Re: Haasonline Simple Trade Bot For BTCe and Bitstamp[Main topic]
I have problem with Cryptsy API i think. No data for benchmark and often failed get data from crryptsy (see log)
Where is the problem ? Cryptsy ?
2014-21-02 11:05:55.052 |TRACKER_INFO| Reading of Cryptsy failed, retrying....
2014-21-02 11:05:56.097 |TRACKER_INFO| Reading of Cryptsy failed, next retry on next update.
2014-21-02 11:05:56.097 |ERROR| Cannot access child value on Newtonsoft.Json.Linq.JValue.
2014-21-02 11:05:56.097 |TRACKER_INFO| Cannot access child value on Newtonsoft.Json.Linq.JValue.
2014-21-02 11:05:57.252 |ERROR| MAX is not a supported currency.
Where is the problem ? Cryptsy ?
2014-21-02 11:05:55.052 |TRACKER_INFO| Reading of Cryptsy failed, retrying....
2014-21-02 11:05:56.097 |TRACKER_INFO| Reading of Cryptsy failed, next retry on next update.
2014-21-02 11:05:56.097 |ERROR| Cannot access child value on Newtonsoft.Json.Linq.JValue.
2014-21-02 11:05:56.097 |TRACKER_INFO| Cannot access child value on Newtonsoft.Json.Linq.JValue.
2014-21-02 11:05:57.252 |ERROR| MAX is not a supported currency.
The biggest coins are cached from the webserver, not all the alt-coins. It would be a impossible job to do.
What you see here is a 2 times try to get a response from Cryptsy and both tries have failed. So its very safe to conclude this is related to Cryptsy. (i see them too sometimes)
Its not about different prices, they are fine. If I did not miss something, candles consists of summaries of prices of fulfilled trades per time period (in this case 15 minutes on both haasbot and Bitcoinwisdom). Even if the candle collection time was not synced, difference should be max half of they time, so 7.5 minutes in this case, so no big deal. But prices are quite the same, what is not is MACD histogram. So basically I mean input is same, but there is huge difference in output. Or does hassbot collect data different way than usually?
Right on. We collect our prices differently. We do not make use of the fixed prices. This gives us the advantage of being unique (even per users of this software) and just be before or just behind all the other trade-bots.
BTW Most of the indicators will never look to a candle, indicators will only look at the current buy price. That is where the math is based upon.
Of course I don´t want to trade at the same times as others do. My point was to try to feed same data to both Bitcoinwisdom and Haasbot and see if the results match each other, because if not, there is something wrong.. Then I can´t fully rely on hassbots logic and find out good strategy. Its like asking two people how many is 2+3 and hearing different results. Or more like asking whats the time - if someone reply with +/-half a hour accuracy, I can get past that, but if its more than 3 hours difference, its useless info.
There is nothing wrong, its almost impossible to say there is anything wrong because pure mathematics are used. It will just respond the same, if the base is the same to. But if it is not then the effect of the differences will grow in the math. Especially on a EMA, in that math we use a exponential for the moving average.
I will be able to prove this to you (and anybody else who want this) later on. At the moment the backtest/benchmark uses its own prices, but in the future it will be possible to import information from the API of bitcoincharts. In here the so called fixed prices will becomes available all the others use too and then i can show it.