Net breakdown (contracts traded):
ZS $229(4), GCL $582(4)
ZS $229(4), GCL $582(4)
| RESULTS FOR DAY | |
|---|---|
| Contracts: | 8 |
| Net $P/L: | 812 |
| Wins: | 2 |
| Losses: | 1 |
| Win%: | 67 |
| Avg$Win: | 486 |
| Avg$Loss: | -160 |
For the week, very happy with this. Not having any losing crude system trades this week certainly helped. I should note, that most of the crude trades have been in my IRA trading account which I have not traded prior to this week in a very long time.
Finished backtesting my latest CL system that I posted about on Monday. Imagine how much higher my confidence would be if I had not gone back any earlier than mid-Aug. '08? But I did look at all data back to March '07 so now I have to figure out if the market has changed since Aug. '08 and I should continue trading this or if I'm being fooled by randomness yet again. You might recall I've been down this road a few month ago. Something for me to think about over the weekend. Any feedback welcome!
Posts
hey
ReplyDeletei read your blog from time to time. I use your blog as a getway to read blog of traders who trade the futures :-P
that chart looks great in 08 and on. I think even a simple moving average crossover would do well in that period. But how long is the drawdown period in 07? You gotta be honest with yourself and ask if you would've stayed with the system being in the red for that long.
Phil McGrew (currensys.com) knows allot about developing systems and I found something in one of his posts from a forum that may be relevant to you.
ReplyDeleteFrom Forum:
{I'm not sure about the meaning of your last sentence - "If the walk forward is valid you should see around 50% of the optimised results on an annual adjusted basis", could you please clarify a bit more?
"You optimize a curve. Let's say you do it for 1 year and it makes $10,000. You then walk forward for 3 months and those 3 months yield $1000. On an annualized basis (comparing apples to apples) $1000 (per 3 months) is $4000. So the optimized makes $10,000 per year and the unseen makes $4000 per year or 40%. My point was that you will rarely exceed 50% of the optimized amount. If you approach 50% you should be very impressed with your efforts."}
Thanks for comments.
ReplyDeleteHilmy, I don't even have to ask myself - there's no way I would have traded that system in 2007 and early 2008. That drawdown was well over a year to return to new equity highs. As for MA crossovers, never seen any of those produce any kind of consistent results. Not that they aren't out there, I just haven't seen any.
Harvey, thanks for link- I'll check it out. I'm very aware of over optimizing and as a result this system is not optimized at all. It's a very simple strategy actually. But the fact that it just didn't work in 2007-2008 is worrisome. But then I'm trading the 2010 market, not the one from a few years ago...
Personally, I can't live with those kinds of oscillations of up and down. M1-R (you can read it) is 70% win rate, only the first 25% was used to create the parameters. It does quite well forward testing.
ReplyDeleteI'll be watching your execution on your new system to see how you withstand those down cycles. I personally, with my own money, just can't stomach down cycles as readily. I've spent hours working around avoiding those kinds of things just to keep my psychological state in a good place.
Are you using some kind of trend following/breakout style system?
ReplyDeleteTC, yes you could classify it as a breakout system. I don't like big oscillations either on the downside but love them on the up side. Can you have one without the other?!
ReplyDeleteWhat do you mean by "the first 25%"? I'm guessing that might mean you only used 1/4 of your available data to optimize your system?
Yes, sorry. I was writing so fast I was skipping words - 1/4th of the data to train.
ReplyDeleteYou can have upsides without huge downsides. Depends on your models. Tell me more about how you think breakouts should be classified. I haven't really gotten any non stat-arb systems to work well.
Those curves you have look good, but I'm curious whether a computer could qualify something as a breakout. In the world of stocks, you can almost always never get liquidity on the fast, volatile breakouts unless you have a really good execution system. OEC and TradeStation do not qualify as good execution systems.