- This system provides investment selections according to stock market climate. During up-markets it is long equity ETFs and switches to bond- and gold ETFs during neutral and negative markets.
- It invests periodically in only two ETFs, appropriately selected for the prevailing market climate.
- Four different approaches are used to assess the market climate. A simple summing grades the market climate into five segments.
- A backtest from Jan-2000 to Jul-2016 shows an average annual return of 16% with a maximum drawdown of -17% and only 61 realized trades during this 16.5 year long period.
Market climate measurement:
The measurements of market climate are in some respects similar to Fred Piard’s system. The system described herewith uses the following economic indicators:
- Russell 1000 Index.
- S&P 500 earnings per share estimate (SPEE), based on the time weighted analyst’s estimates of current year and next year.
- Short Interest Percent of Float for the S&P 500 (SPSI).
- The Aggregate Spread, similar to the Enhanced Aggregate Spread, is a recession indicator with a long lead time, which uses:
- 20-year US Treasury bond yield (Y20),
- Target Federal Funds Rate (FFR),
- Consumer Price Index (CPI),
- Unemployment Rate (UER).
- The Hi-Lo Index of the S&P 500 measures the excess of the number of shares reaching new 3-month highs versus the number of shares forming new 3-month lows.
The above five indicators are used in four models which indicate a weak market climate when:
- The unemployment rate is on the increase: UER now > UER 3-mo ago.
- The earnings estimate of the S&P500 declines: SPEE now < SPEE 20-wks ago.
- The markets are declining: Short moving Average (MA)< long MA of the Russell 1000 Index.
- The three following conditions are true:
- Recession indicated: (Aggregate Spread 10-mo ago < 150 bps) and
- Negative market sentiment: (short MA < long MA of the Hi-Lo Index) and
- Negative market sentiment: (short MA > long MA of SPSI).
Market Climate Score:
Scoring 1 when any of the above four market climate models (1. thru 4.) is true, and summing the scores, results in the market climate score (MC-score).
MC-score = 0 (best market climate)
MC-score = 1 (positive market climate)
MC-score = 2 (neutral market climate)
MC-score = 3 (negative market climate)
MC-score = 4 (worst market climate)
See the Appendix for the performance of SPY during the five MC-score defined market climates.
The iM MC-Score System:
Using the market clime score, a trading system is devised that selects from following ETFs:
| ETF | Description |
|---|---|
| XLP | Consumer Staples Select Sector SPDR Fund ETF |
| USMV | iShares Edge MSCI USA Minimum Volatility ETF |
| VNQ | Vanguard REIT ETF |
| EMB | iShares JPMorgan USD Emerging Markets Bond Fund ETF |
| SH | ProShares Short S&P500 ETF |
| GLD | SPDR Gold Trust ETF |
The system always uses two ETFs (equal dollar weighted), and invests in alternative market segments during different market climates, as follows:
For MC-score = 0 : USMV and XLP.
For MC-score = 1 : USMV and XLP.
For MC-score = 2 : VNQ and EMB.
For MC-score = 3 : GLD and EMB.
For MC-score = 4 : GLD and SH.
Notes:
- The model was backtested using the on-line portfolio simulation platform Portfolio 123, which also provides extended price data for ETFs prior to their inception dates calculated from their proxies.
- USD Emerging Markets Bond Fund EMB is used, because according to Blackrock and Research Affiliates USD EM debt is expected to produce much higher returns than US Bonds and Treasuries.
Trading:
ETFs are bought and sold on the first trading day of the week after a signal is generated.
Transaction slippage and brokerage fees were assumed to be 0.1% of each trade amount. The model is rebalanced weekly, but portfolio turnover is low, about 2-times a year on average, or about 4 buy- and sell-trades per year.
The only buy rule specifies the two ETFs for each of the five MC-score defined market climates. There is only one sell rule to ensure a minimum holding period of about three weeks. There is no ranking system needed.
Historic performance of the iM MC-Score System:
In the Figures-1 and 2 the red graph represents the model and the blue graph shows the performance of benchmark SPDR S&P 500 ETF Trust ETF SPY.
Figure-1: Simulated Performance 2000-2016: Annualized Return= 16.0%, Max DD= -16.9%. The backtest period was from Jan-2000 to Jul-2016.
