Topic: (Closed) Butter Bot!: Premier Bitstamp, and BTC-E EMA Trading Platform (Closed) - page 65. (Read 274808 times)
November 10, 2013, 02:58:51 PM
I meant if it automatically summarized it at the top like it does the min profit...
November 10, 2013, 02:53:02 PM
I know weve been talking about new features we'd like to see, and i for one love the backtest engine. Howabout a forwardtest engine?
Seriously, It would be nice if the backtest engine had a little more info like number of losing vs winning trades, so we could also try to see if we could find settings that at least in theory minimized the number of trades where we just turned around and bought it back slightly higher.
Seriously, It would be nice if the backtest engine had a little more info like number of losing vs winning trades, so we could also try to see if we could find settings that at least in theory minimized the number of trades where we just turned around and bought it back slightly higher.
But, it does. You can scroll down and see every trade, win & lose, & by how much.
November 10, 2013, 02:49:13 PM
I know weve been talking about new features we'd like to see, and i for one love the backtest engine. Howabout a forwardtest engine?
Seriously, It would be nice if the backtest engine had a little more info like number of losing vs winning trades, so we could also try to see if we could find settings that at least in theory minimized the number of trades where we just turned around and bought it back slightly higher.
Seriously, It would be nice if the backtest engine had a little more info like number of losing vs winning trades, so we could also try to see if we could find settings that at least in theory minimized the number of trades where we just turned around and bought it back slightly higher.
November 10, 2013, 02:35:22 PM
Ah... you are using double gates... so in affect, we are doing the same thing, just different ways. With double gates, your poll time is 30 min, but trades are at 1hr (has to pass through 2 gates @ 30min each). I am running a single gate @ 1hr.
End result, same.
End result, same.
well, yes and no. the end result is that your settings sold at $288, whereas mine sold at $318. that's almost 10% difference.
Further examination of the charts uncovers another reason you had a better sell point then I did. You sold slower... later... with your slower gate setting (0.09 v 0.01).
Nice work.
November 10, 2013, 02:30:40 PM
Ah... you are using double gates... so in affect, we are doing the same thing, just different ways. With double gates, your poll time is 30 min, but trades are at 1hr (has to pass through 2 gates @ 30min each). I am running a single gate @ 1hr.
End result, same.
End result, same.
well, yes and no. the end result is that your settings sold at $288, whereas mine sold at $318. that's almost 10% difference.
Oh, sir, be so very careful not to get wrapped up in that number. Our settings could be identical, with tick-offset 5 minutes apart and have vastly different buy/sell numbers.
It is a simple matter to go back and "pick" the moments it was optimum to buy and sell. It is not possible to predict the next optimum moment.
Thus, it is important to establish a good solid strategy, and stick with it, regardless of the luck factor of exactly when the order falls.
Refer to my suggestions on experimenting with tick-offset.
Ours is a perfect example. We are both trading the same EMA, both trading at 1hr interval. The gate setting is different (mine is faster), but the bulk of the difference in trade point was the tick-offset; i.e. luck.
November 10, 2013, 02:20:01 PM
Ah... you are using double gates... so in affect, we are doing the same thing, just different ways. With double gates, your poll time is 30 min, but trades are at 1hr (has to pass through 2 gates @ 30min each). I am running a single gate @ 1hr.
End result, same.
End result, same.
well, yes and no. the end result is that your settings sold at $288, whereas mine sold at $318. that's almost 10% difference.
November 10, 2013, 02:17:36 PM
looking at my tests, if I go to 1 hour, I actually have a higher percentage profit, because the bot bought earlier (at $175), but also sold earlier ($308)
at 30m, he bought at $195, sold at $318.
so, what you want is the 1 hour buying for you and the 30 minute selling for you.
at 30m, he bought at $195, sold at $318.
so, what you want is the 1 hour buying for you and the 30 minute selling for you.
November 10, 2013, 02:12:00 PM
Ah... you are using double gates... so in affect, we are doing the same thing, just different ways. With double gates, your poll time is 30 min, but trades are at 1hr (has to pass through 2 gates @ 30min each). I am running a single gate @ 1hr.
End result, same.
November 10, 2013, 02:08:28 PM
Your mileage will vary depending on your tick-offset, but here is the 1hr v 30 minutes for the last 3 months:
There is a significant penalty for anything under 60 minutes in all the simulations I run.
When I go back manually to see why, it comes full circle to that all important momentum timing.
Sean
There is a significant penalty for anything under 60 minutes in all the simulations I run.
When I go back manually to see why, it comes full circle to that all important momentum timing.
