Timing Model Returns for 2008

Results are below for 2008, frictionless (no fees or trading costs are deducted), and monthly and yearly rebalances. The model is out-of-sample since 2006. I will be doing an update to the white paper by the end of the month, so stay tuned!

S&P 500: -36.77%
TIMING: 1.33%

EAFE: -43.06%
TIMING: -8.17%

10 YEAR: 21.51%
TIMING: 21.51%

GSCI: -46.49%
TIMING: 1.08%

REITs: -37.33%
TIMING: -9.67%

Monthly Rebalance, equal-weighted:
B&H: -29.76%
TIMING: 1.59%

Yearly Rebalance, equal-weighted:
B&H: -28.43%
TIMING: 1.22%

That's 36 profitable years in a row for the timing model (just barely, thanks to bonds ripping in November and December!).

Click on table to enlarge:

Comments

[BWEL fan:]

Ok, so what are you waiting for? If your model is this good, why are you sharing it? You should be running a hedge fund......

Jan 07 10:26 AM

[Augustus:]

Any model which does not have a losing year is certainly remarkable. When I've looked at it in the past I noted that it does have long periods of underperformance vs the SPX when that index is on a several year uptrend. I've wondered if the model could be improved by trying to capture more of the gains when the different sectors are comming off their bottoms. As we've seen in 2008, avoiding the selloffs and maintaining profits earned is a good method. Can it be improved by using a different ma for the re-entry. Maybe using the 100 dma as the get back in would capture more of the gain from the bottom. Use the 100 dma until that one is above the 200 dma, then continue following the 200 dma as you've previously done. Perhaps you can run your data through that adjusted system and let us know how it turns out.

Jan 07 12:56 PM

[John Lounsbury:]

Mebane - 2008 was a great year for most timing models. The real test of a timing model is how it perfroms over a full business cycle. I agree with Augustus. You should keep on running hypotheticals against your models and keep looking for advantageous refinements. But I expect you already know this.

Thanks for publishing your work. My comment to Ben is that I have always gotten more (from readers) out of publishing my ideas than I have given away. Even the act of formally writing down a description sharpens the details in my own mind.

Jan 07 01:10 PM

[User 343684:]

Mr. Faber,

I have also discovered that using the 200EMA is an excellent tool for timing entries or exits for someone using the Yale portfolio allocation. My backtesting using index funds goes back to 1996. I created another one using ETFs that goes back to 2003. My backtesting does take into account maint, trading fees & dividends. The margin cost is not taken into account since I do not have exact figures yet. What's a reasonable % to use? That will make the results more accurate.

In any case, my entry/exit rules are slightly different from yours but I still use the 200EMA on monthly prices:

The backtesting yielded the following results:

Using index funds and 50% margin:
Annualized gain: 12.45%
Total gain: 172%
Worst drawdown: -0.91%
Worst loss: -0.77%
Total months: 100
Winning trades: 14
Losing trades: 5
Longest trade: 56 months
Shortest trade: 1 month

Using ETFs with 50% margin:
Annualized gain: 15.70%
Total gain: 126.92%
Worst drawdown: -0.36%
Worst loss: -0.37%
Total months: 66
Winning trades: 8
Losing trades: 3
Longest trade: 56 months
Shortest trade: 1 month

The drawdown is computed using closed trades and not the market's daily volatility. In other words, once a trade was closed, I took last peak in equity and computed its drop caused by the closed trade. The same applies to the "worst loss" field. This part of the research needs improvement.

Margin could be bumped to 100% w/o incurring a huge loss since the 200EMA works quite well in the Yale portfolio. In such a case the worst loss was 1% & 0.40% and the ann. gains were 15% & 19.46% respectively.

Sincerely,

Jan 24 10:49 AM