Topic: Goomboo's Journal - page 38. (Read 281461 times)

Hey all,

I decided to backtest the hourly strategy on hourly data for the past 5 months (as much data as I could easily get my hands on).  This is the 10 / 21 exponential crossover from September until now.

Below is a summary of how the signals have performed for the past 123 trades.  As always, remember that the dollar values are meaningless (this is simply trading a flat 100 BTC every trade).  Your actual performance should vary with your risk-management methodology.

My key takeaways:
-34% success rate
-.83 trades per day on average
-29 hours in an average trade
-Both long and short profitable across the time period studied

Goomboo, there was a book recommendation you made in this thread a while ago.  What was it?

Thank you very much for the compliments and I'm more than happy to answer your questions.  We all were new to it at some point, so don't worry :p

1.  The moving average number just says how many data points it is using to calculate the average - in our case closing price of candles (hours).  The number 10 means that it is an average of the 10 past closes.  It's an exponential average which means (skipping the math) that it gives a more "smooth" picture.  For a comparison, I'd look at a simple moving average versus an exponential moving average of the same period, side-by-side and you'll get the "aha!" moment.  The idea behind using a moving average is that it gives us a picture of the trend, or which way prices are tending to travel at that moment.

2.  Liquidity is a fancy finance word that basically is about quantity, not time.  It's an electronic exchange, so when I click "buy", it should be executed pretty much immediately, so time is not really what I'm worried about.  If you look at that website (http://bitcoin.clarkmoody.com/) it shows you the amount of bitcoin you can buy or sell at certain prices.  I'm worried about a concept called slippage, which I'll explain in the below example/picture.

Let's say I want to buy 1,000 BTC right now.  For me to buy, there has to be someone willing to sell me these BTC - and I need to know what his price is.  Since I'm trying to trade for a profit, the price I receive (the average price I pay per BTC), is important because after all my profit is just the difference in my entry / exit prices minus commissions.  Below is a screenshot of the liquidity in the market and the current price at Mt. Gox.  That price at Mt. Gox is fine if you are just curiously looking at BTC, but if you are actually trading them, you need to know what your trading price will be.  For example, to buy 1,000 BTC, you would end up paying an average of around $5.00 per BTC.  This is called slippage - you see the price is $4.98, but your trade will actually cost of $5.00 per coin - you slipped!

Putting it in fancy finance words - liquidity helps prevent slippage.




I wish you the very best in your trading and I am very happy to see that you are doing the research and protecting your hard-earned money.

Haha, I've never seen anything describing this forum more accurately, spot on, dude, spot on.

No! You should have been Long (Holding BTC) since $4.3x, right around the open of the 22nd, and still holding.

It is a zero-sum game out there, which means if you lose your money someone is taking it.  Go read this whole thing before starting.

Also.  Did the EMA10/21 crossover fail to catch the move up to $5?

Ain't that the truth

Frequenting these fora, shall make for some entertaining times ahead!  

Edit:

I'll help by saying;
Yes! The EMA's referred to would be the 10/21. The time period is purely up to you. A decision you make based on the amount of time/risk you wish to invest. An hourly EMA would require much more time spent watching for the next cross-over. Usually once a day, but can be more or less. Whereas a daily EMA would have you trading once or twice a week.

Hi Goomboo, thanks so much for posting this epic thread. I am relatively new to all this, so please excuse my ignorance if my questions are a bit basic. I would really like to ask them though, as I am determined to never use a method which I do not fully understand.

Firstly, the exponential moving averages; the values are 10 and 25. But 10 and 25 what? Is it, as I suspect, 'ticks' which to my understanding is the time period which you select - in this case hours?

Secondly, you mentioned in your original post about checking for liquidity. I understand that this means that you want to make sure there is enough capability of movement in the market such that you don't end up posting a buy or sell order which takes a large amount of time to be accepted, exposing you to the risk of market movement against you in the meantime. Could you please explain how you read the charts related to the website that you linked to in order to determine if there is sufficient liquidity to cover your moves?

I am in BTC more for the fun aspect and because I believe in the technology than to necessarily make a large profit; I have only invested a small amount which I can afford to lose. But I sincerely hope I can make a nice profit so I can give you a generous tip not only for your great advice, but also for your general ethos; one thing I have noted about this site is that people are almost jumping of the roof of the exchange / cracking champagne bottles just about every few minutes depending on the market. It's really refreshing to hear from someone with a level head and real world experience.

Cheers.

It resets when I'm really happy with the previous trade.  It's subjective, and I look at a lot of other information.  The whole point is to spend less when it is likely to go sideways.  There is also a cap of how much I'm willing to put down on a given trade, but I try to keep that around 4-5X so that odds are there will be a trend before I hit it.

How do you determine what's large enough to be considered a "profitable move"?  Is that the only condition that will reset the multiplier?

Basically yes, just the amount of capital allotted for a given trade. Perhaps as a %. Less investment = less risk.
You wouldn't just set it free with your entire account balance would you? Especially until well tested. 




I like this! Automate what I already do. 

+1 to this. Nice idea. Let us know how it goes

Intresting. kind of like playing blackjack and always upping the stakes.
Let us now how it goes 

The method I'm testing out is scaling in after a big move.  The next cross over I trade 1X, then 2X, then 3X, then 4X, etc. until there's another large, profitable move.  Then I go back to 1X.

Can you elaborate a little on how risk is managed with volume?  Do you mean because of the higher frequency of trades involved in an hourly trading system?  Or perhaps amount of capital involved?

You would manage risk with volume. Decision making would be a simple Boolean statement.

EMA10 > EMA21
True -> Buy
False-> Sell

yes and compare with shorther and longer term macd and other indicators to calculate a percentual risk, buy/sell if risk low enought

other ideas?

Bot would detect it, then wait for the next candle to confirm. As you (a human) would.

How do you detect false crossovers?  If you can do it then a bot can as well.

In a situation like this it is possible you have like 2/3 crossovers. An automated system would trade on every crossover and this is would lose your money.
But ofcource if you make the system smart enough there is no problem