Goomboo's Journal

[TradesLikeAPotato]

I disagree. .4% of your trade really adds up, and if BTCe has less volatility it is only marginally so.

[1714195530]

1714195530

[1714195530]

1714195530

[sukiho]

I disagree. .4% of your trade really adds up, and if BTCe has less volatility it is only marginally so.

I agree it adds up but it still seems to me that mtgox has higher highs http://bitcoin.stoto.net/

[Goomboo]

Goomboo, I have a question for you. Hourly trading is only inferior to daily because of fees, so trading on a BTCe with a .2 fee may yield better results. The problem is, you cannot backtest this. Would using MtGox candles with a .2 fee be a good indicator of potential profits on BTCe? Or is the arbitrage between the two enough to nullify the data?

The question really can't be answered without directly testing it.  There are several variables at stake here:

- How much is the typical slippage on any given trade?  Is it different between exchanges?
- Is there a delay in execution between the different exchanges?

In general, the only thing I can say is that arbitrage would virtually ensure that significant price movements would be mimiced across exchanges if money could be easily and efficiently moved between the exchanges. However, in the current environment, this is uncertain as well :p

So to directly answer your question: I don't know, there are too many moving pieces.  But in general, yeah the movement captured should be relatively the same in the long run.

[TradesLikeAPotato]

Thanks for your replies Goomboo and sukiho.

https://i.imgur.com/yBjoAjN.png

 Here is a poorly photoshopped graph comparing the highs and lows between the exchanges. It's true that MtGox generally has slightly higher highs and lower lows, but (ignoring the recent arbitrage due to MtGox issues) I think the fees would outweigh the profits from MtGox. Whether trading hourly on BTCe or daily on MtGox would be more profitable I am still unsure.

[RyNinDaCleM]

Goomboo, I have a question for you. Hourly trading is only inferior to daily because of fees, so trading on a BTCe with a .2 fee may yield better results. The problem is, you cannot backtest this. Would using MtGox candles with a .2 fee be a good indicator of potential profits on BTCe? Or is the arbitrage between the two enough to nullify the data?

Why can't you back test BTC-e?
Bitcoincharts.com has all the trade history.

[TradesLikeAPotato]

It's possible. The problem is getting the candles. The ones for MtGox are easily downloadable, while to get the BTCe ones I'd need to figure out the bitcoincharts API. I'm not a programmer but I'll take a look tomorrow and see how much work it would take.

EDIT: Yeah I have no idea what I'm doing.

[kenshin23]

Bitcoincharts doesn't have any documentation for their charts API, so what I ended up doing to backtest was the following:

Get the chart with the timeframe that I want to backtest (for example: 5 months, hourly.)
Click the "load raw data" link below the chart.
Select the first line of the data, hold the mouse button, then press the "End" key on the keyboard.
Drag the mouse to the end of the last line, then copy the data (Ctrl+C or right-click, then Copy)
Then, paste into a new text file and modify the data to suit your backtesting program (I use Gekko, so I wrote a simple PHP script that gets each candle, formats the date to a UNIX timestamp and then returns a CSV file that Gekko can read.)

I hope that helps.

[TradesLikeAPotato]

You beat me to it, I was about to post a similar solution.
You can convert to unix timestamps in excel actually, just use this formula:
=(A2-DATE(1970,1,1))*86400

 I extracted MtGox candles for the period in this thread: https://bitcointalk.org/index.php?topic=239815.0  Strangely, I end up with completely different backtest results.

Could you try and test it yourself? I'd give you my candles but those are precisely what I want to test. Start at June 26, 2011 and paste each period of 5 months to get a full candles table.

[TradesLikeAPotato]

I just created a topic detailing this technique:

https://bitcointalk.org/index.php?topic=277704

[oda.krell]

I just created a topic detailing this technique:

https://bitcointalk.org/index.php?topic=277704

Extremely useful, thanks! This is what finally got me started using gekko myself, doing some proper backtesting of EMA methods.

