Dissecting the 10 Minute Strategy

We analyze a popular trading strategy with some surprising conclusions.

Like what you're seeing?
Try our FREE 30 day trial And test your strategies against historical data immediately

Written by Liam Flavelle on 13 July 2017

Summary

  • I recently published a backtest of Lino Patti's 10-Minute Strategy showing 17% annual gains for the past 17 years.
  • In this article, the strategy is analyzed to identify which factors actually add value and which ones don't.
  • We demonstrate how popular factors don't always work well together.

In Lino Patti's recent article, '10-Minute System' Beats Market By 17% In 17-Year Backtest', he shows off a backtest we recently completed based roughly on the rules he uses to identify good, undervalued companies. This backtest showed us that the strategy would have returned 17% a year over a 17-year period.

Lino's article attracted some interesting comments and constructive criticism, namely:

  • The typical stocks selected were micro- or nano-companies.
  • Rules such as P/E less than 10 and P/Book less than 1.2 are too restrictive.
  • The rules are fairly complex (there are a lot of data points to track).

In this article, I'll refine our initial strategy in an attempt to address these issues.

You can access extended statistics, charts, and position data for this model on the InvestorsEdge.net platform by clicking on any of the charts in this article or by clicking here. All the versions discussed in this article can be accessed by clicking on the history button in the left menu.

Our Baseline

Lino's original strategy is based on Benjamin Graham's ideas and looks to identify quality companies that are deeply undervalued. This results in few trading opportunities as few stocks meet all of his criteria. To address this, I have adjusted his trading strategy to equally rank all qualifying companies by his 11 factors and selecting the stocks that appear at the top of the list.

Our baseline strategy is as follows:

  • Our stock universe will consist of all stocks that meet the following criteria:
    • US shares and depository receipts
    • Market cap greater than US$ 50m (removing nano stocks from the equation as per one of the critiques)
    • Annual EPS greater than EPS five years ago
    • Debt to NCAV ratio is greater than 0
  • We will start with US$ 10,000 in cash.
  • Each transaction will attract a US$ 7 flat fee commission charge.
  • Each month we will open positions in the top 30 stocks ranked by:
    • Current Ratio
    • Debt/NCAV
    • EPS/previous quarter’s EPS
    • Number of dividend increases in the last 10 years
    • Yield
    • Payout Ratio
    • Free cash flow/Dividends
    • 5-yr Dividend growth
    • 3-yr Dividend growth/5-yr Dividend growth
    • Price to Book
    • P/E

I have backtested this strategy from 1st January 2000 until 7th July 2017 with the following results:

Baseline results

Not a bad result – the strategy would have turned our initial US$ 10,000 into US$ 144,000 and, ignoring the 2007-8 financial crisis, would have had maximum drawdowns of around 30%:

Baseline drawdowns

If you analyze the results, you can see that over 85% of our returns came through capital gains with the rest through distributions to shareholders and spin-offs, which is to be expected as this Benjamin Graham-inspired strategy focuses on quality companies that have been mis-priced by Mr. Market.

Refining the Universe Selection

Over the last 17 years, the universe selection has returned between 330 and 800 stocks out of the 6,000 or so US shares and depository receipts with market capitalizations greater than US$ 50m.

The universe selection included two further rules:

  • Last twelve-month rolling sum of a company’s EPS should be greater than the previous quarter’s rolling total.
  • Long Term Debt to Net Current Asset Value should be greater than 0.

When I disabled these rules individually and re-ran the backtest, the numbers came out as below:

Rule

CAGR

Drawdowns

Ex EPS

12.2%

60%

Ex Debt / NCAV

13.9%

63%

Removing either of these rules would have resulted in our returns dropping from 16% to around 13% a year.

I realized when analyzing this strategy that I had made a mistake with this ratio – Lino’s original rules called for the classic Graham rule of a Debt to NCAV ratio of less than 1.1. Adding this original rule reduced returns to 10% a year, so it was a useful mistake to make. Since debt can’t be a negative number, I can state that the best rule for this strategy is to only select stocks with a positive NCAV.

Refining the Ranking Factors

Below is a table displaying results of backtesting our strategy, removing each of the 11 factors one by one:

Factor

CAGR

Drawdowns

Ex Debt/NCAV

19.4%

61%

Ex Current Ratio

18.8%

64%

Ex Num. Div Increases

18.4%

61%

Ex EPS Growth

17.0%

63%

Ex Payout Ratio

16.9%

61%

Ex 5Yr CAGR Dividends

16.8%

63%

Ex FCF/Dividends

16.7%

63%

Ex Yield

16.5%

63%

Ex 3yr/5yr CAGR Dividends

15.9%

63%

Ex P/Book

11.2%

63%

Ex PE

9.1%

61%

You can see from the table that our two classic value factors, Price to Book and Price to Earnings, contributed the most to our strategy, whilst the other factors generally acted as a drag on returns.

Building Our New Model

When the best seven factors are combined together, we get the following simulated results:

Best Original Factors

Combining the best performing factors together has increased our returns by 7% annually over the 17 years. Capital gains still would have made up just over 85% of our profits, and our trades would have ended in profit around 60% of the time.

Best factors market cap

You can see from the market capitalization chart that the companies that the revised strategy invests in are still mainly micro and small (between US$ 50m and 2bn). You often find this with value strategies - the companies that we are looking for are often regarded either as junk by the market, or there has been an overreaction to a series of events that have driven the price (and therefore the market cap) down.

So, Can We Trade This Strategy?

A 23% annual gain over 17 years is great, but how did our modified strategy perform recently? If we had started trading this system in 2010, the answer is "less great":

Best factors in sample

When we look at the annual performance and drawdowns chart, it becomes obvious as to why:

Best factors in sample annual performance

Best factors in sample drawdowns

The euro crisis of 2011 and the oil price drop and taper tantrum of 2015/16 would have dramatically impacted our returns.

Will this strategy come back to profit? My gut feel is 'probably' - value strategies generally have had a torrid few years recently, and a lot of market commentators are predicting that they will return to form in the near future.

Your Takeaway

Some factors go in and out of fashion, sometimes for years. Other metrics you use may be eminently sensible (who doesn't want to buy stocks that have low debt, great dividend histories, quality balance sheets and low prices?), but do you really know that they actually improve your trading results?

The key reason that factors don't play well together all the time is that we are really using them for two purposes:

  • To identify that they are cheap stocks with unrealized value.
  • To identify that now is a good time to buy them.

Combining factors to identify quality companies works well pretty much all the time, but they can and often do trigger conflicting signals as to the right time to enter and exit positions.

While this hasn't been a critique of Lino's original strategy, or any other Benjamin Graham-inspired models out there, I believe that there are some useful lessons to be learned here about how mixing factors together doesn't always make for safer investing decisions.

Now, that we've all learned about some factors that don't work well together, the second part of this series will identify ones that do so consistently and profitably.

Comments

You can read comments but to make comments you need to be logged in.

Comments
Like what you're seeing?
Try our FREE 30 day trial And test your strategies against historical data immediately