ULTRA System Report (11/05/02)

The Thursday Seasonal System

This system is new in ULTRA 7.10 even though ULTRA has contained the ability to model the phenomenon using the FRIDAY system for many years. Thursday strength often carries into Friday. This has been a very reliable phenomenon in real-time for many years. There are many ways to model this phenomenon with price and breadth. We prefer breadth because it is more stable than price in the last 15 minutes of trading.

For example:
It is common for the SP500 price will move wildly in the last 15 minutes of trading but the NYSE Advance-Decline Line won't change significantly. Therefore, if one wants to act on a signal in the closing minutes of trading they are more likely to be correct when using a breadth indicator than a price indicator.

One way to model Thursday strength is via:

ADR = NYSE Advances / NYSE Declines
PRE-EOW = Day before the last day of the week (usually Thursday)
BUY: DAY is the PRE-EOW and ADR > 1.0
SELL: At the close of the last day of the week (usually Friday)

The Historical Results (Jan42-Nov02) of this system are:

ADR > 1.0 1.2 1.3 1.4 1.5 1.6
Trades/Yr 28 22 19 16 14 12
Winners 66% 66% 67% 67% 68% 70%
Max Drawdown -8.7% -8.7% -8.7% -8.7% -8.7% -6.7%
CAR While Invested +64% +74% +83% +78% +83% +99%

While all the ADR values greater than 1.0 produce excellent results, requiring higher values of ADR make the system more selective resulting in lower drawdowns and higher CARs while invested. (CAR = Compounded annual return).

We settled on ADR = 1.3 for the default setting of this system.

Knowing that any value over 1.0 is profitable is valuable. Assume that on a given Thursday nearing the close and your trading decision deadline. The ADR > 1.3 and you take a long position but then the market sells off in the last few minutes and ADR closes at 1.25. Even though you ended up on the wrong side of your system, you are still set up for a likely bullish scenario on Friday according to the historical results in the chart above.

Testing the THURS System

In ULTRA, systems are tested via the Timing/Historical Analysis menu item. Systems with optional settings are set up via the Options/Timing System Options menu item.

  1. Select Options/Timing System Options/THURS (v)
  2. Select "*.Original Settings"
  3. Click on "SELECT" An Options Screen will be displayed where the THURS options can be modifided.
  4. Note that the system is set up as: _X_ Buy on DAY when(BBL) 1.3 < A/D Ratio < 10.0. Sell next day(BBH). DAY = PRE-End of week. These settings will buy on the PRE-End of week (Usually Thursday) if the A/D ratio (ADR) is between 1.3 and 10.0. The system will sell on the next day (I.e. only holds the position one day).
  5. Click on OK to close the Options Screen.
  6. Select Timing/Historical Analysis, click on THURS, and then OK.
  7. Type in 01/01/42 for Start Date, Click on SP500, SAME-DAY, and then OK. ULTRA will begin processing dates testing the THURS system as it was set up via Options. When the analysis is complete a file, thurs.dmp will be displayed containing the historical results.

The BWB Seasonal System

This new system models what is commonly referred to the Warwick Bounce after its creator Ben Warwick. Generally the system is described as: "Buy at Friday's close if the SP500 is down > 2.0% during the previous Monday-Thursday, and the SP500 is up on Friday. Sell at Monday's close."

ULTRA's BWB allows you to test this phenomenon and vary the parameters to whatever you desire. Historically, BWB as defined above has performed fantastically in recent history (Prior to 1978 the phenomenon was not significantly profitable).

BWB Jan78 - Nov02 Historical Results

Trades: 3.1 per year
Winners: 62%
Max Drawdown: -4.9%
CAR While Invested: +102%

BWB successfully identifies days when the SP500 historically appreciates at an annual rate of over 100%.

The VIXBP Seasonal System

The VIX indicator is a measure of the expected volatility of the SP100 index based on the pricing of OEX options. A VIX reading of 30.0 reflects that put/call pricing suggests the expected volatility of the SP100 to be around +-30% over the next year. High levels of perceived future volatility as measured by the VIX have historically coincided with market bottoms.

This phenomenon can be demonstrated by the simple system:
Buy the SP500 on the close on the NEXT-DAY after VIX > 48
Sell the SP500 on the close on the NEXT-DAY after VIX < 32

A VIX of over 48 is rare but has been extremely bullish since 1986 which is as far back as the VIX data series exists:

Trades: 7 (0.42/yr)
Winners: 100%
Max Drawdown: -13.3%
Percent Invested: 3.9%
CAR While Invested: +181.8%

Capturing the 3.9% of the time the SP500 appreciates at at a +178.2% annual rate is extremely valuable since most systems are not good at picking bottoms. As the SP500 rises, this system will sell out quickly as the VIX drops below 32. There is also risk with this (or any bottom picking) strategy since it is possible that the SP500 could just continue to drop even though the VIX is greater than 48. Another variation of capturing this bullish phenomenon that has a much higher sample of trades is the following system:

Buy if: (VIX > BGT2 and VIX > VIX_MA )
Sell if: (VIX < VIXMA) or (VIX < SLT2)
Where: VIX_MA = 10-day Moving Average of VIX

Historical Results trading the SP500 on a NEXT-DAY basis:

BGT2=SLT2 28 29 30 31 32
Trades per year 6.3 5.3 4.7 4.5 3.5
Winners 73% 72% 72% 75% 74%
Max Drawdown -19% -19% -19% -19% -14%
% Invested 8.6% 7.3% 6.4% 5.5% 4.5%
CAR While Invested 107% 114% 138% 183% 225%

These are truly spectacular results.

In the > 32 case you are capturing the 4.5% of the time when the SP500 appreciates at a +225% annual rate on a NEXT-DAY basis with a very simple system.

However, this system has the same potential problem as the as the VIX > 48 variation. That is, even though VIX is greater than 28 and greater than its moving average resulting in a buy signal, the VIX could just keep moving up and the market down. One way to solve this would be a simple percentage SP500 stop. For example, The "Stop Sell" would sell if the SP500 dropped > 4% below the SP500 value on the day the buy signal was generated. After a "Stop Sell" you could not buy again until a day occurred when VIX < VIX_MA. Otherwise, the system would just buy back on the next day after the "Stop Sell" in almost all cases.

Buy if: B1: (VIX > BGT2 and VIX > VIX_MA )
After a S2, VIX must rise above VIX_MA for a buy signal to become eligible to occur.

Sell if: S1: (VIX < VIXMA) or (VIX < SLT2)
S2: 4% Stop Sell based on the SP500 on the day the buy signal B1 occurred.

BGT2=SLT2 28 32
Trades per year 6.3 3.5
Winners 68% 69%
Max Drawdown -11% -7%
% Invested 6.8% 3.4%
CAR While Invested 113% 244%

Notes:

  1. At values below 28 the VIX seems to lose its predictive value although we are working with some systems that work with lower VIX values that may prove valuable.
  2. Other stop loss values work well also. The 4% value is an example and the default value we have settled on.
  3. We are working on other stop loss strategies that would make this system purely based on the VIX instead of having an SP500 price component.
  4. Although VIXBP is pretty amazing as is, it still remains a "work in progress".

ULTRA Financial Systems Inc.
P.O. Box 3938, Breckenridge CO 80424
Phone: 970-453-4956 Fax: 970-453-2467

© 2004 ULTRA Financial Systems, Inc.