Back testing is for losers

I had so much fun on this week’s podcast with @kevinrmuir who is really one of the smoothest interviewers in the podcast business making you feel like you are just shooting the breeze over a couple of beers. But as I started to think this about this week’s column my mind drifted to an episode of the markethuddle.com from a few weeks back when Andrew Beer - a man who specializes in offering low cost hedge fund replication strategies to retail investors - made a throwaway comment that made my ears perk up.

Andrew noted that in their studies of hedge fund returns almost all the alpha occurred within a few first years of performance and for the rest of the time superstar managers simply coasted on their reputation. Don’t believe it? Just pull out any copy of the Market Wizard series by Schwager and take a look at the most immediate ten year returns of the great traders profiled in the books  and you will be shocked at the mediocre results. 

There are of course exceptions to that rule including Soros and Drukenmiller and especially Renaissance - but they are exceptions that  prove the rule as all three titans closed off access to their investment strategies a long time ago making it impossible to participate in their alpha even if you wanted to.

More insidious, in my opinion, is the case of Bill Miller who beat the S&P for 15 years in a row only to absolutely destroy investors portfolios during the GFC of 2008-2009. But here is the thing that kills me. Miller eventually recovered his momo ways and if you average his returns over the whole lifetime of his career he managed to squeak out a tiny premium to the S&P 500. But do you think investors who started with him in 2008 have anything to show for it? No, they are probably still underwater after more than a decade of being in the market or at breakeven at best.

All of this is not a rant against Bill Miller or any other money manager out there - god knows we’ve all had our drawdowns - but rather against the absolute folly of  seeking validation for your investment through a track record or a back test. In many other aspects of life, especially in physical matters like engineering a backtest is the absolute best way to make a decision. We couldn’t build roads, buildings, rockets if we didn’t have the historical data to prove that they work. But the very same mindset applied to finance is a disaster waiting to happen. In fact I would say that the single worst way to evaluate the performance of a trader is by their track record. 

How many times have you read stories of Morningstar research showing that top performing money managers over the past 3 years almost always end up in lowest quintile over the next 3 years  while the worst performing managers, if they are allowed to stay in business, almost always beat the market during that period.  In fact, the much better strategy in investing is to always look at the absolute worst performing funds over the past few years and buy them all because you will have a much better chance of beating the market that way, rather than buying the superstars of the moment.  Alas that really isn't practical as investors will flee those funds and they will close up shop.

Which brings me to my point - track record seeking or backtesting is simply a variation of the FOMO theme. Investors and traders are just chasing the profits of the past when the actual money is to be made in the unknown future. That's why every beautiful backtest fails the moment it trades live and every bet on a hot manager almost always turns sour very quickly.

Does that mean that I am arguing for no backtesting whatsoever? No, of course not. Anytime you have an idea for a setup you need at least a modicum of validation that it has worked in the immediate past. But a backtest is just the very start of the process. The true art and alpha of trading lies in technique. Just as your grandmother’s recipe for ragu or gefilte fish can taste sublime while yours tastes blah, just as one surgeon can make an incision with minimal blood loss while another one will spout a gusher, just as one basketball player can make a fadeaway jumper seem effortless while another banks a brick after brick the difference between a pro and hack in almost every discipline of life is not knowledge but technique. 

And technique requires human practice and refinement. That’s why endlessly optimizing backtests of your algo or running thousands of spreadsheet screens on money managers is not only lazy but counterproductive. You’ll never find your investment success in reams of statistics of the past, but only in insights  about market behavior of the present and immediate future. 

My favorite way to “practice” and refine my techniques is to run countless hours of price action on the Trading View replay function - it beats the hell out of watching endless hours of TikTok ( though I do that too) and occasionally you will see something tiny that can make a huge difference in your edge.

Lastly, let me leave you with a disclaimer. Everything I write pertains only to trading and even more particularly day trading. There IS actually a very provable model of investing that has worked  for centuries specifically in US equity markets - dollar cost averaging into the index. This process eliminates both idiosyncratic and duration risk by sampling equity exposure over a wide variety of securities and time periods. So yes if you want to believe in a backtest that works - this is the only that does.