This does raise an interesting question, however: how many existing, successful investment strategies would survive a similar back-test of performance to the 1970s? Not many, I would be willing to bet. For many non-equity strategies the historical data is not available to conduct such a test, which is probably their good fortune.

I can only speak to equity strategies:

Cliff Asness of AQR gave an excellent talk that touched on these issues at the 2014 CFA Institute Annual Conference: “21st Century Asset Management: Facing The Great Divide.” He illustrated that the value anomaly persisted from 1926 through 2011. Of course, there were rough patches lasting several years each: AQR almost didn’t make it in the 1999 after the HML (their value factor) had an ~50% draw-down. HML was back in the green by 2001. This incident also illustrated that knowing economic reasons behind the losses of a factor helps manager stay the course and eventually make money. Managers treating this factor as a black box would have probably cut their losses at the worst possible time. This chart tells the story: http://greenbackd.files.wordpress.com/2014/03/asness-long-short-value-v2.png

Though I have not seen momentum factor returns back to 1920s, they probably similarly persisted.

Many (most?) quantitative strategies can ultimately be decomposed into some sort of a value/contrarian/mean-reversion component and/or a momentum/trend following component. Since the underlying value and momentum factors have persisted for many decades, many strategies that rely on them have done similarly well.

If you have a system that's been tested vigorously with a positive expectancy, you could also see if it's possible to increase the number of trades by using additional securities in the same asset class or other asset classes.You would increase returns by sheer volume, assuming,of course, expenses are reasonable.You also can have multiple strategies that perform well in different market environments as part of a larger aggregate strategy,where drawdowns would be diminished by offsetting profitability of one system vs. another. Understanding how a system is expected to behave under different market conditions should help investors feel comfortable during normal drawdown periods.I agre with Greg that,for most traders, a small number of trades usually account for the bulk of profitability and losses in any given year.As to increasing the win/loss percentage, I don't believe that's the answer to improving system results in many cases. I have seen too many traders focusing in on high win rates, while not focusing in on the more important factor, which is average dollar /percentage gain on a winning trade and average dollar/percentage loss on a losing trade. you could have a system that has a win percentage of 80% and still lose money if the average loss is many multiples of the amount gained on a winning trade.

Yes, you obviously want to pay attention to other performance metrics also, such as profit factor. In fact, there is no reason why you shouldn’t consider an objective function that explicitly combines various desirable performance measures, for example:

net profit * % win rate * profit factor

Another approach is to build the model using a data set spanning a different period. I did this with WFC using data from 1990, rather than 1970. Not only was the performance from 1990-2014 better, so too was the performance during the OOS period 1970-1989. Profit factor was 2.49 and %Win rate was 70% across the 44 year period from 1970. For the period from 1990, the performance metrics increase to 3.04 and 73%, respectively.

So in this case, it appears, a most robust strategy resulted from using less data, rather than more. At first this appears counterintuitive. But it’s quite possible for a strategy to be over-condition on behavior that is no longer relevant to the market today. Eliminating such conditioning can sometimes enable strategies to emerge that have greater longevity.

More here: http://jonathankinlay.com/index.php/2014/07/more-on-strategy-robustness/

Nicholas brings up an important point about diversification through multiple uncorrelated strategies. It is also very true that when a trading system is “too robust,” it can be a reason to worry.

It often turns out that a strategy that has a very high win/loss ratio is merely selling insurance. You can easily construct a strategy that has a win percentage of ~95% -- sell naked out-of-the-money options with delta of 0.05. If you sell catastrophic insurance, win percentage can be even higher.

These types of strategies may make money for years, only to give up all prior gains, and then some, in a single event. Or for decades, as buying junk mortgages did. Of course, if you can run these types of strategies in a systemically important financial institution using other people’s money and central banks’ liquidity, it is the economically rational thing to do...