Figure-2: Simulated Performance 2009-2016: Annualized Return= 16.3%, Max DD= -12.5%. The backtest period was from Jan-2009 to Jul-2016.
Rolling 1-year Returns:
Over any 1-year period, starting Jan-2000 and 1 week offset intervals, this model showed returns ranging from -6% to +56%, whereas SPY returns ranged from -39% to +60% over the same periods. There were only 13 sample periods out of 813 which had small negative returns, whereas for SPY there were 238 sample periods with negative 1-year returns.
Risk Measurements:
Table-1 lists the annualized returns, maximum drawdowns, and various risk measurements for the trailing three years and for the full backtest period.
Trading Statistics:
This model assumes a starting capital of $100,000, and trading costs/slippage of 0.10% of each trade amount. Note, that the total amount of all Losers is only about 7% of the total amount of all Winners. This model has a high win rate of 81% of all realized transactions.
Number of days invested from 1/3/2000 to 7/29/2016 (6,052 days) in:
| XLP | 4,498 |
| USMV | 4,491 |
| VNQ | 586 |
| EMB | 1,374 |
| SH | 217 |
| GLD | 938 |
Performance Statistics:
Following the Model:
The backtests indicate that a successful strategy would be to invest in ETFs having styles that are applicable to prevailing market climate as defined by the MC-Score System.
This model can be followed live at iMarketSignals, where it will be updated weekly starting end of August 2016.
Appendix: Performance of SPY during various market climates
MC-score = 0 (best market climate)
MC-score = 1 (positive market climate)
MC-score = 2 (neutral market climate)
MC-score = 3 (negative market climate)
MC-score = 4 (worst market climate)
Disclaimer:
One should be aware that all results shown for the MC-Score System are from a simulation and not from actual trading.
All information for this model is back-tested, based on the methodology that was in effect on the launch date. Back-tested performance, which is hypothetical and not actual performance, is subject to inherent limitations because it reflects application of a methodology and selection criteria in hindsight. Actual returns may differ from, and be lower than, back-tested returns.
All results are presented for informational and educational purposes only and shall not be construed as advice to invest in any assets. Backtesting results should be interpreted in light of differences between simulated performance and actual trading, and an understanding that past performance is no guarantee of future results. All investors should make investment choices based upon their own analysis of the asset, its expected returns and risks, or consult a financial adviser.
Hi, is this the new market timer that can be used for the hedge entry rules for your other systems like BestogaX-5, Bestoga3 and HiD-LoV-7? Does it improve performance over using the Standard Market Timer model?
Also a second question- if the EMB bond fund is expected to produce higher returns than US funds, would it be a better choice than IEF/TLT for the models in Combo5 that switch to bonds when they are out of equities?
Thanks.
The MC-score can be used for hedging purposes:
enter hedge when MC-score > 2
exit hedge when MC-score < 3 Performance is better when using the Standard Market Timer signals for hedging. Ultimately investors have to decide for themselves what type of bond funds to invest in. With rising interest rates on the horizon the past performance of bonds is not indicative of expected future performance.
I like this model. I appreciate seeing the performance of SPY for each score.
I am sure one could suggest a number of different investment permutations depending on the score. But could you run one where the investment remains only in VTI when the score is 0 or 1? Total and annual return would be of interest.
This model needs two ETFs per MC-Score. With VTI and VV when the score is 0 or 1, from Jan-2000 to Aug-2016:
Total Return….862.36%
Annualized Return….14.59%
Annual Turnover….179.63%
Max Drawdown….-19.99%
Overall Winners….(47/63) or 74.60%
Sharpe Ratio….1.01
How might one use this system with self-directed retirement accounts that restrict investements to mutual funds (no ETFs or Stocks)? Thanks.
We are preparing a model for Vanguard ETFs which all have Mutual Fund proxies. There are six ETFs for each MC-Score.
From Jan-2000 to Aug-2016:
Annualized Return….13.66%
Annual Turnover….122.86%
Max Drawdown….-16.07%
Overall Winners….(88/110) or 80.00%
Sharpe Ratio….1.24
On most of your models, you show trade history.
