@theflyingdutchman thanks, understood

@captain-nidoran

Hello.

We are making some changes before starting competitions, it affected the statuses of some submissions and they became invisible on the client side. Now it should be ok.

We carefully preserve users' data, we are making backups every day. So in the worst case of data loss, we are able to restore most data from these backups. Don't worry a lot. And thank you for the report.

Regards.

@magenta-grimer Hi, the lookback period in the backtester function is expressed in calendar days, and the indicators are expressed in trading days. So as far as the lookback period is long enough to include all indicator needed periods, all choices are fine.

If the longest period needed for an indicator is 10 trading days (2 weeks), a lookback period of (for example) 20, 50 or 100 for the backtester function will deliver the same results. The shorter the better for efficiency.

@support Thank you once again. You guys are the best!

@penrose-moore We finally released a template which allows you to perform a retraining, it is available in the "Examples" section of your private space or publicly in the Documentation:

https://quantiacs.com/documentation/en/examples/machine_learning_with_a_voting_classifier.html

@support Thanks a lot!

@illustrious-felice Hi, that is not automatic as all trades are punished by slippage, and that is a subtraction from profits irrespective on the sign of the weights.

@magenta-kabuto Hi, did you try it? The slice function is designed to return all weights.

@antinomy
Thank you!

@support Thanks, mailing you!

@sheikh Thank you for the comments.

The BTC spot we are using are daily data, and open/close/high/low are computed by matching the opening and closing time of the BTC Futures contracts. As we are working with daily resolution, we believe the spot BTC is ok for the current contest. Having the weekend data at disposal, signals can be built by using more information. Moreover, the previous time series were built matching the past spot data with the short futures data.

This change is not disruptive, as old algorithms developed on the BTC Futures will still run on the BTC Futures. You pointed in another thread the change in the leaderboard: many thanks, please allow us to check and fix it.

Most of the quants requested this change, so we went for it. Maybe in the future we will allow for both options, but please note that the BTC Futures have a short history, so they will have anyhow to be patched with spot data. Thanks for the suggestion.

@algotime Dear Algotime, all three strategies will participate in the contest. We will update the contest leaderboard once all eligible strategies have finished processing. Thank you for your patience.

@support Thank you very much!

I hope to don't disturb you so often in the future @support , but before that I need help to "get going".

@cyan-gloom Hi, we are busy with checking the submissions which are coming, as we are close to the end of the Q20 contest, sorry if we were not responsive so fast. We will inform as soon as we are ready.

@support
Yes, pip is way faster. Thanks!
I might have found an even faster solution but I guess I have to wait a few hours to find out if it really works.

Here's what I did:

I created a folder in /root/books called "modules" to install cvxpy there to make it persistent: !mkdir modules && pip install --target=modules cvxpy Then if the import fails in the strategy, it creates symbolic links in /usr/local/lib/python3.7/site-packages/ that point to the content of /root/books/modules/ try: import cvxpy as cp except ImportError: import os source = '/root/book/modules/' target = '/usr/local/lib/python3.7/site-packages/' for dirpath, dirnames, filenames in os.walk(source): source_path = dirpath.replace(source, '') target_path = os.path.join(target, source_path) if not os.path.exists(target_path) and not os.path.islink(target_path): os.symlink(dirpath, target_path) continue for file in filenames: source_file = os.path.join(dirpath, file) target_file = os.path.join(target, source_path, file) if not os.path.exists(target_file) and not os.path.islink(target_file): os.symlink(source_file, target_file) import cvxpy as cp

Creating the symlinks only takes 0.07 seconds, so fingers crossed 🙂

UPDATE (a few hours later):
It actually worked. When I just reopened the strategy, the environment was newly initialized. First I tried just importing cvxpy and got the ModuleNotFoundError. Then I ran the strategy including the code above: cvxpy was imported correctly and the strategy ran.

I'm not sure if that solution works for every module because I don't know if pip might also write something to other directories than site-packages.

Anyway, I'm happy with this solution.
Regards

Dear @antinomy,

There has been some lag in processing, thanks for pointing it out.

Regards

@lookman The answer we sent to your e-mail address bounced back.

@magenta-grimer That is correct, but you can for example train a model locally on your machine, and then submit a code which uses the results of the training (for example stored in numerical form) but does not reveal the underlying idea.

Thanks for bringing this up. I had a similar question when looking through the example. From what I understand, it can seem forward-looking at first because of how the data is prepared and the model is trained. The important part is whether the prediction at each point is based only on information that would have been available at that time. If that's the case, then it may not be true look-ahead bias, just a backtesting implementation detail. That said, it's definitely worth double-checking because data leakage in machine learning strategies can be surprisingly easy to miss and can make results look much better than they would be in live trading.

@sun-73 it should be ok