My search starts from the assumption that (a) more recent history is more relevant for present day prediction than older history, (b) overfitting of the method must be avoided, so results need to be compared across different time frames, (c)
trading is slightly more expensive than assumed by default in gekko, because of signal delay & slippage

I'll post my results when I'm done with the testing. I have to do it partly manually because I want to control the treshold parameter, so I'm not searching through all EMA combinations, but only those that previous posters have mentioned as promising. Maybe someone else can fill me in in case I forgot a promising combination.


A quick question for Goomboo

Looks like we're turning this thread into a discussion about backtesting EMA strategies, finding the optimal parameters, etc. If this is starting to become too tangential to your original idea of this thread, i.e. you think we're hijacking your thread, let us know.

[Goomboo]

Looks like we're turning this thread into a discussion about backtesting EMA strategies, finding the optimal parameters, etc. If this is starting to become too tangential to your original idea of this thread, i.e. you think we're hijacking your thread, let us know.

No worries, thank you for asking.  Been much to busy to post lately, and the discussion is certainly on-topic, so I'm fine with it.

Goomboo

[TradesLikeAPotato]

I just created a topic detailing this technique:

https://bitcointalk.org/index.php?topic=277704

Extremely useful, thanks! This is what finally got me started using gekko myself, doing some proper backtesting of EMA methods.

My search starts from the assumption that (a) more recent history is more relevant for present day prediction than older history, (b) overfitting of the method must be avoided, so results need to be compared across different time frames, (c)
trading is slightly more expensive than assumed by default in gekko, because of signal delay & slippage

I'll post my results when I'm done with the testing. I have to do it partly manually because I want to control the treshold parameter, so I'm not searching through all EMA combinations, but only those that previous posters have mentioned as promising. Maybe someone else can fill me in in case I forgot a promising combination.


A quick question for Goomboo

Looks like we're turning this thread into a discussion about backtesting EMA strategies, finding the optimal parameters, etc. If this is starting to become too tangential to your original idea of this thread, i.e. you think we're hijacking your thread, let us know.

Glad I could help, I'm looking forward to your results. While you do your testing, maybe you could speculate the cause of the absurd BTCe profit values I have encountered here.

[oda.krell]

[...]
Glad I could help, I'm looking forward to your results. While you do your testing, maybe you could speculate the cause of the absurd BTCe profit values I have encountered here.

I can reproduce the result. The reason: you switched 'long' and 'short' EMAs (you defined short to be 5, and long to be 1). If you switch them around, you will get (as expected) a -100% profit, with the same number of trades, which makes me think that the EMA method written by the gekko author breaks under wrong user input.

Sorry, if I had more time I could look at the code and confirm, but for now I'd say chances are pretty good that's the reason.

[TradesLikeAPotato]

I think you are correct.

[kvm]

Please have a look at my thread on how to get bitcoin realtime charts in MetaTrader 4:

https://bitcointalk.org/index.php?topic=160791.0

I will be glad to hear your suggestions and recommendations! I'm thinking to add a range bars generation! Will it be useful for backtesting?

[Marcus Antonius]

Question for Goomboo: do you do anything different when you see a parabolic spike approaching?

[0xz[.]]

Goomboo, do you offer  private advices? do you have subscribers?

[Goomboo]

Question for Goomboo: do you do anything different when you see a parabolic spike approaching?

In the context of this thread, unless you can quantitatively describe a parabolic rise and have backtested a method of taking profits during periods of mania, then I would say to wait until the crossover has occurred prior to exiting.  "The trend is your friend until the bend at the end".

[Goomboo]

Goomboo, do you offer  private advices? do you have subscribers?

Sorry, I do not.

Best of luck!

[Xiaoma]

I'm personally very wary of EMA during extreme situations like current one, where you don't have a trend but a few whales stomping all over a weak market.

My observation here is that whale buys big as soon as lines show time to sell. You end up losing a lot if you follow the averages now. I'm back to instinct mode for a couple weeks, until I feel market is healthy. Or Gox goes bankrupt lol, I moved to bitstamp at a loss but I feel safer.

I wonder is Gomboo still trading nowadays? I learned a lot from you