Can you do the same for this model?
Symbol …. Open … Close … Days … Pct
XLP …. 06/06/16 … … …
USMV … 06/06/16 … … …
EMB …. 05/16/16 … 06/06/16 … 21 … 0.07%
VNQ …. 05/16/16 … 06/06/16 … 21 … -1.77%
XLP …. 04/25/16 … 05/16/16 … 21 … 2.20%
USMV … 04/25/16 … 05/16/16 … 21 … 1.23%
EMB …. 12/21/15 … 04/25/16 … 126 … 4.47%
VNQ …. 12/21/15 … 04/25/16 … 126 … 3.92%
XLP …. 02/06/12 … 12/21/15 … 1414 … 40.72%
USMV … 02/06/12 … 12/21/15 … 1414 … 39.81%
VNQ …. 12/19/11 … 02/06/12 … 49 … 12.37%
EMB …. 12/19/11 … 02/06/12 … 49 … 1.86%
USMV … 10/25/10 … 12/19/11 … 420 … 12.45%
XLP …. 10/25/10 … 12/19/11 … 420 … 7.74%
VNQ …. 10/04/10 … 10/25/10 … 21 … 5.19%
EMB …. 10/04/10 … 10/25/10 … 21 … 1.71%
USMV … 07/13/09 … 10/04/10 … 448 … 26.13%
XLP …. 07/13/09 … 10/04/10 … 448 … 17.38%
EMB …. 10/27/08 … 07/13/09 … 259 … 28.16%
VNQ …. 06/22/09 … 07/13/09 … 21 … -0.74%
GLD …. 06/02/08 … 06/22/09 … 385 … 6.82%
SH …. 10/06/08 … 10/27/08 … 21 … 19.71%
EMB …. 12/24/07 … 10/06/08 … 287 … -10.28%
VNQ …. 05/05/08 … 06/02/08 … 28 … -3.75%
GLD …. 12/24/07 … 05/05/08 … 133 … 6.17%
XLP …. 09/25/06 … 12/24/07 … 455 … 14.61%
USMV … 09/25/06 … 12/24/07 … 455 … 12.06%
EMB …. 09/05/06 … 09/25/06 … 20 … 1.58%
VNQ …. 09/05/06 … 09/25/06 … 20 … 0.04%
USMV … 05/27/03 … 09/05/06 … 1197 … 40.47%
XLP …. 05/27/03 … 09/05/06 … 1197 … 23.83%
VNQ …. 05/05/03 … 05/27/03 … 22 … 4.45%
EMB …. 05/05/03 … 05/27/03 … 22 … 3.11%
USMV … 03/10/03 … 05/05/03 … 56 … 12.10%
XLP …. 03/03/03 … 05/05/03 … 63 … 3.02%
VNQ …. 02/18/03 … 03/10/03 … 20 … 0.07%
EMB …. 10/07/02 … 03/03/03 … 147 … 13.20%
GLD …. 12/02/02 … 02/18/03 … 78 … 8.40%
VNQ …. 10/07/02 … 12/02/02 … 56 … 6.37%
XLP …. 08/05/02 … 10/07/02 … 63 … -4.19%
USMV … 08/05/02 … 10/07/02 … 63 … -5.57%
VNQ …. 05/20/02 … 08/05/02 … 77 … -5.57%
EMB …. 05/20/02 … 08/05/02 … 77 … -6.35%
XLP …. 04/08/02 … 05/20/02 … 42 … -1.84%
USMV … 04/08/02 … 05/20/02 … 42 … -2.02%
VNQ …. 02/19/02 … 04/08/02 … 48 … 9.77%
EMB …. 01/14/02 … 04/08/02 … 84 … 4.40%
SH …. 01/28/02 … 02/19/02 … 22 … 4.01%
GLD …. 09/04/01 … 01/28/02 … 146 … 1.98%
SH …. 12/24/01 … 01/14/02 … 21 … 0.25%
EMB …. 10/08/01 … 12/24/01 … 77 … 3.79%
SH …. 08/20/01 … 10/08/01 … 49 … 8.82%
EMB …. 08/13/01 … 09/04/01 … 22 … 2.28%
GLD …. 02/05/01 … 08/20/01 … 196 … 3.87%
SH …. 06/11/01 … 08/13/01 … 63 … 4.85%
EMB …. 04/23/01 … 06/11/01 … 49 … 9.12%
SH …. 04/02/01 … 04/23/01 … 21 … -7.08%
EMB …. 03/12/01 … 04/02/01 … 21 … -2.57%
SH …. 02/20/01 … 03/12/01 … 20 … 7.91%
EMB …. 11/20/00 … 02/20/01 … 92 … 6.86%
VNQ …. 11/20/00 … 02/05/01 … 77 … 8.12%
XLP …. 01/03/00 … 11/20/00 … 322 … 18.40%
USMV … 01/03/00 … 11/20/00 … 322 … 0.71%
Thanks
Can you list the periods when the score was 0? Can’t tell from the above data since score 0 and score 1 are invested the same.
