I have been backtesting using CME currency (ease of use) whereas in my actual trading I am using cash FX (although I could use CME if needs be). I know this breaks the golden rule (trade what you test, test what you trade...). Anyway I have a couple of issues;
1) When i substituted cash FX for CME FX in my backtesting I got quite different results. Enough to make me question the foundation of what I was doing. I am doing a detailed comparison of the trade logs to hunt it down but this may take some time. My question is, shouldnt the results under either method using a simple MA system be very similar? Has anyone else found this issue?
2) In choosing whether to trade CME of Cash I have read some comments on the site. I am still interested in other users comments as I cant really see the benefit of CME over FX. But that is just me.....surely liquidity and costs mean you have to go the Cash route.
Kind Regards
Jim
CME currency contracts versus cash FX market? Which one?
-
blinkybill
- Roundtable Knight
- Posts: 196
- Joined: Thu Mar 27, 2008 6:44 pm
Maybe the sample data that Blox supplies, could shed some light? For those currencypairs that are tradeable as "Forex," and also as "Futures" (such as GBPUSD, USDCHF, and so on), available in the Blox sample data, ... do you get the backtest results that you expect?
It would remove one variable from the equation, namely, the data vendor and data setup within Blox. It is probably a pretty safe assumption that the Blox sample data and its setup in Blox software, is not brimming with errors.
It would remove one variable from the equation, namely, the data vendor and data setup within Blox. It is probably a pretty safe assumption that the Blox sample data and its setup in Blox software, is not brimming with errors.
-
svquant
- Roundtable Knight
- Posts: 126
- Joined: Mon Nov 07, 2005 3:39 am
- Location: Silicon Valley, CA
- Contact:
As you investigate futures vs spot keep in mind the following
- Differences in the close time from the different data sources - they maynot be the same.
CME has the yield differential built-in while spot does not. Need to have good interest rate data and make sure it is accounted for in your cash fx work assuming you hold positions for a bit.
A google search for the phrase: "currency futures vs forex" yields over 2 million results.
From a very quick scan of the results on page 1 & 2 , it seems that those online companies that specialise in Cash Forex trading all recommend . . . . <fanfare please> . . .Cash Forex Trading!
Those that deal with Futures reckon that Futures are the way to go (surprising that!)
Here's a 4 web-page article that appeared in the April 2008 issue of Futures magazine
http://findarticles.com/p/articles/mi_q ... _n25136958
No prizes for guessing which way Futures Mag tilts on this matter, but still an interesting article regardless.
From a very quick scan of the results on page 1 & 2 , it seems that those online companies that specialise in Cash Forex trading all recommend . . . . <fanfare please> . . .Cash Forex Trading!
Those that deal with Futures reckon that Futures are the way to go (surprising that!)
Here's a 4 web-page article that appeared in the April 2008 issue of Futures magazine
http://findarticles.com/p/articles/mi_q ... _n25136958
No prizes for guessing which way Futures Mag tilts on this matter, but still an interesting article regardless.
-
blinkybill
- Roundtable Knight
- Posts: 196
- Joined: Thu Mar 27, 2008 6:44 pm
Sluggo
I decided to take up your advice and ran the following test on sample data (5 major currencies ex Euro) from 1995 to present on the DMA simple system as provided with TBX download. The results were;
Stepped Parameter Summary Performance Test Ending Balance CAGR% MAR Modified Sharpe Annual Sharpe Max Total Equity DD Longest Drawdown # Trades
FX 2,035,466.91 5.18% 0.14 0.39 0.12 36.8% 56.3 296
CME 2,277,903.25 6.02% 0.16 0.48 0.19 37.1% 61.8 252
These are quite similar results although number of trades is quite a bit different. I am happy with this. However I then decided to run the same test on my own data download from CSI on the same system. The results were as follows;
FX 2,100,694.45 5.44% 0.15 0.39 0.13 37.2% 56.3 294
CME 1,786,121.26 4.23% 0.12 0.35 0.07 36.2% 61.9 293
Whilst the CME numbers were a little different, perhaps due to differences in the CSI download format from my data to the sample data the overall results still seem within expectations and the number of trades now match!
But now lets do the same test on data back to the start of 1987 where the AD starts trading and see if the relationship holds;
data to come...
I decided to take up your advice and ran the following test on sample data (5 major currencies ex Euro) from 1995 to present on the DMA simple system as provided with TBX download. The results were;
Stepped Parameter Summary Performance Test Ending Balance CAGR% MAR Modified Sharpe Annual Sharpe Max Total Equity DD Longest Drawdown # Trades
FX 2,035,466.91 5.18% 0.14 0.39 0.12 36.8% 56.3 296
CME 2,277,903.25 6.02% 0.16 0.48 0.19 37.1% 61.8 252
These are quite similar results although number of trades is quite a bit different. I am happy with this. However I then decided to run the same test on my own data download from CSI on the same system. The results were as follows;
FX 2,100,694.45 5.44% 0.15 0.39 0.13 37.2% 56.3 294
CME 1,786,121.26 4.23% 0.12 0.35 0.07 36.2% 61.9 293
Whilst the CME numbers were a little different, perhaps due to differences in the CSI download format from my data to the sample data the overall results still seem within expectations and the number of trades now match!
But now lets do the same test on data back to the start of 1987 where the AD starts trading and see if the relationship holds;
data to come...
-
blinkybill
- Roundtable Knight
- Posts: 196
- Joined: Thu Mar 27, 2008 6:44 pm
I havent posted the data here but what I notice is that the impact of carry is much higher in the period 1987-1991 and for some reason this leads to a much bigger difference in the results in that period. I would have though that the futures contracts would have accounted for the carry but it doesnt seem to be the case. Will keep digging.
aa ha essentially it comes down to a data issue. I should have thought of this but takes some nutting out. Prior to about 1995 FX data consists mainly of purely closing prices. hence when you calculate ATR is probably distorted to the downside by less volatile price movement (i.e close to close less volatile than either close to close or prev close to low etc)..this totally distorts the number of contracts one can buy for a given risk and hence impacts the carry!!! I reran the numbers based on a fixed % of equity allocated per trade and the results are much more similar..
aa ha essentially it comes down to a data issue. I should have thought of this but takes some nutting out. Prior to about 1995 FX data consists mainly of purely closing prices. hence when you calculate ATR is probably distorted to the downside by less volatile price movement (i.e close to close less volatile than either close to close or prev close to low etc)..this totally distorts the number of contracts one can buy for a given risk and hence impacts the carry!!! I reran the numbers based on a fixed % of equity allocated per trade and the results are much more similar..
It sounds as though you are gradually "peeling the onion" and getting some Eureka Moments during the process. Congratulations!
In the area of "reminding you about things you already know": Some of the builtin trading systems that ship with Blox, use ONLY the close prices, so they are adapted to instruments that only report one price per bar (a Close, rather than an Open-High-Low-Close). Such as mutual funds, spot commodities (cash trading), and perhaps, old FX data. Maybe they might be another useful tool in your investigations.
In the area of "reminding you about things you already know": Some of the builtin trading systems that ship with Blox, use ONLY the close prices, so they are adapted to instruments that only report one price per bar (a Close, rather than an Open-High-Low-Close). Such as mutual funds, spot commodities (cash trading), and perhaps, old FX data. Maybe they might be another useful tool in your investigations.