Interesting thread. Grids are for mean reversion really, just another method of doing that. Of course it will get killed after a while, that's why you need trend following strategies in your folio as well, that should be your second account. Too bad you don't want to backtest it, i'm not sure whats holding you back but ok.
TY braincell for reading and the kind post. I was in hopes that I could get some participation and someone would backtest and post (90% or better quality). You see I am not selling anything here and I am not in denial either. I agree with your assumptions. I would require a third party backtester to post. I do not want the responsibility. Someone please post a backtest that has 10 years of "accurate" data in the MT4 history folder and set the readers of this thread straight. I hope that the readers here are gaining knowledge of market movement that this system illustrates and I hope you have taken something to heart with you. This is not a win or lose thread...it is an exercise of framing market movement. I have learned that some of you that I have a lot of respect for are lurking in this thread...We are all waiting for one thing....BLOOD IN THE STREETS! ES
I see, you don't want to be seen advocating anything. Well, I have the data but I never used MT4, so I'm not even sure if there's a free version that would enable me to test it like that (with offline data). Is there? I might take the time. Anyway, I think it's almost a fact that grids work best on noisy instruments with small range. For example the VIX derivatives (stocks or futures). I'm sure you could get some much better results trading those on a grid. Also the index ETFs are good. Forex has one of the biggest trending capabilities of any market type so it's really among the worst choices for a grid. I have no idea why they specifically created this EA for that, maybe they didn't quite get it. Also I think the best use of a grid is when you can decide (discretionary) that the market is not about to make big trendy moves for a while. I'm not sure it makes sense to run it 24/7/365.
braincell, Thank You for understanding. I am first a trader then a moderator...but when its my own thread, I must be neutral. I am very glad to volunteer here in ET to do my part in the community. Latley my responsibilty has increased as a moderator and it is like a breath of fresh air to come in here to my thread as a trader! However, I try to be sensitive to NOT fill the inboxes of the subscribers to this thread with endless updates. I am thankful to the readership of my thread, but I do not want to abuse it. I have a tendency to get on my soapbox and debate endlessly. How am I doing? Have I improved? LOL... MT4 is free, but testing with offline data and those technical questions that you bring up I do not know the answers to. Many of the Retail Spot Forex Traders use Alpari Forex Data. I am told that MT4 is a very sloppy backtesting tool and to not rely upon it for anything other than to see if your code works correctly...lol..its like a spell check tool in other words. I agree with you about using this grid on a noisy instrument with a narrow range. I believe a grid can be calibrated for any instrument and any range, if the incremental "gridspace" is adjusted for historical volatility. However your point of CAPACITY FOR VOLUME AND TRENDING AND VOLATILITY" is duely noted. Now...finding the sweet spot for the instrument while maintaining a neutral directionaless strategy is illusive indeed. "Putting the Market in Can" may require grid-spacing that reduces NAV to just 5% APR. There is always a time in history that widens the spacing to a point where it is not worth the risk as you must lock in the max range. I have deduced that it is actually better to start with less money and add-to or remove funds as needed. I suppose one could try to anticipate noise vs. trend and adjust up or down the capital....but again, this should be a directionaless venture. braincell, Thank-You for your most excellent points and your time for putting thought into this. ES P.S. This particular illustration might end prematurley because there is only a "static 4K" used. I am suggesting for those of you that are grid-fans that you can calibrate on a daily, weekly or monthly basis how much capital you tie up. Naturally, you would need one of these offshore dealers that use a credit card for deposits and withdrawals that are fee-free. I suspect that during a max drawdown year (or two years) that you might need to tie up 10K per account. Naturally when you are moving money between grids and accounts you need to understand your exposures, which braincell hinted at. You can manage millions this way in pieces. P.P.S. The money managers that I have worked with seem to meet their demise with a perfectly volatility-neutral type strategy by thinking that they have to pick a direction. I put to you professional traders out there...is it really worth taking on a direction to make that yield? I would rather net 10-15% APR for my investors after labor for decades than to be a flash in the pan. I am trying to illustrate 30% APR in this thread....but it needs 10K not 4K...so do the math to figure your APR ok? There is a spreadsheet somewhere to project APR with variable balances...money managers do it all the time. P.P.P.S. By the way I do not manage OPM, so do not even ask. I am not interested to lose money for others ☺
by the way braincell...anybody who can code as you in assembly language, well... I have a lot of respect for (yes, I do read your posts). ES
It would be very interesting if somebody could perform a test for 1 year running only grids on a broadly diversified portfolio, like 100 instruments. I have a "hunch" that the results would be very good if the grid is lucky enough to auto-hedge at the right time and the right price for multiple symbols - if you know what i mean, or if somebody had enough time to construct a careful folio and grids that would correlate in a statistically significant way (whatever time frame). Hehe and i get this on a trading forum? Thanks, it's great to know the language, but it would be even better if it had more practical uses. Besides, learning assembly isn't as hard as most people think it is, really. It's good to know somebody reads the stuff i post and possibly takes something from it, even if a contrary view. Then again, you're just being nice 'cause youre a moderator.
sooo true...sometimes it sucks...but assembly language gets to the heart of the thing. ES Then again, you're just being nice 'cause youre a moderator.
This would be brilliant! but only a dream for me. It would be very interesting if somebody could perform a test for 1 year running only grids on a broadly diversified portfolio, like 100 instruments. I have a "hunch" that the results would be very good if the grid is lucky enough to auto-hedge at the right time and the right price for multiple symbols - if you know what i mean, or if somebody had enough time to construct a careful folio and grids that would correlate in a statistically significant way (whatever time frame).
Pretty sure that's what the user 'fullautotrading' does in futures. Read his posts. http://www.elitetrader.com/vb/search.php?s=&action=showresults&searchid=4063482
Actually he only forward tests it. G-bot is a kind of grid system, but only looks at specific ranges. It requires discretionary input. The user 'fullautotrading' never really backtested 1 year of data. However, he gave me permission to use his software and do it myself. It is apparent that the limited backtest ability it has is not realistic because you cannot implement discretionary decisions. But my results were very positive, running a large number of stocks. However the ending PnL was usually only 10% of max drawdown value. This is a huge difference, which can be controlled with some discretionary input. So it's still a dream to do it for 1 full year with a very large folio and get published results. Maybe he does do it eventually, but i know he has been using G-bot for 2 years now (and others) and is profitable overall.