Thanks.
Open – Close
01/03/00 – 08/07/00
10/02/00 – 11/06/00
04/22/02 – 05/06/02
09/08/03 – 04/05/04
05/03/04 – 08/16/04
11/08/04 – 12/05/05
01/03/06 – 07/17/06
10/02/06 – 02/05/07
03/05/07 – 07/02/07
11/05/07 – 12/03/07
02/08/10 – 05/03/10
06/07/10 – 07/06/10
11/08/10 – 12/06/10
01/03/11 – 07/05/11
08/08/11 – 08/15/11
02/27/12 – 09/24/12
10/22/12 – 12/31/12
03/04/13 – 12/22/14
06/08/15 – 08/31/15
06/13/16 – 09/06/16
Thank you. The latest score is 0. When did that begin?
The latest score is 1.
Do you have a time frame for the Vanguard ETF version? (Six for each score sounds very interesting.)
We are launching a 4-ETF version in a few days.
Hi Georg and Anton
Interesting and certainly worthy of analysis and consideration….Thanks.
I’ve been staring at Figures 4-8 under the Appendix: Performance of SPY during various market climates – but can’t understand them.
The blue line indicates the SPY(?) The red is the Model(?) I don’t understand the relation between the 2 lines as the MC score changes. Could you please explain these graphs a bit more??
Do you have any thoughts about using 2x funds with this strategy??
Many thanks,
Vman
In Figures-4 to 8 the blue line represents the performance of SPY and the red line the performance of SPY for a certain Market Climate Score (MCS).
So for example, in Figure-4 the red line shows the performance of SPY for MCS=0. The horizontal flat lines represent the period when the MCS= 1, 2, 3 or 4. Since we only wanted to show performance for MCS=0, the flat lines represent a cash only position.
In re-assessing which iM models I’m utilizing, I multiplied out the returns for all five MC 0 to 4 segments, but only get a about a 102% cumulative return. I’m not sure if the 5 charts are incorrect or if the total model 16 year hypo is off, but my assumption was that by stringing the 5 segments’ total returns together, it should come out close to the full model results. I’m not sure where the errors are coming from but it’d be great to recheck all the inputs and rerun in Portfolio123. Please correct me where I may have some misunderstanding of the backtest results posted. Thanks
Returns for all five MC 0 to 4 segments (Figs. 4 to 8) are shown for SPY – cash.
The model holds different ETFs for each of the 5 segments. Therefore stringing the 5 segments’ total returns together does not match the return for the model.
I have what I think is a common problem. I need a “quarterly” timing system for my 401k. All the funds have minimum 90 day holding periods (even the Vanguard Index funds), which make it very difficult to deal with the whipsaws of weekly or monthly timing systems (MAC-US etc.) The penalties for violating the holding periods are very high (2% of your entire account).
Do you guys have a system that meets a quarterly time-frame? Or is there a system out there somewhere (that beats buy and hold)?
What funds are available in your 401k?
Or between which funds would you like to alternate?
Do you display this MC score on the site? If not, what is the current MC score according to this model?
Also, this is the MC score that is used in the iM Low Frequency Timer, correct?
Thanks.
Tom C
For 3/30/2020 the MC-score = 2 for this model. It is not necessary to display it on the site because the ETF holdings are indicative of the score.
The same formula for MC-Score is used in the Low Frequency Timer together with several others.