > c[1]+c/333 Then Begin sell this Bar at c; value1=c; value2=c-c[1]; End; {*** short exits ***} exitshort at value1-value2*0.66 limit; exitshort value1+value2*2 stop; if time=1720 then exitshort; ------------------------ I think this time it's better. It finally made sense to have long trades as well short trades, though they don't do as well. But now they both make profits. There is no trend indicator, as I have tried to get rid of as much unnecessary stuff as possible for fear of overoptimisation. It makes 28 trades in 6 months, that is about one trade per week. This time the requirement is that the gap be at least 30 points, or it won't trade. That's what got the longs to be profitable. At this point, I think I am done. I will have to look for one big 30+ ticks gap at (my) 15.30 and go in the opposite direction, until it almost closes it, or until I lose twice as many ticks as the gap was wide. This pretty much sums up everything we have been saying about this opening gap on this thread. And it should give me about 100 dollars per trade (including losing trades). Once again, it happens about once a week. What's interesting is that, without even knowing about this gap thing, by testing it, I came to the exact same conclusion of those who talk about and trade the gap. Go in the opposite direction of the gap, and wait until it gets closed. But be prepared to get screwed by a breakaway gap, which will make you lose a lot of money two times out of ten. A lot more than you make by trading the gap, since the stoploss is larger than the gap.
John Carter's method of trading gaps is more than what you have mentioned... He goes in a piece at a time depending on the market situation. He looks at what the different sectors are doing with the banks, brokers and semi's the most important. And there is a lot more to it than what I have mentioned... Like pivots points, tick, trin, ect...
Here is a link to a webinar from John Carter and his YM gap strategy. http://www.cbot.com/cbot/pub/cont_detail/0,3206,1058+17036,00.html
So lets see: [*]If YM gap (use the 9:30 est open) >=30 ticks, fade the gap at 9:35 est. [*]Target = 2/3rds of the gap (round down) or 11:20 est [*]Stop = 5 times the gap. (round up) or 11:20 est Is that it? Trade once a week and in the last 6 months it has never lost? Michael B. P.S. I guaranty you the first time I trade it, it will reach its stop before I get my target!
FWIW, I got tick data for the YM for the month of April from CBOT. (They give you one month free). I went through the data by hand using my origional rules: Gap > 6 pts. Profit target 6 points Stop loss: Close trade @ 9:55 AM For some reason the data was missing one day, April 29th, so I couldn't do that day. 17 trades. +37 points. Figure 17 points to cover commissions. 4 losing trades. (26, 5, 6 and 10 points respectively). The 26 point loss was on 4/5. This is assuming that I didn't make any mistakes which is a pretty big assumption and also dependent on good fills etc. etc.
Curve fitting a system to make 24 trades profitable by setting a huge stop loss is really, really dumb.
At the Denver Trading Group on Thusday night (05/19/05 7:00pm mst-10:00pm mst), they will have a speaker (John Carter) talking about trading with a full time job. .....it's $15.00 at the Denver Marriot Tech Center You might fly in to check it out. Michael B.