Sean
November 10, 2013, 02:08:13 PM
November 10, 2013, 02:06:55 PM
He's already posted his numbers before.
See a couple of posts back on this page.
See a couple of posts back on this page.
November 10, 2013, 01:57:51 PM
I like the 1hr poll times, but to slow the bot back down I use the 20/42 EMA. EMA is not about "agile", it is about "timing".
The 20/42 EMA results in a greater spread and makes it slower to cross. This eliminates a lot of trade noise. I then use very fast gate timing (-0.2/0.01) so that when it does cross, it trades. The crosses are much sharper with 20/42 settings as opposed to the more traditional 10/21.
The 20/42 EMA results in a greater spread and makes it slower to cross. This eliminates a lot of trade noise. I then use very fast gate timing (-0.2/0.01) so that when it does cross, it trades. The crosses are much sharper with 20/42 settings as opposed to the more traditional 10/21.
I like 20/42, but with 30m. It seems to perform better, especially with big/quick drops, like this weekend. Your 1 hr setting sold at $288, whereas the 30m sold at $320.
I think going back to 1hr might be better for times with less volatility.
That is very curious. Would you mind posing your numbers? Dropping to 30 min always shows a significant hit in my tests:
November 10, 2013, 01:45:04 PM
I like the 1hr poll times, but to slow the bot back down I use the 20/42 EMA. EMA is not about "agile", it is about "timing".
The 20/42 EMA results in a greater spread and makes it slower to cross. This eliminates a lot of trade noise. I then use very fast gate timing (-0.2/0.01) so that when it does cross, it trades. The crosses are much sharper with 20/42 settings as opposed to the more traditional 10/21.
The 20/42 EMA results in a greater spread and makes it slower to cross. This eliminates a lot of trade noise. I then use very fast gate timing (-0.2/0.01) so that when it does cross, it trades. The crosses are much sharper with 20/42 settings as opposed to the more traditional 10/21.
I like 20/42, but with 30m. It seems to perform better, especially with big/quick drops, like this weekend. Your 1 hr setting sold at $288, whereas the 30m sold at $320.
I think going back to 1hr might be better for times with less volatility.
November 10, 2013, 12:38:41 PM
Have you read this? http://www.theverge.com/2013/11/7/5077856/google-for-windows-to-prohibit-browser-extensions-not-in-chrome-web-store
Chrome will not allow manual extensions installations anymore
Chrome will not allow manual extensions installations anymore
I read it - but from what I understand you'll still be able to use the dev-versions to install these extensions. If they really make it impossible through Chrome (which would be very bad and dissapointing) it'll probably still be possible through Chromium / Comodo Dragon
November 10, 2013, 11:40:52 AM
Hmmmm ... maybe it's time for me to move to BTC-E.
The uber-low fees make it my favorite.
I'm going to wait for an upward trend to start and then transfer my bitcoins over - 0.2% vs 0.6% is a huge difference!
Thanks for the settings; a 1hr frequency is a little bit more 'agile' than the 2hr one I've been using until now.
I like the 1hr poll times, but to slow the bot back down I use the 20/42 EMA. EMA is not about "agile", it is about "timing".
The 20/42 EMA results in a greater spread and makes it slower to cross. This eliminates a lot of trade noise. I then use very fast gate timing (-0.2/0.01) so that when it does cross, it trades. The crosses are much sharper with 20/42 settings as opposed to the more traditional 10/21.
November 10, 2013, 11:32:10 AM
Hmmmm ... maybe it's time for me to move to BTC-E.
The uber-low fees make it my favorite.
No, recommend Coinbase as a portal.
November 10, 2013, 11:18:44 AM
Hmmmm ... maybe it's time for me to move to BTC-E.
The uber-low fees make it my favorite.
I'm going to wait for an upward trend to start and then transfer my bitcoins over - 0.2% vs 0.6% is a huge difference!
Thanks for the settings; a 1hr frequency is a little bit more 'agile' than the 2hr one I've been using until now.
November 10, 2013, 11:17:18 AM
Hmmmm ... maybe it's time for me to move to BTC-E.
The uber-low fees make it my favorite.
November 10, 2013, 10:55:08 AM
Hmmmm ... maybe it's time for me to move to BTC-E.
The uber-low fees make it my favorite.
November 10, 2013, 10:49:29 AM
Have you read this? http://www.theverge.com/2013/11/7/5077856/google-for-windows-to-prohibit-browser-extensions-not-in-chrome-web-store
Chrome will not allow manual extensions installations anymore
Chrome will not allow manual extensions installations anymore
Jump to: