|
Z10 Questions
#472632
05/09/18 02:35
05/09/18 02:35
|
Joined: Jun 2013
Posts: 1,609
DdlV
OP
Serious User
|
OP
Serious User
Joined: Jun 2013
Posts: 1,609
|
Hi jcl. After playing a bit with Z10 have a few initial questions:
a) The manual says the prices come from CryptoCopy - should that be CryptoCompare? cryptocopy.me seems to be down (dead?)...
b) Z10 seems to download the entire price history for each asset all over again each new day, rather than just the change - the few latest prices. This is obviously not a big problem given this would only be done every 3 weeks, but is it as expected?
c) I am able to Test to the AR posted in the manual, but in the Zorro window the Monte Carlo AR is wrong - it is not the 50% value from the Performance Report - in fact it isn't any of those values. Is this a bug?
d) Changing the AssetsZ10.csv to (only) add USDT back in drops the CAGR by around an order of magnitude. Other asset combinations can make the system a loser, which seems to validate the manual's comments about the riskiness of Z10 and likely selection bias. How then to best proceed? Is the best choice to include as many of one's exchange's coins as possible (that have reasonable history) to minimize selection bias and therefore risk, at the expense of (Test) CAGR?
e) To clarify another manual comment, should one or the other of USDT and DGD be included, but never both? They would seem to serve different purposes, but the AR difference noted above when both are included gives pause...
f) Re. brokers, newbie question: Rather than a crypto exchange, would it be reasonable to use Shapeshift from a wallet?
Thanks.
|
|
|
Re: Z10 Questions
[Re: Nicola]
#472683
05/14/18 08:19
05/14/18 08:19
|
Joined: Jul 2000
Posts: 27,986 Frankfurt
jcl
Chief Engineer
|
Chief Engineer
Joined: Jul 2000
Posts: 27,986
Frankfurt
|
Yes, it*s Cryptocompare. From online sources, price history is normally loaded in full since many don't have a setting for limiting it anyway. Zorro cannot calculate the Montecarlo AR of reinvesting systems, therefore it is replaced by a script calculation in the performance report, but not in the window. This will be fixed, as also the number of decimals in the CSV.
When editing the asset list, do that only by fundamental criteria, not by test results, or else you'll get awful selection bias. You can add or remove what you want, but then you must stick to it no matter how bad the backtest gets. USDT is excluded because it is not traded at the exchange we're using.
I did not understand question H).
|
|
|
|