the other algorithm is this one
https://www.quantopian.com/live_algorithms/59a423b4084e5f0010daf7f6

Can someone tell me why it is not appearing in leaderboard?

thanks
-kamal

Your backtest beta is 0.5414. I believe backtest beta must be less than 0.3 in order to earn a market-neutral badge and qualify for top rank. Overall your backtest stats are not so great, so it's likely that as the paper-trade progresses it's performance will deteriorate significantly. You should aim for a sharpe at least over 1.0 on the backtest.

Sometimes it takes a couple days for algorithms to appear on the leaderboard.

Hi Viridian,
My also is ranked #1 in paper trading score and #343 in backtest. When I backtested from jan 1 2002 to aug 8, 2017 beta was much lower and so I submitted it to contest. I don't know how to get the same result as the backtest here.
The beta in paper trading is 0.03074,with a sharpe ratio of 21.07 . So maybe it isn't all that bad an algorithm.
What is the likely outcome if it remains #1 for some more time or for all 6 months? How much will they pay me and when?

thanks
-kamal

I took a look at the linked entry. The backtest has a beta of .5414, which is too high to earn the beta badge. You need all three badges in order to rank highly.

The deadline for that contest is still several days away. You can stop your entry and submit a new, better algorithm that has better control over your beta.

If you have a contest entry that isn't on the leaderboard, the most common reason is that it's a very new entry. Your entry needs to have 2 trading days before it appears on the leaderboard. If it's still not there, please contact support.

@Kamal, I believe the backtest period for the contest is 2-years. Hope that helps.

Please keep in mind that point estimates of beta are often misleading. Beta to market is effectively an average of all betas over your backtest period. If your algorithm alternates between -1 and 1, you will still have an average of 0. It is much better to look at the rolling beta over time to determine what your risk exposure characteristics look like. The primary thing is they need to be consistently low, jumps or inconsistency indicate poorly managed risk. For more see here:

https://www.quantopian.com/lectures/portfolio-analysis
https://www.quantopian.com/lectures/instability-of-estimates
https://www.quantopian.com/lectures/factor-risk-exposure

I fixed my algorithm and it is here
https://www.quantopian.com/leaderboard/33/59a83654b6bc617e3ceebc07

funny thing is that a good back test yields a bad live trading performance and vice versa. I believe something needs to be fixed in the back test so that it is more in sync with live trading.

thanks
-kamal

My beta is between 0.2 and 0.3 across time frames, but needs a few days to come down. Same for sharpe ratio. I don't have access to the algorithm by Persian Blue Leopard. So, what will I get by looking at his stats?
Master what step for live performance? I could have gotten higher performance but with higher beta i.e the more I hedge, the lower is the performance.

thanks
-kamal

He is around 78% whereas highest is ~90%
What is a factor sort and how can I use it?
I shorted based on some morningstar data which will raise value of portfolio when whole market tanks.
If I short more, the beta will come down but so will the toppling performance and sharpe ratio.

thanks
-kamal

My safari browser does not load notebooks for some reason. So, is it possible for someone to cut-paste the entire program given in "factor analysis"?

thanks
-kamal

Kamal,

Research requires you to allow third-party cookies in your browser. Can you try enabling cookies and see if that allows you to load the notebook?

I had originally allowed cookies from current website. When I changed to allow cookies always. it loaded the notebook. But I still cannot see the whole progrsm, only pieces of it with explanation. I wanted to cut-paste the whole program discussed in factor analysis lecture.

thanks
-kamal

Hello Kamal,

The lecture on factor analysis is presented in Jupyter notebook format, a cell based code execution environment. For more info see here:

https://www.quantopian.com/lectures/introduction-to-research

For quantitative research, this format is often better as it allows you to isolate pieces of analysis and have visibility into your data. Notice that this line at the end of the factor analysis lecture will run a full analysis all at once:

al.tears.create_full_tear_sheet(factor_data, by_group=True)

You still have to perform the setup done earlier in the lecture to get the required format for the data in factor_data.

Thanks. When I execute [2], it yields

NameErrorTraceback (most recent call last)
in ()
----> 1 class MyFactor(CustomFactor):
2 """ Momentum factor """
3 inputs = [USEquityPricing.close,
4 Returns(window_length=126)]
5 window_length = 252

NameError: name 'CustomFactor' is not defined

I am actually on the lookout for a module which yields a sharpe ratio of > 2 and beta of < 0.3 as indicated by the poster on this thread.

thanks
-kamal

I cut pasted the snippets into a new algorithm and got this build error:-

24 Error Importing alphalens raised an ImportError. No modules or attributes with a similar name were found.Why?

60 Warning Undefined name 'get_pricing'

The Why for the build error indicates that i cannot import alphalens.
OliveCoyote has written above:-
"Honestly, just follow the lectures on factor analysis and long/short equity, use alternative data sets, and you'll find it's quite straightforward to achieve backtests with Sharpe Ratios of 2 or more."

So, can someone tell me how to get an algorithm with a sharpe ratio of 2 or more with a suitable beta?

thanks
-kamal