tag:blogger.com,1999:blog-261139923818144971.post8286005772656612979..comments2024-03-27T07:58:49.946+00:00Comments on This Blog is Systematic: Random data: Evaluating "Trading the equity curve"Rob Carverhttp://www.blogger.com/profile/10175885372013572770noreply@blogger.comBlogger10125tag:blogger.com,1999:blog-261139923818144971.post-88987051547934396832021-08-07T13:03:41.734+01:002021-08-07T13:03:41.734+01:00You linked to it in your post, but essentially it ...You linked to it in your post, but essentially it was a collection of objective rules based around classical price patterns like double bottoms and retests. Unfortunately it was my first real attempt at building an automated trading system, so it was fragile, did very well for a year and then fell apart into a heap. These days I run a mix of trend (futures), momentum (stocks) and I adapted Andreas Clenow's momentum system for crypto (which has taken full advantage of the mania there, I've never had a win quite like it). Anyway, I'm sincerely indebted to you for writing your books. Seriously, without people like you and Andreas Clenow I'd still be doing stupid things which absolutely are guaranteed to blow me up eventually. I wish you long life, happiness and wealth. You are making a signficant contribution to my world.Scott Phillipshttps://www.blogger.com/profile/18059179119265640522noreply@blogger.comtag:blogger.com,1999:blog-261139923818144971.post-59072452757945466442021-08-07T07:05:19.046+01:002021-08-07T07:05:19.046+01:00Hi Scott. Remind me what the crazy ivan system is?...Hi Scott. Remind me what the crazy ivan system is??Rob Carverhttps://www.blogger.com/profile/10175885372013572770noreply@blogger.comtag:blogger.com,1999:blog-261139923818144971.post-36485399595537946902021-08-06T03:59:41.845+01:002021-08-06T03:59:41.845+01:00Rob, sincere thanks for your superb blog and books...Rob, sincere thanks for your superb blog and books. I had one of those embarrassing Dunning-Kruger moments when I first came across your stuff (hobby traders, amirite?). Anyway, I designed the crazy ivan system you mentioned (about 10 years ago when I was just starting to design proper systems) and it was, unsatisfyingly one of those ideas that looks great on paper and failed miserably in the real world. Turns out that particular system (and others) performs most strongly coming immediately out of drawdown and equity curve trading missed out that performance, which made it a poor idea, in risk adjusted and absolute terms. Of course comms were lower so it was better, but not enough to justify the complexity. I eventually rejected all these type of things as "polishing the turd"... ideas that were marginal and needed tricks to push them over the line of profitability. Suggest you delink it, I wouldn't want people getting the wrong idea and going down a dead end.Scott Phillipshttps://www.blogger.com/profile/18059179119265640522noreply@blogger.comtag:blogger.com,1999:blog-261139923818144971.post-85664516726974604392018-10-29T18:00:48.528+00:002018-10-29T18:00:48.528+00:00Yes, assuming you are only interested in whether t...Yes, assuming you are only interested in whether the other strategy produces longs or shorts, which you would then translate into forecasts of +10 and -10Rob Carverhttps://www.blogger.com/profile/10175885372013572770noreply@blogger.comtag:blogger.com,1999:blog-261139923818144971.post-86908582869807349692018-10-29T17:35:26.104+00:002018-10-29T17:35:26.104+00:00Hi Rob,
Supposing I have a list of transactions i...Hi Rob,<br /> Supposing I have a list of transactions in CSV format from running a strategy in a different programming language, i.e. strategy returns, which script in PST should I target for feeding these in? Forecasting.py?Chad Bhttps://www.blogger.com/profile/13026562498196984544noreply@blogger.comtag:blogger.com,1999:blog-261139923818144971.post-67455521298043250472015-11-18T16:49:50.461+00:002015-11-18T16:49:50.461+00:00Thanks Rob, I'll have to look into that. Haven...Thanks Rob, I'll have to look into that. Haven't tried anything using the drawdown as a filter. I suspect that it might not work, but my intuition is usually wrong about trading. Which is why I prefer robots and computers. These sort of of shortish MR trades tend to have many smaller wins, with the occasional invigoratingly large loss. A system that shuts off on drawdown thresholds might simply reduce the amount of little wins needed to scrape back into the black. But until I've tried it, I won't know.<br /><br />Thanks for your comments and your blog. I've added it to my regular reading list.Matt Haines Photographyhttps://www.blogger.com/profile/13895187802989669254noreply@blogger.comtag:blogger.com,1999:blog-261139923818144971.post-18515502494133866352015-11-18T06:23:08.939+00:002015-11-18T06:23:08.939+00:00Interesting. I guess a system where you cut after ...Interesting. I guess a system where you cut after the drawdown reached X% might protect you from a drawdown larger than you ever saw in simulation; i.e. the situation when the MR relationship breaks down.Rob Carverhttps://www.blogger.com/profile/10175885372013572770noreply@blogger.comtag:blogger.com,1999:blog-261139923818144971.post-15176282469391454112015-11-18T05:15:47.784+00:002015-11-18T05:15:47.784+00:00I've tried using a simple MA on win rates for ...I've tried using a simple MA on win rates for several mean-reversion systems, and the results have always been dismal. Except for that one time I forgot to check for a future leak, and that one worked like a charm! 😃 The problem, as I saw it, is that these systems all had a very high win rate, 60-75%, and so losing trades were few and far between. There was no clustering to speak of, so the switching on and off merely removed good grades. <br /><br />-Matt<br /><br />Http://www.throwinggoodmoney.comMatt Haines Photographyhttps://www.blogger.com/profile/13895187802989669254noreply@blogger.comtag:blogger.com,1999:blog-261139923818144971.post-83276439731515423272015-11-10T14:59:23.855+00:002015-11-10T14:59:23.855+00:00I can think of a few reasons:
Intepretability
I ...I can think of a few reasons:<br /><br />Intepretability<br /><br />I have been using Sharpe Ratios a long time, and I don't really have a feel for other performance statistics. If you told me something had a calmar ratio of 0.5 I wouldn't have a clue what you were talking about, and even if you explained that meant the average annual return to max drawdown was 0.5 I still wouldn't know if that was good or not.<br /><br /><br />Consistency<br /><br />I create forecasts as proportional to expected returns scaled for standard deviation, which is basically a Sharpe Ratio. I also calculate risk for position scaling by using the standard deviation. Consistently using standard deviation as a measure of risk has its flaws, but at least I'm using the same measure throughout my system.<br /><br /><br />Symmettry:<br /><br />Although the downfall of SR is that it assumes symmettry, I actually think it is a good thing to be as scared of an expected rise as a fall.<br /><br /><br />Statistical robustness:<br /><br />When we calculate the standard deviation we use the whole distribution. With average drawdown we don't and with maximum we use only a single point. Return / max drawdown in particular is a terrible statistic to use with real data, since it is very much random what it comes out at.<br />Rob Carverhttps://www.blogger.com/profile/10175885372013572770noreply@blogger.comtag:blogger.com,1999:blog-261139923818144971.post-16935376582076010822015-11-10T14:41:42.144+00:002015-11-10T14:41:42.144+00:00Another solid post, Rob. Why do you prefer SR over...Another solid post, Rob. Why do you prefer SR over the return/drawdown measure that was used in this post? Throughout the book, you use the SR to standardized risk but isn't the ret/dd ratio a more specific measure of that? Darrinhttps://www.blogger.com/profile/01441588918436763565noreply@blogger.com