wat
I taught myself how to backtest futures systems in closer to 5-10 minutes, and for free. I haven't come up with a winning strategy yet, but the backtesting is the easy and cheap part. Forward testing live is when it gets expensive.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Back testing futures
- Thread starter meanreversion
- Start date
wat
I taught myself how to backtest futures systems in closer to 5-10 minutes, and for free. I haven't come up with a winning strategy yet, but the backtesting is the easy and cheap part. Forward testing live is when it gets expensive.
Finally someone making sense.
I'm not even sure why those two bothered to reply to my initial post if they had nothing to say.
I've been using Pinnacle data, which seems ok so far, although I heard CSI is meant to be good, any thoughts on that?
I'm not even sure why those two bothered to reply to my initial post if they had nothing to say.
I've been using Pinnacle data, which seems ok so far, although I heard CSI is meant to be good, any thoughts on that?
I've used Pinnacle Data for EOD data since the mid 90s. Solid vendor.
http://www.pinnacledata.com/
The also have lots of market internals, which I find very useful.
http://www.pinnacledata.com/
The also have lots of market internals, which I find very useful.
IT is "easy" to backtest but it is "hard" to interpret the results to your benefit.
Quote from meanreversion:
Finally someone making sense.
I'm not even sure why those two bothered to reply to my initial post if they had nothing to say.
I've been using Pinnacle data, which seems ok so far, although I heard CSI is meant to be good, any thoughts on that?
I used CSI for 3 years with Tradingblox, together they worked without any problems. I attempted CSI with Amibroker and that was a hassle. A few thoughts on CSI
1. End of day only
2. Rollover logic is adjustable by date, OI and volume or any combination.
3. Tech support has One guy whose good, the other few were not.
Ah, it wasn't just me then. I attempted to use CSI with Amibroker but it wouldn't work with the latest version (5.30). An engineer from CSI remote accessed my machine for an hour or so but couldn't work it out. In the end I abandoned CSI and went with Pinnacle .. getting that data into Amibroker is easy. Only thing is, I had no idea if Pinnacle were any good or not, but judging by comments I've seen so far, they're decent.
Quote from meanreversion:
Ah, it wasn't just me then. I attempted to use CSI with Amibroker but it wouldn't work with the latest version (5.30). An engineer from CSI remote accessed my machine for an hour or so but couldn't work it out. In the end I abandoned CSI and went with Pinnacle .. getting that data into Amibroker is easy. Only thing is, I had no idea if Pinnacle were any good or not, but judging by comments I've seen so far, they're decent.
Yeah I gave up after 6 months of being the Ping Pong ball between CSI and Amibroker, I go so fed up with "it's there fault, not ours" that I blamed the whole thing on me, both parties seemed ok with that...
I've done most of my backtesting on FX spot, and now I've got the data, am comparing system results between spot and futures. So far, the results are kind of similar, but where testing on spot falls down is lack of carry input into the P/L. Futures already have interest rate differentials built into them, but there are presumably other issues with backtesting, and that's why I started the thread.
Take a breakout system - buy if price breaks highest high for last 10 days, for example .. that's fairly simple in spot, but when transferring this method to futures, is it necessary to shift the spot high by the pip differential between spot and futures? Or can you simply analyze the continuous futures contract itself, and leave static orders in the lead month based on 10 day highs/lows for the continuous contract?
I'm guessing it doesn't make a whole lot of difference, i.e. just leave static orders..
Take a breakout system - buy if price breaks highest high for last 10 days, for example .. that's fairly simple in spot, but when transferring this method to futures, is it necessary to shift the spot high by the pip differential between spot and futures? Or can you simply analyze the continuous futures contract itself, and leave static orders in the lead month based on 10 day highs/lows for the continuous contract?
I'm guessing it doesn't make a whole lot of difference, i.e. just leave static orders..