I’ve written a few articles lately that look at the use of valuation metrics on investing strategies.
Those articles prompted a reader to ask if I could test a strategy called the EV/EBITDA pullback.
The idea behind this strategy is to buy a stock when the EV/EBITDA ratio deviates 30% or more below the historical average and sell when it returns to more normal levels.
(In case you’re wondering, EV/EBITDA is a fancy alternative to P/E).
I think this is an interesting idea because when a stock is trading below its historical earnings ratio it can be a good buying opportunity.
For example, Visa typically trades at an ev/ebitda of around 24. But when the ratio drops below its average it can be a good time to get long:
Setting Up The Test
In order to test this strategy I will be using data downloaded from Sharadar which includes fundamental daily metrics. I am using a database of 5700 US stocks which includes delisted securities but avoids micro caps.
The exact strategy rules we are going to test are shown below:
- EV/EBITDA is 30% or more below its 200-day average
- AND EV/EBITDA > 0 (using negative EV/EBITDA ratios makes no sense here).
- EV/EBITDA moves back above 200-day average
- OR after 200 days
- Universe: 5,700 US stocks provided by Sharadar
- Starting Capital: $50,000
- Max Portfolio Size: 25 positions
- Position Size: 4% (equal weight)
- Compound Returns: Yes
- Ranking: Enterprise Value (smaller stocks preferred)
- Liquidity: Avg. Turnover > $250,000
- Commissions: 0.05% per trade
- Execution: All trades placed on next day open
- Backtest Period: 1/2000 – 1/2014
System In Plain Language
In other words, we are going to buy stocks when the ev/ebitda ratio is more than 30% below the 200-day average and sell when the ev/ebitda ratio moves back above the 200-day average, OR after 200 days (whichever comes first).
We will hold a maximum of 25 positions at any one time and rank duplicate signals by enterprise value, preferring smaller companies first. Position size will be equal weighted and we will also use a liquidity filter to weed out illiquid stocks based on a 10-day average turnover.
Initial testing revealed that a raw signal of buying EV/EBITDA pullbacks was profitable on a large sample of trades. As the table shows below, all of the EV/EBITDA pullbacks tested beat the all trades benchmark result:
The next step is to turn this trading idea into a full blown portfolio investing strategy.
Example Trade Setup
The following chart shows exactly the kind of trade setup we are looking for with this strategy in Zillow Group (Z). Price is shown in the top pane and EV/EBITDA is plotted in the bottom pane:
You can see that EV/EBITDA is 39.7 on the 19th December 2012 which is more than 30% below the 200-day average (orange line). We therefore go long on the next day open at a price of 8.08.
Later, on the 11th March 2013, the EV/EBITDA ratio moves back above the 200-day average (red circle) so we exit on the next day open (red arrow) for a total profit of 98.63% after transaction costs.
Now we know the rules behind this system we will backtest the strategy on an in-sample period between 1/2000 – 1/2014.
Doing so achieved the following statistics and equity curve:
- # Trades: 721
- Annualised Return: 22.09%
- Risk-adjusted Return: 23.56%
- Max Drawdown: -52.37%
- CAR/MDD: 0.42
- Win Rate: 61.86%
- Sharpe: 0.28
As you can see, our pullback strategy has performed pretty well. Although the drawdown is large at -52%, the strong annualised return makes up for it.
For comparison, a buy and hold return over the same period was only 1.74% with a max drawdown of -56.47%.
Although our strategy looks pretty good so far I decided to also run a large scale optimization where I optimized the buy and sell thresholds and the average holding period.
The following table shows the top three permutations I tested based on risk-to-return:
In other words, the best strategy from our optimization was to buy stocks when EV/EBITDA is more than 15% below the 200-day average and sell when EV/EBITDA is more than 30% above the average, OR after 100 days.
This best performing run produced an annualised return of 25.64% with a win rate of 58.63% and a return-to-risk score of 0.47.
Full Sample Test
The top performing strategy from our optimization showed an excellent annualised return of over 25% in-sample so it’s time to put it to the test across the whole dataset and see how the performance holds up in the out-of-sample period.
The following results and equity curve now show the performance of our top performing strategy across the whole data period between 1/2000 – 11/2020:
- # Trades: 1403
- Annualised Return: 18.49%
- Risk-adjusted Return: 19.55%
- Max Drawdown: -54.10%
- CAR/MDD: 0.34
- Win Rate: 56.95%
- Sharpe: 0.26
The fact that this strategy moved to new highs in the out-of-sample is encouraging and the win rate is fairly stable over the sample period.
However, the annualised return has dropped from 25.6% in the in-sample to 18.5% in the full sample and the return-to-risk score has dropped to 0.34.
Some underperformance is to be expected since the in-sample segment was optimized. However, the strategy has actually underperformed buy and hold in the out-of-sample and the system is now in a drawdown since 2018.
In this article we looked at a strategy called the EV/EBITDA pullback and produced some interesting results.
A strategy that buys stocks when the EV/EBITDA is 15% below the 200-day average and sells when EV/EBITDA is 30% above the 200-day average (or after 100 days) shows modest profitability which extends into our out-of-sample period.
However, the out-of-sample performance was not very promising and results show it performed worse than buy and hold on SPY.
It should be also noted that a good percentage of gains appear to come from the ranking method that selects smaller cap companies. When the ranking method is reversed, the results are not nearly as good suggesting a lot of the outperformance could come from the size effect.
That said, the EV/EBITDA pullback did show some promise during the all trades test where it showed a positive return across a large sample of trades that comfortably beat our benchmark result.
It’s entirely possible that the idea does have merit but we went in the wrong direction with our portfolio settings. The pullback may have some use as part of a more sophisticated approach but more thorough testing is necessary before trading this strategy on its own.
Charts and simulations produced in Amibroker using historical data from Sharadar. Data is adjusted for cash dividends and includes delisted stocks.
Thank You For Reading
Joe Marwood is an independent trader and the founder of Decoding Markets. He worked as a professional futures trader and has a passion for investing and building mechanical trading strategies. If you are interested in more quantitative trading strategies, investing ideas and tutorials make sure to check out our program Marwood Research.
This post expresses the opinions of the writer and is for information or entertainment purposes only. It is not a recommendation or personalised investment advice. Joe Marwood is not a registered financial advisor or certified analyst. The reader agrees to assume all risk resulting from the application of any of the information provided. Past performance, historical or simulated results are not a reliable indicator of future returns and may not account for real world settings. Financial trading is full of risk and margin trading can lead to financial losses totalling more than what is in your investment account. We take care to present accurate analysis but mistakes in backtesting and presenting of analysis regularly occur. Please read the Full disclaimer.
Thank you to everyone who takes the time to leave a comment. Your feedback, constructive criticism and identification of mistakes is welcome. In order to concentrate on work I may not have time to respond to all comments.