HORRIBLE day for the system...System got burnt bad after the fed minutes..... nq: -$204 er2: -$1348 es: -$133.60 ym: $130.18
05-25-06 06:03 PM method made $60 today. Tomorrow I am thinking of possibly running the method on ym, nq, and es and see how it does... only issue is this bogs down my primary computer too much...... *** 05-26-06 03:25 PM Today the system was ran on ES, NQ, and YM. I just recently turned the system off: Here are the P/L after commisions. ES +74 NQ -423 YM -110 *** 05-30-06 03:55 PM Today I ran the slightly improved system on the NQ, ES, YM, and ER2 NQ: +$95 per contract YM: -$232 per contract ER2: -$234 per contract ES: +$355 per contract *** 05-31-06 04:19 PM HORRIBLE day for the system...System got burnt bad after the fed minutes..... nq: -$204 er2: -$1348 es: -$133.60 ym: $130.18 *** Just humor me bro. Reverse your Buy/Sell parameters. What now constitutes a "Buy Signal", turn it into a "Sell Signal". What now is a "Sell Signal", turn it into a "Buy Signal". If you reverse the parameters, you should reverse the numbers listed above. Best, Jimmy
reversing the method will not reverse the numbers.... the problem lies in the exit criteria. the vast majority of trades go positive generally VERY positive only to have it all given back and turn negative before the trade is complete.... Reversing the entry criteria will more than likely result in MUCH larger losses as the trade will normally be very in the red from the start and eventually stoped out
Out of curiosity I decided to backtest the method with the entry criteria reversed, just as expected it had a net loss of 158,000 over the testing period, and had a net loss EVERY quarter
OK, OK. How about my suggestion of locking in a profit (say at a decent $200 per contract ES, or the relative point spread for any/all of ther other indices) and then brining the rest to breakeven plus commission and slippage? This is how I actually trade, and it's a very basic and very profitable for discretionary trading (that's because the markets have a tendency to make their profits and THEN REVERSE). As your current system shows, they don't just constantly trend (at least I think that is what it shows, but I can't see the logic behind the code, so I'm just guessing). Anyway, give it a shot. Best, Jimmy
jimmy, Just like always my bot system is up nearly $700 right now.... but by time it actually takes profits on these trades it will only lock in maybe $200 of it.... thanks for your suggestions. I decided to test a few different PT scenarios: These tests were run on the ES data. #1 Benchmark, system as it is today -Net +101,411.000 -Draw down: -3320.00 -Max days loose in row: 10 #2 Same method, only added take profits at same distance as stop: -Net +99,636.000 -Draw down: -2960.00 -Max days loose in row: 7 #3 same method, take profits smaller than stop -Net + 96,862.000 -Draw down: -2840.00 -Days in row: 8 #4 same method, take profits significantly smaller than stop -Net +92,988.00 -Draw down:-3230.00 -Days in row:8 Win % MUCH higher nearly 75% in this method. #5 same method, take profits significantly smaller than stop -Net +62,190.00 -Draw down:-2572.50 -Days in row: 7 Winning % was insanely high #6 This one uses same pt as #2 but stop is made smaller. -Net +87,592.00 -Draw down: -3547.50 -Days in row:8 #7 his one uses same pt as #2 but stop is made even smaller. -Net +54,106.00 -Draw down: -2917.50 -Days in row: 10 #8 this one uses a smaller stop and smaller PT -Net 52,106.00 -Draw down: -2370.00 -Days in row: 8 #9 this one uses a smaller stop and smaller PT -Net 88,585.00 -Draw down: -3027.50 -Days in row: 8
I'm sorry if this has been touched on already.....are you utilizing some sort of "Look-inside-the-bar" option in your analysis ?
Hey Frost, "I ain't made at'cha!" To be quite honest with you, I wish I could program like you can! Then I wouldn't have to arduously write down my rules, back-test them, commit them to memory and then every day make discretionary decisions, while making sure every trade I take "falls within the parameters of the rules". This is how all the live traders HAVE to do it. You've progressed way beyond that stuff. Now, just trade in multiples of two, implementing option #5 on the first set of contract(s), while moving the protective stop in place on the second set of contracts so that you can take advantage of any trend that develops. In six months you'll be skiing (or traveling in Europe, or whatever you WANT to be doing), while your "money tree" is bearing fruit (of course you'll be monitoring it with your WIFI system). ha ha ha Best to you, Jimmy
AC3, that is a valid question. In my earlier days of developing automated systems I did have a few systems that suffered from that problem. Since then I have put in many fail safes into my code to prevent that sort of thing from happening. This system here I am 100% certain it does not use any sort of "look inside the current bar" analysis. Thus far all of the "real" trading of the system is inline with the expectations of the system. The system was mainly developed for the ES and that part is functioning very inline with how it should. The other markets such as YM no backtest were done on that market. So I do not know what to expect. But the NQ and ES results are inline.