Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
What's the meaning of the backtest result ?

RETURNS
4.7%

ALPHA
0.03

BETA
-0.01

SHARPE
0.59

DRAWDOWN
2.1%

Algorithm
1.2%

Benchmark
5.6%

Can anyone explain these keywords for me? I'm new to here and don't know much about them.

4 responses

Hello Roy,

Returns: if you started with $10,000 you finished with $10,470

Alpha: +ve Alpha means you had a return in excess of what would be assumed for the risk you took

Beta: -ve Beta means your algo is is producing returns opposite to the benchmark i.e. -ve returns when the benchmark has +ve returns and vice versa

(Alpha and Beta relate to the Capital Assets Pricing Model (CAPM) and are the intercept (Alpha) and slope (Beta) of a linear regression (see http://www.moneychimp.com/articles/risk/regression.htm )

Sharpe: a risk-adjusted performance measure i.e. a low Sharpe may suggest a level of risk with which you might not be comfortable

Drawdown: the maximum difference between a high and low portfolio value (peak to trough) during a given time period

Algorithm: the algo returns. This can be a function of your starting capital if you are trading fixed quantities of shares, A $1,000,000 algo that loses 8.6% could become a $100,000 algo that loses 85.9%.

Benchmark: the benchmark returns but at present without dividends reinvested so this figure is lower than it really is and as a result the algo returns look better by comparison.

P.

Hi, Peter, thanks for your explain, it makes more sense to me now.

But I still can't understand what Benchmark is, can I just ignore this feature ? I think Algorithm returns and drawdown are the most important features, right ?

Hello Roy,

The benchmark returns are like a Buy & Hold strategy i.e. you bought the SPDR® S&P 500® ETF at the start of your backtest period and did nothing else i.e. it does not need to be actively managed/traded. A benchmark (you could define your own and it could be a 'portfolio of benchmarks' - see https://www.quantopian.com/posts/question-regarding-the-benchmark) is needed or how can you judge your algo?

Drawdown feeds in to how risk averse or risk tolerant you are. I suppose. If you can watch 90% of your equity disappear in the belief that you will finish up +10%.....

P.