Back to Community
Investors acceptable drawdown.

Anyone think 45.5% can be an acceptable drawdown for an investor if the sharpe is not bad and returns are really high?

From 2011-01-04 to 2017-06-30

Total Returns
Benchmark Returns
Max Drawdown

6 responses

You could not suffer such MDD when you really put majority of your money in. In my opinion, MDD should be less than 15% even the strategy is very profitable.

There aren't a lot of investors who would be interested in an algo like that. You'd have to have a cast-iron stomach on that roller coaster. I'm also curious to see how well it does out-of-sample; most algos I see with parameters like that tend to fall apart over time.

One practical problem: How would you know when to turn it off? Almost all algorithms have a lifespan, and when it stops working, you have to stop allocating to it. In this case, how would you know that it was time? When it loses 50%? 60%?


The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by Quantopian. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. No information contained herein should be regarded as a suggestion to engage in or refrain from any investment-related course of action as none of Quantopian nor any of its affiliates is undertaking to provide investment advice, act as an adviser to any plan or entity subject to the Employee Retirement Income Security Act of 1974, as amended, individual retirement account or individual retirement annuity, or give advice in a fiduciary capacity with respect to the materials presented herein. If you are an individual retirement or other investor, contact your financial advisor or other fiduciary unrelated to Quantopian about whether any given investment idea, strategy, product or service described herein may be appropriate for your circumstances. All investments involve risk, including loss of principal. Quantopian makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances.

That 45% is with the historical data (not even including 2008). There's nothing to say that in the future it couldn't be worse. Imagine putting all your money in right before that kind of drawdown. How long would it take to even break even again? How consistent is the algo? Are you getting those 100+% returns every year? Or does it depend on a few idiosyncratic events to boost the returns (which would suggest overfitting). I think there are a lot of things to consider, and historical returns is just one. It can be misleading.

I already have quite a few algos that would pay for a time machine if could just work out the paradox and get back to 2006 to invest in them.

I'd work on evening out the results and make sure the algo continues to perform out-of-sample. I'd consider combining this strategy with another that doesn't share its same weakness.

You could not suffer such MDD when you really put majority of your money in. In my opinion, MDD should be less than 15% even the strategy is very profitable.

Then you had better stick to holding 10 year high quality sovereign bonds!

Just publish the algo an we all can look at it and see where the overfitting is if there is any

Real life drawdown is almost always worse than backtested (due to inevitable overfitting or even just market efficiency) so I would not go over -15% max drawdown on an actively traded strategy in a backtest. In real life you'd need to have a pre-determined point at which you make the call "this algorithm is not performing as intended" and pull the plug. If your backtest drawdown is -45% do you pull the plug at -50%? -60%? The line gets very blurry at such insane drawdowns and almost impossible to recoup your losses. How much are you willing to lose on a potentially non-viable algorithm?