| stevehopwoodforex.com https://www.stevehopwoodforex.com/phpBB3/ Print view |
|
| Gold/Silver Trading https://www.stevehopwoodforex.com/phpBB3/viewtopic.php?t=325 |
Page 1 of 2 |
| Author: | passgo [ Sat Feb 04, 2012 11:35 am ] |
| Post subject: | Gold/Silver Trading |
Let me say at the outset that this is not a well thought out trading system, more a series of ideas which I would like to test out. My task is to develop an expert advisor to optimise trading in the gold and silver markets based on the LBMA gold fix as set out at http://goldenlightswitch.blogspot.com/p ... rules.html: "Trading System Rules The rules are simple: - sell a gold futures contract at the AM fix to be short and cover at the PM fix - buy a gold futures contract at the PM fix to be long and sell at the next day's AM fix The name Golden LightSwitch is derived from the switching long to short, always being in the market, flipping positions like a lightswitch. The objective is to shadow the gold bullion banks as they fix gold prices." (I trust I am not breaking any copyright by reproducing here with full attribution.) I am no coder but I think it might be not too difficult for you wizards! While I have searched the web and downloaded a number of ea's, none of them really trade to these rules and none allow optimisation. To use an ea for this purpose I believe it is important to be able to vary day of the week, time and, ideally, probability by day, time + candle pattern...... (Timing is all - to improve probability there is a need to have time+event but cannot afford too much lag. As I type, I am thinking that it must be possible to formulate a spreadsheet that would provide the analysis to start to identify these coefficients. Will think on!) My last thought, for what it is worth, is that there could also be a way to use the ideas of parrondo's paradox ttp://www.eleceng.adelaide.edu.au/Groups/parr ... troduction. The main idea of Parrondo's paradox is that two individually losing games can be combined to win via deterministic or non-deterministic mixing of the games. My maths is too far ago to apply this paradox to trading, but I feel that the LBMA gold fix may in be an opportunity. |
|
| Author: | GT3RS [ Sat Feb 04, 2012 1:12 pm ] |
| Post subject: | Re: Gold/Silver Trading |
Hi, I have seen something similar mentioned on another forum. More info here> http://en.wikipedia.org/wiki/Gold_fixing I decided to have a quick look at this, because it needs some parameters throwing around it in order to test and develop. I just went for a 25 pip SL. Go short at 10:00 and go long at 15:30 UK time. Test is from start of January to date. I've accounted for a 5 pip spread in my results. Results are interesting. 46 trades 15 winners Strike rate = 33% Total point count = 1,359 [!!!!] but this includes a MONSTER deal of +613 pips. Even if you take this out of the equation, the pip score is VERY good, with an EXCELLENT risk/reward. Mind you, it needs to be, with a 33% strike rate!! Lol! Risking 2% of your account on every deal, this equates to a 105% increase in equity........ And you only have to look at the daily chart to see the uptrend going on for the whole of January! Whether results would continue in this manner is highly unlikely. This was a VERY rough and ready backtest, so don't hold me to these figures please!!!! You'd need balls of steel to trade this for real as there are frequent long losing streaks. Losing streaks of 4 trades are common. Longest losing streak was 7!!! Passgo - have you traded this kind of strategy yourself? If so, what parameters did you use? And how were the results? Maybe my little backtest will give someone with far more brains/experience than myself some ideas to refine further? Cheers!!! |
|
| Author: | garyfritz [ Sat Feb 04, 2012 3:09 pm ] |
| Post subject: | Re: Gold/Silver Trading |
I don't have access to XAUUSD with my crim so I tested it with CME futures data on Tradestation. London fixes are at 10:30 and 15:00 London time. The CME data is in US Central time. I used a 6-hour offset from Chicago time, shorting at 10:30 - 6 hours = 04:30 CST, and buying at 15:00 - 6 hours = 09:00 CST. There may be some Daylight Savings Time errors in there but I'm not sure. The results are pretty good, if a bit inconsistent. Here are the last two years: The profit is in "$ per oz." So in the last 2 years this system captured $600 per oz profit. (FWIW during this time gold went up about $700 per oz.) I got much better results with "fix times" times of 07:30 and 13:00 (00:30 and 06:00 Chicago time): 13:00 PM fix made 50% more profits than any other time, but any "PM fix" time of 9:00 - 12:30 made more money and had a smoother equity curve than the "proper" 15:00 London time. With "fix times" of 07:30 and 13:00 London time, you're spending almost 80% of your time long. That might not have worked so well if gold hadn't been in such a raging bull market lately. But gold topped out on 8/22/11, and the equity curve above doesn't top out until 11/1/11. So this system can pull some profits out even when gold is going down. This really only started working in early 2009. If you want consistent performance for the last 10 years, you should use 07:00 and 10:00: But once again you're only spending 3 hours a day short, and that probably wouldn't work as well if gold wasn't going ballistic the whole time. |
