I recently asked a similar, though not as complete question. All I have heard back thus far is crickets....

Perhaps you more in depth question will elicit a response.

Leverage < 3
Alpha - don't know
Beta close to zero
Sharpe > 2
Drawdown < 5%
Returns > 12 % per annum

Thanks Paul and Pravin

Hopefully we can have some others join in on this. I think it would be helpful if these kind of questions were added to a FAQ or something on the site for new traders. It would just be nice to know what to go after and what would be considered a successful algo.

Leverage < 2.9 (Contest rules), <3.5 (personal use)

Beta: +- 0.2 (contest), doesn't matter for personal use.

Annual returns to be > 5x max drawdown

Leverage to be adjusted such that max drawdown < 7.5% (Contest algo), 20% (Personal algo)

Few mil algo capacity (contest), sufficient capacity (for personal use)

Hedged component (contest)

I also look at the consistency of returns. (returns graph should look similar to this: http://imgur.com/dbAPmQb)
It should also make money regardless of market condition/financial crisis, or be smart enough to exit all positions when market condition is unfavorable.

Thanks Kayden for the great information. This is very helpful.

Thanks Kayden for the great information. This is very helpful.

Kayden, thanks. Regarding drawdown vs returns, is that over any single year, or cumulative?

For example, I have an algo that over the period from 04/2015 to 04/2016 shows a max drawdown of 11.5% and a return of 28.39%, but over the period 4/2011 to 4/2016 the numbers are 25.6% vs 553.8%.

Unfortunately the beta is -.29, but it has 0 leverage based on $10,000

The numbers are different at $1,000,000 due to slippage I guess. 27.1%max drawdown vs 336.7%profit with a beta of -.16.

Specifically what does hedging include? The algo above has two sets of stocks that based on simple market conditions it chooses to buy/sell.

Paul, I use annualized returns over max drawdown from a full length backtest.

In your example, your algo had roughly 40% annualized return and 25.6% max drawdown from 4/2011 - 4/2016. I think it's important to see how your algo fares during the financial crisis and certain stress. A good way to do so is running your backtest through Robert Shanks' tearsheet in this link https://www.quantopian.com/posts/contest-8-winner-robert-shanks.

About hedging, Quantopian requires an algo to hold both long and short positions at the same time to be insured against strong market movements in either direction. Essentially you bet your long positions will outperform your short positions under any market condition. I tend to write intraday strategies and have not found a hedging strategy that makes it worth the slippage and commissions, perhaps someone with more experience can elaborate on this.

All of this is wishful thinking. In my opinion you would be better advised to explore what risk / return profiles have been achieved in the real world by real find managers, hedge or traditional.

Judging by the figures in this thread I think you will find such exploration both enlightening and alarming.

Most of you guys are simply howling at the moon.

A few years back for instance I too long term CTA track records spreading back over many, many years.

On average I seem to recall an MAR of around 0.2 but frankly I can't remember the exact figure.

Very sobering.