Topic: Goomboo's Journal - page 26. (Read 281462 times)

Is this typical? When I first used it and didn't have my account linked it had a nice display of the history there. Now it's blank. It's also been stuck on 0.17 for quite some time now.

Thanks!

What're your thresholds if you don't mind me asking? Also you should keep in mind this is through a very volatile period, as activity settles out you may see smaller and smaller returns.

Glad you have learned from the thread - best of luck and hello to you too!

This thread helped me prototyping a very raw trading bot (for Bitstamp).
Here's its behaviour so far, while simulating 1 BTC trading over past 90 days.
It does miss a lot of the spiking, but over 100% gain isn't too bad, with few loss (4) vs gain (11) orders.

90 Days Bitstamp 1BTC trading Report (SIMULATION)

Now trying to improve it.

Agree, best thread in the speculation subforum.

Hi Goomboo,
Budhist says hello to Christian . Three past evenings I was reading this Journal. Every comment from every user and it was quite a ride I have to tell.

Red and green arrows where to buy/sell, EMA, beautiful colourful tables, predictions about BTC when it was much weaker and vision of it's crashing. Also few haters, many appreciations and kindness and calm of Gomboo.

I have to say thank you a lot for a great education. I am looking forward for some new knowledge:)

Glad you have enjoyed the thread, thank you for the compliments.

I test both in NinjaTrader and Microsoft Excel.  I use Excel for the majority of my testing / trading research.

Best of luck

Draw EMA's? and the volume available at a certain order.

Hey GoomBoo,

If you have some spare time, can you please take a quick look at http://bitcoin.stoto.net/ and give me some suggestion, what to add, to make it more useful for traders like you?

Goomboo; you never crease to amaze me both quality of your analysis and your willingness to give it away for free. You make some awesome contributions that you not at all obligated to share.  Thank you.

The clear and concise way you present time as it is relevant to trading is the best I've seen this concept described.

In what language do you write analytic code to test current theory and past performance?. Is it the same as your predictive models?

In the rawest of forms, I have found Python to serve these purposes well 

Keep up the good work goomboo.

Alright thank you for all this advises I'll test it and validate. It confuses a bit but if you're just looking at the direction of the highest one it's not that confusing
Thank you anyway for all this good tips you give and all this explications!

Yeah, in that situation, you'd exit your position if you were relying on multiple timeframes.

The longer timeframe tells you which direction you should be trading.  The shorter timeframe tells you when you should be trading.

My personal opinion is that multiple timeframes tends to confuse things, so I prefer to stick with analyzing a single timeframe.  If you're able to use it to your advantage, then by all means, test and trade!

Buy half sell half in that case^^

You sell half? 

Ok for the multiple timeframe analysis and if something like this happen what should I do? (using this method)




higher time frame say "hold" and primary trend say "sell", I think it's sell.

edit:
I mean what information the longer time frame give? The trend of the market? That's what I suppose.

I'm happy to hear that you are profitably trading.

Your idea of looking at the trading system across different timeframes (1-hour, 4-hour) is certainly valid.  It's a concept known as "multiple timeframe analysis".  It's based on the theory that the trend of any given timeframe will more than likely be influenced by a higher timeframe.  In trading jargon, the trend of the higher timeframe is known as the "primary trend".  So, the idea would be to only hold BTC on the 1-hour timeframe if the 4-hour timeframe says you should hold as well.  Or in trader speak: "Only trade with the primary trend".

I do suggest testing it and you can use Excel, Ninjatrader, or even test it by hand.  You can express all of the moving averages in a single time period to make testing easier.  For example, the 10 period moving average on the 4-hour chart is the similar to the 40 period moving average on the 1-hour chart.

The general steps in NinjaTrader
-Import the data you're trying to test - it has to be in this format: http://www.ninjatrader.com/support/helpGuides/nt7/index.html?importing.htm
-Create the strategy using NinjaTrader's strategy builder (or code it yourself...strategy tester is pretty easy for most scripts...and NinjaTrader already has a moving average crossover pre-programmed)
-Backtest the strategy across the data you imported using the Strategy Analyzer

To my mind that's the key!!!! I'm a student (electronic PHD) and I have many teachers almost all are doctors (8 years or most studying a specific field). You can’t understand some of them even if they say they are detailing something very simple and easy but they take 1 hour to explain it. Others always start with “Look it’s very simple” and can detail something way more complex in 2 minutes.

In our case if your trading system is easy to understand then that don’t mean that he is to simple for trading but that this system make sense. That’s all. And with sense you’ll be able to keep trading even in a 7 loss row and keep your feelings apart.
If your system is too complex you’ll have much more chances to be confused and lost.

I’m just a trader for 15 days now so I can’t really speak with my experience but I’ve elaborated a bit the EMA system:
I’ve added a warning message (python bot) when I have BTC (that mean the EMA 10 is over the EMA 21) which calculate the differentiation of the EMA 10 (a*x+b function) if a<0 then the bot send me a message like “look at BTC price”.  This does not enter in the trading system I’ve just added it as a warning because I’ve noticed that under a certain a coefficient the market tend to be short quiet often.

With this method I can trade using a Hourly based EMA. And by the way I’m also considering to use 2 EMA base or more in the trading system. Like checking on a 4-hour base to have the trend of the market and selling when the 1-hour  base EMA are crossing. Is it an admissible idea?^^ I suppose you’ll tell me to back test this method. I tried to do so but failed when I needed to use Ninjatrader :/ is any one able to supply me some help or documentation on back testing using ninja or any other tool?

Eventually, on 10 days of active trading I made 3 full trades (USD->BTC->USD) and I had 6.7% profit, -4.3% loss, 7% Profit: ~9% of profit (including the Mt. Gox fees), I’m definitely a Pro trader^^ (just kidding). But I never though that trading would be following a system and keeping away “information influencing the price of BTC” and I support this now

Uh? There was an "imminent" crossover for almost 24h before the last little upward move.

It looks like a crossover is imminent. Its been too long!

Using the daily candles would have made you to sell at the very bottom ($50-60), I would have killed myself if I did such a mistake. Then, on the daily chart EMAs crossed again at $100-120, so you would have bought back at a huge loss.

To the contrary, using hourly candles you would have sold at $200ish (well.. If Gox queue let you, which is not obvious ), and bought back at $100ish after the bounce from $50.

I really do not think Goombo is using daily candles to check if the 10 and 21 EMAs crossed, can he confirm?