Pages:
Author

Topic: Goomboo's Journal - page 13. (Read 281467 times)

sr. member
Activity: 378
Merit: 250
August 16, 2013, 02:49:12 PM
if I read this well, this model does not outperform buy and hold: only +643%

our newsletter has fared better :-) (had to throw this in :-))


If your newsletter is so great, why do you need to advertize everywhere in every topics ?

Stop spamming and make your newsletter public during 2 weeks, people will choose on real results.
newbie
Activity: 41
Merit: 0
August 16, 2013, 01:27:17 PM
Absolutely. I've run some tests and a threshold of .25 is the most profitable. It surprised me that the default setting was the best one for once.

EDIT: That's only under my trading conditions actually. You'd have to test for your specific EMA and time period.
newbie
Activity: 57
Merit: 0
August 16, 2013, 06:50:56 AM
OK, now I see the problem. I had no threshold and I didn't have enough initial candles. Now I get your result. Amazing how dramatically the threshold affects results.
newbie
Activity: 41
Merit: 0
August 15, 2013, 09:31:40 PM
I get 3415%. Odd. Something's amiss.
Indeed. Just to verify, daily candles, 100 initial EMAs, .6 fee, and .25 threshold, correct? Anything you can think of that may be skewing the numbers?
newbie
Activity: 57
Merit: 0
August 15, 2013, 08:33:12 PM
I get 3415%. Odd. Something's amiss.
newbie
Activity: 41
Merit: 0
August 15, 2013, 03:55:52 PM
With Gekko?
Yes, all the default parameters other than EMAs.
newbie
Activity: 57
Merit: 0
August 15, 2013, 03:26:27 PM
With Gekko?
newbie
Activity: 41
Merit: 0
August 15, 2013, 12:02:17 PM
Quote
12/29 daily gives 3456.193% profit, higher than any other EMA, using the time frame of the candles in this thread: https://bitcointalksearch.org/topic/downloadable-mtgox-historical-candles-239815

That's not as high as the 20/21

That's bizzare, 20/21 gives me 2312.392% profit on the same time frame.
newbie
Activity: 57
Merit: 0
August 15, 2013, 06:19:45 AM
Quote
12/29 daily gives 3456.193% profit, higher than any other EMA, using the time frame of the candles in this thread: https://bitcointalksearch.org/topic/downloadable-mtgox-historical-candles-239815

That's not as high as the 20/21
newbie
Activity: 41
Merit: 0
August 15, 2013, 12:42:35 AM
That's strange. In my tests the 12/29 EMA has been most profitable.

What numbers do you get?

12/29 daily gives 3456.193% profit, higher than any other EMA, using the time frame of the candles in this thread: https://bitcointalksearch.org/topic/downloadable-mtgox-historical-candles-239815
newbie
Activity: 57
Merit: 0
August 14, 2013, 10:23:40 PM
That's strange. In my tests the 12/29 EMA has been most profitable.

What numbers do you get?
newbie
Activity: 57
Merit: 0
August 14, 2013, 04:44:04 PM
Fair comment Smiley I just figured you already have it set up and only need to change the start date to get the answer, but maybe there is manual fiddling involved. Really appreciate you sharing what you already have. Thanks.

I do it all manually.
newbie
Activity: 10
Merit: 0
August 14, 2013, 04:20:35 PM
So having spent a couple of days reading the thread, I've taken on board to have only 1% of your investment in a given fund and realised that I need to look beyond bitcoins alone to apply these techniques.

I've also taken on board the need to be selling short in down trends and that if you don't do that, you're halving the potential profit. So, I've been looking for places to invest that allow short selling and have a wide range of things to invest in. CFDs keep coming up. These seem ideal for a small time trader like me as there are no trading fees, however there seems to be an "overnight interest payment" or "overnight premium" associated with them if you hold at the end of the day. Is it possible to use CFDs for medium term trend following, or is it really for day trading only? How about sellling at the end of each day and then buying again at the start of the next to avoid the overnight interest - or is that a crazy idea?

