I am using a strategy I made using the Johansen Cointegration. And even with slippage adusted to
set_slippage(slippage.NoSlippage(max_volume=0))
I am getting varying results
Im using the raw price data and orders use order_target.
Im getting
Dataset
Crypto
Start Date
31 Dec 2017
End Date
31 Dec 2018
Capital
10000
RETURNS
-97.63%
ALPHA
-0.85
BETA
-0.06
SHARPE
-2.15
DRAWDOWN
-97.63%
And
Dataset
Crypto
Start Date
31 Dec 2017
End Date
31 Dec 2018
Capital
100000
RETURNS
44.21%
ALPHA
0.37
BETA
0.04
SHARPE
0.97
DRAWDOWN
-6.55%
How can this be?
I can send the code please let me know what email address to send it to.
Also I get this without slippage adusted
Dataset
Crypto
Start Date
31 Dec 2017
End Date
31 Dec 2018
Capital
100000
RETURNS
-17.98%
ALPHA
-0.10
BETA
0.00
SHARPE
-0.45
DRAWDOWN
-36.9
Dataset
Crypto
Start Date
31 Dec 2017
End Date
31 Dec 2018
Capital
10000
RETURNS
376.86%
ALPHA
10.45
BETA
-0.38
SHARPE
1.39
DRAWDOWN
-88.82%
Hello Jane,
You have presented 4 strategy performances.
Could you please send to me the characteristics of each model to comprehend better what it's happening?
Thanks
José Carlos
I emailed you the code at quantra@quantinsti.com. The first 2 are with silppage reomoved and the last 2 its with. The only thing that was changed was capital.
Hey Jane,
The difference in the performance is due to capital only. Let me explain with an example.
On 2018-03-29, the algo is trying to place the below order. The capital required to place the below order is more than 10,000. So while running the strategy with 10,000 as the capital some of the orders will not be placed as there is no sufficient capital to place the order. But that is not the case when the capital is set to 100,000. Order placed for the below assets will be completed with 100,000 as capital. This is causing the difference in the returns and other stats.
Asset Name Quantity * Price BTC/USDT-1.0 * 7927 BNB/USDT -218.7811 * 1.6333 ETH/USDT 5.4532 * 446.48
Note:
1. while sharing the code you can copy-paste it in notepad and save as a file with .py extension. You can upload it on google drive or other sharing platforms and share the link along with the community post. Don't forget to make the link public while sharing on community. This can also help others benefit from this community post.
2. If you want to compare two backtest, run the full backtest (using the New Backtest Orange button) and compare the transactions and round trips. These can explain the difference in most of the cases.
Ishan Shah your comments as usual are a little shallow. Kindly check the codes in question. Also im waiting for Mr Jose Carlos Gonzales Tanaka to respond. And explain why the codes are performing so very differently.
I think adding mroe capital would make the same amount using order_target. But Im adding more money and losing without adjusting the slippage.
Hello Jane,
The explanation given by Mr Shah is actually the explanation of what it's happening with the first two strategy performances.
Can you please provide the different parameters you applied between the first two strategies and the last two?
In order to compare appropriately strategy performances, we need more information you have in each strategy. We need to know the differences you have on each strategy compared to the rest of the other strategies. This will let us know what exactly might have happened to the performances.
You might think Mr Shah is giving you a short answer, but actually, his answer corresponds to the sole information you are given. It's not only about providing you code, but you can also share kindly the differences between the four strategies, so we can help you more precisely to acknowledge what is happening with your strategies.
We'll be glad to receive the information of the differences between the four strategies to grasp better what it's happening with their performances. Could you please upload your strategy code on a .py extension in a google drive and then send the link to us so we can check it quickly?
Thanks,
José Carlos