Backtesting The 12% Solution Monthly Trading Strategy
by David Alan Carter
Backtesting is simply looking back over the years to see how a particular investing or trading strategy would have performed under market conditions. I believe in backtesting: in the book, The 12% Solution was backtested for the 10 full years of the Buffett Bet (that would be the years 2008-2017).
In addition, in Appendix A of the book, I ran a full 19-year backtest using a modified strategy: I replaced the ETFs in our model with mutual funds tracking similar indices. This was necessary as not all of the ETFs in the model existed prior to 2007. This modified model gave us the opportunity to see how the strategy would have performed going all the way back to the year 2000, and capturing the Dot Com Crash of 2000-2002.
While not perfect, backtesting provides a measure of light in an otherwise dark sea of investing.
In addition, in Appendix A of the book, I ran a full 19-year backtest using a modified strategy: I replaced the ETFs in our model with mutual funds tracking similar indices. This was necessary as not all of the ETFs in the model existed prior to 2007. This modified model gave us the opportunity to see how the strategy would have performed going all the way back to the year 2000, and capturing the Dot Com Crash of 2000-2002.
While not perfect, backtesting provides a measure of light in an otherwise dark sea of investing.
Backtesting vs. Live Trading
I frequently get asked some variation of the following:
- "Are performances derived from backtests or from real 'live' portfolios?"
- "Is there any actual results anywhere or is the backtest the best way to present the material?"
All the performance results you see in the book, on the website, in the monthly newsletters and on the Members Page (available to readers of the book) are technically backtested results, even though live trading in the strategy has been ongoing and documented since 2017. By backtested results, it means I use the strategy's algorithm to display the returns, the volatility, the drawdowns, and the like - whether looking back one day, one month, or ten years.
Why post backtested results vs. live trading results? Because backtested results represent the strategy in its purest form.
For example, the algorithm behind the model calculates its rebalancing trades at the close on the last trading day of the month. It then places those trades either:
And it's those closing prices it uses in calculating returns, drawdowns, etc. [By the way, the rebalancing trades are the same regardless of the date trades are placed.]
On the other hand, live trading is subject to a number of variables depending on who is doing the trading. While it's certainly possible to emulate the model and rebalance at the close on the last (or first) trading day of the month, most folks - myself included - find it easier to place their trades prior to the close - sometimes hours earlier and at their convenience. As the market is always moving, if I make a trade at noon and the strategy makes the same trade at close, that four-hour price swing will either be to my advantage, or detriment, by the end of the month when the strategy posts results. So, I might beat the strategy or lag the strategy for the month - although such differences are usually slight over time.
Another example: humans have been know to panic (shocking!). If the market is going haywire on rebalancing day, a human investor might delay a trade for a day or two to see how things shake out. Or that human might decide to allocate a portion of the portfolio to cash (not a bad idea sometimes). Or that human might decide he knows better than the model which ETFs will fare better in the upcoming month, and switch to something else.
Not so the algo. The algorithm is a cold-hearted machine that follows the plan, executes trades at precise times, and never succumbs to emotion when the going gets tough.
So, in the interest of accuracy and consistency, I use the algorithm when posting results. In my experience, and to sum things up, any differences between live and backtested data can be attributed to 1) the slight difference in execution times for the trades, and 2) whether or not one sticks with a particular strategy through thick and thin, as opposed to acting on emotion and trading in and out of the strategy.*
_____
For more detail on modeling vs. live trading, see the section marked "Additional Risk Disclosure for System Trading" in the website's Disclaimer.
Why post backtested results vs. live trading results? Because backtested results represent the strategy in its purest form.
For example, the algorithm behind the model calculates its rebalancing trades at the close on the last trading day of the month. It then places those trades either:
- At the close of the last trading day of the month (as per the book), or
- At the close of the first trading day of the month (as per the newsletter and Members Page).
And it's those closing prices it uses in calculating returns, drawdowns, etc. [By the way, the rebalancing trades are the same regardless of the date trades are placed.]
On the other hand, live trading is subject to a number of variables depending on who is doing the trading. While it's certainly possible to emulate the model and rebalance at the close on the last (or first) trading day of the month, most folks - myself included - find it easier to place their trades prior to the close - sometimes hours earlier and at their convenience. As the market is always moving, if I make a trade at noon and the strategy makes the same trade at close, that four-hour price swing will either be to my advantage, or detriment, by the end of the month when the strategy posts results. So, I might beat the strategy or lag the strategy for the month - although such differences are usually slight over time.
Another example: humans have been know to panic (shocking!). If the market is going haywire on rebalancing day, a human investor might delay a trade for a day or two to see how things shake out. Or that human might decide to allocate a portion of the portfolio to cash (not a bad idea sometimes). Or that human might decide he knows better than the model which ETFs will fare better in the upcoming month, and switch to something else.
Not so the algo. The algorithm is a cold-hearted machine that follows the plan, executes trades at precise times, and never succumbs to emotion when the going gets tough.
So, in the interest of accuracy and consistency, I use the algorithm when posting results. In my experience, and to sum things up, any differences between live and backtested data can be attributed to 1) the slight difference in execution times for the trades, and 2) whether or not one sticks with a particular strategy through thick and thin, as opposed to acting on emotion and trading in and out of the strategy.*
_____
For more detail on modeling vs. live trading, see the section marked "Additional Risk Disclosure for System Trading" in the website's Disclaimer.
When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works.
|