|
| Author: | passgo [ Sat Feb 04, 2012 4:14 pm ] |
| Post subject: | Re: Gold/Silver Trading |
HI RT3S and Garyfritz That was quick! I think to a degree you have hit on the core of why this works: the uptrend. I have not traded this system with the pure rules but I have been using it to time entries. The difficult part of trading gold (silver is a little less daunting) is taking the decision whether to ride the trend with all its ups and downs or harvest the profits. My idea is that using the switch is a way to harvest. The times you have identified are really interesting - I will have a closer look: just using my eye I had thought about 10.00 to 12.00 but not quite made the leap. My thinking was that if you could link the timing to some event during the 10.30 - 15.00 window, this would seriously increase the performance but shortening the window probably makes more sense - would it be worth building a pending order with a buffer zone? a confirmation signal? |
|
| Author: | GT3RS [ Sun Feb 05, 2012 9:17 pm ] |
| Post subject: | Re: Gold/Silver Trading |
I just tried start of Jan to present with a 50 point SL. Result in points is only slightly less, BUT, equity gain is considerably smaller due to smaller position sizes. Gary, the time of this is critical IMHO, otherwise you aren't trading the fix. Passgo - do you have any suggestions for filtering/entries? Cheers! |
|
| Author: | passgo [ Sun Feb 05, 2012 10:58 pm ] |
| Post subject: | Re: Gold/Silver Trading |
Have tried trend and oscillator indicators, but never found much use in gold trading. Two thoughts - use pending orders with a buffer and corresponding decrease in SL? Need an ea to test and identify level. Is it possible to backtest candle patterns as confirmation - bearish pattern for sell during am/pm fix, bullish pattern for long after 15.00? Will reduce TP but might improve overall performance. |
|
| Author: | garyfritz [ Mon Feb 06, 2012 5:56 am ] |
| Post subject: | Re: Gold/Silver Trading |
I understand my alternate times aren't trading the fix. But they seem to work better. |
|
| Author: | passgo [ Mon Feb 06, 2012 8:59 am ] |
| Post subject: | Re: Gold/Silver Trading |
Good morning - garyfritz's hours certainly working this am. Using a trailing stop - would this boost the % and/or change the optimum hours? Your questions given me some ideas which I hope to work on today and get back to you with some data - want to look at volatility by day as looking at the charts think there are some key differences. Also looking through my notes, we need to look at option expiry dates. |
|
| Author: | passgo [ Mon Feb 06, 2012 3:18 pm ] |
| Post subject: | Re: Gold/Silver Trading |
Stimulated by discussion looking at various sources of gold volatility data came across the goldfix data on the Kitco website. See below results of analysis for 2011: GOLD UP, GOLD DOWN COUNT 129, 125 AVERAGE 9.6, 7.0 MODE 3, 1 STDEV 11.11, 7.20 MIN 0.25, 0.25 MAX 80, 42 Doesn't support the gold switch trading system: so perhaps need to change objective away from the idea of the gold fix to a trading system that can take advantage of the daily volatility we see in gold and silver markets. Can we support the hours (7.30-13.00) thrown up by Garyfritz's analysis with an explanation of market behaviour that makes some fundamental sense? Has computer trading reduced the relevance of the gold fix? Are we seeing trading tending to start at the beginning of the day with position squaring and later the long term trend re-establishing itself? How is that correlated to the morning's USD/CHF trading? Mataf's gold hourly volatility chart shows the peak hours are 13.00-15.00 and Thursday has highest volatility. I want to test using 7.30 - 13.00 + opposite to trend confirmation for first trade, then 15.00 re-entry with long term trend confirmation. Can use Steve's multipurpose trade manager to lock in profit, trailing stop etc. |
|
| Author: | passgo [ Mon Feb 06, 2012 5:07 pm ] |
| Post subject: | Re: Gold/Silver Trading |
Think this shows the way forward to develop an expert advisor - anyone ready to give it a go? |
|
| All times are UTC | Page 1 of 2 |
|
Powered by phpBB® Forum Software © phpBB Limited |
|