One good thing, a lot of the sites have unlimited "simulated" trading on their platform, which is free.

Separately:
I don't know, dude, why don't you pitch in a little and tell us?
 Grin

Fair comment Smiley I just figured you already have it set up and only need to change the start date to get the answer, but maybe there is manual fiddling involved. Really appreciate you sharing what you already have. Thanks.
newbie
Activity: 41
Merit: 0
August 14, 2013, 02:02:31 PM
That's strange. In my tests the 12/29 EMA has been most profitable.
newbie
Activity: 57
Merit: 0
August 14, 2013, 01:09:11 PM
I don't know, dude, why don't you pitch in a little and tell us?
 Grin
newbie
Activity: 10
Merit: 0
August 14, 2013, 07:29:02 AM
Hi, I'm new here and just caught up on this thread! Great thread.

More interesting backtest results. I compared Ichimoku, the 10/21 EMA, and the 15/24 EMA, looking at almost the full history of BTC trading, starting when one bitcoin went for a dollar, and ending today (1 BTC = $108.74). With the EMAs, I traded at the closing price of a candle which indicated a crossover, no threshold. With Ichimoku, I bought BTC when the price was over the Tenkan Sen over the Kijun Sen, and Chikou over price. I sold when the candle closed below the Kijun Sen. Trades were executed at the closing price.

Here are the results:

10/21 EMA P/L = 28,530%
Ichimoku P/L = 46,317%
15/24 EMA P/L = 50,028%

 Shocked

I share all this information gladly because I've learned so much from this forum.

Wow those are startling percentages! Regarding 15/24, this makes me think of what Goomboo said about curve fitting. It is not surprising that this optimal combination for June 2011 to January 2013 also gives the best results up till now, because there is a big overlap in the data. What happens if you remove the overlap (i.e. only look at Feb 2013 to now)?
newbie
Activity: 57
Merit: 0
August 14, 2013, 06:14:58 AM
Daily.
legendary
Activity: 2478
Merit: 1362
August 14, 2013, 04:29:21 AM
More interesting backtest results. I compared Ichimoku, the 10/21 EMA, and the 15/24 EMA, looking at almost the full history of BTC trading, starting when one bitcoin went for a dollar, and ending today (1 BTC = $108.74). With the EMAs, I traded at the closing price of a candle which indicated a crossover, no threshold. With Ichimoku, I bought BTC when the price was over the Tenkan Sen over the Kijun Sen, and Chikou over price. I sold when the candle closed below the Kijun Sen. Trades were executed at the closing price.

Here are the results:

10/21 EMA P/L = 28,530%
Ichimoku P/L = 46,317%
15/24 EMA P/L = 50,028%

 Shocked

I share all this information gladly because I've learned so much from this forum.

Thanks for sharing !

Which timeframe did you test for these results?
newbie
Activity: 57
Merit: 0
August 13, 2013, 08:51:18 PM
More interesting backtest results. I compared Ichimoku, the 10/21 EMA, and the 15/24 EMA, looking at almost the full history of BTC trading, starting when one bitcoin went for a dollar, and ending today (1 BTC = $108.74). With the EMAs, I traded at the closing price of a candle which indicated a crossover, no threshold. With Ichimoku, I bought BTC when the price was over the Tenkan Sen over the Kijun Sen, and Chikou over price. I sold when the candle closed below the Kijun Sen. Trades were executed at the closing price.

Here are the results:

10/21 EMA P/L = 28,530%
Ichimoku P/L = 46,317%
15/24 EMA P/L = 50,028%

 Shocked

I share all this information gladly because I've learned so much from this forum.
newbie
Activity: 41
Merit: 0
August 11, 2013, 06:11:46 PM
I'm curious, is it best to use data from the conception of bitcoin or only data from recent months? Using all available data allows for testing under various market conditions, but using recent data gives you an idea what's most effective in current market conditions. I'm seeing conflicting reports on the most profitable EMA/time period, so I'm having trouble deciding the most profitable strategy.
Pages:
Jump to: