We all know that the “simple” positons we can have with our stock market investments are:
1.) Long – Own equities (and bonds)
2.) Short – Own “inverse” equities (betting against the market)
3.) Cash -- Own only cash
All quantitative trading strategies have formulas and indicators that they use to determine if the market is going up or going down,
hence telling the developers whether they should be Long or Short. These formulas and indicators include:
- following the trend,
- calculating the strength of a market’s momentum,
- calculating the “rate of change” (slow or fast),
- using formulas like “Fibonacci”,
- tracking the numbers of stocks hitting new highs (and new lows),
- tracking the number of stocks that are above/below their 50-day, 100-day and 200-day moving averages,
- using the MACD formulas, and
- including literally hundreds more.
From the strength of the combination of the above (and many more), quantitative models tell investors whether to be Long or Short.
The objective is to make money in both up and down markets by being Long in up markets and Short in down markets. Good models
do just that most of the time, and their followers make money in both up and down markets.
Since we do not make money (other than a tiny amount from bank interest or money market funds) when our money is in cash,
Cash signals seem to be useless.!!!
So, where do Cash signals come from, and what does being in Cash do for us?
What Cash positions do for Investors:
Cash positions do not make money for us, BUT cash positions often prevent us from losing lots of money (as in "avoiding the crash")
Where do Cash signals come from:
To my knowledge, there are no algorithms, indicators, formulas, etc. that calculate and “issue” cash signals. In fact, many timing
models only issue Long and Short signals (no Cash). If model developers (like MIPS) want the safety of occasionally being in an
all-cash position, the developers themselves need to come up the ideas on how to determine when we should be in cash, and then
“program this” into their models.
How MIPS timing models were programed to issue Cash Signals:
Because the MIPS algorithms are proprietary, we do not explain exactly how we issue cash signals.
1.) Traditional “Stops”
MIPS goes to cash after it has lost a certain (not fixed) percentage amount from a “high point” determined
within the model; or from a certain maximum drawdown amount. Stops really mean “I don’t know why,
but you are not doing well so I am going to shut you down until further notice”. That is a Cash Signal.
2.) Too Close to Call Decision
MIPS literally uses hundreds of indicators and algorithms combined with mathematical equations to
determine the % that is Long and the % that is Short, To issue a new signal, the winning position must be
ahead by a significant amount, or the model will default into a Cash position (too close to call). For
example, if the resulting decision is 55% long and 45% short, the model may issue a Long Signal; but if
the result is 51% long and 49% short, it would almost surely issue a Cash Signal (to close to call Long or
3.) Volatility Too High
There are many different definitions of volatility (like the VIX, standard deviation, etc). The greatest
impacts on quantitative models are (a) how fast are the market changes and (b) how “large” are the changes
relative to other similar changes in the past. Recently, we experienced some of the biggest changes in the
history of the US stock markets. In March 2020, the SPY fell the fastest and lowest in one week than at
any other time since 1933 (almost 90 years ago). The daily changes in 2020 are shocking when compared to
the long-term performance of the market.
For example, it is reported that the US stock market (Dow or SP 500) has risen close to 7% per year (after
inflation) for over the last 80 years. That means that the annual percent change would have been between
9-10% per year with inflation.
To make my point, let’s use 10% gains per year. That amounts to 0.8% per month (or approx. 0.04% / day).
- In the first 5 days of March 2020, the SPY dropped over 11.5%, which is more than the 1 year average
change in the past.
- One month after 2/19/20, the SPY was down 29% (took almost 3 average years in the past to change that much).
- Between Mar 2 - Apr 15, there were 15 days with changes greater than 4%; and 4 days with greater than 9%.
In the past, where one might have lost an average of 1.5-2,5% in five bad days, one could have lost
over 25% in three bad days in 2020. With odds like that, the risk/reward ratio does not fit well for an
investor concerned about preservation of capital. In this environment, the potential losses (risk) are far
higher than possible gains (rewards). This high risk show lead to more time in cash.
4.) Market Moving too Fast
In a fast-moving market, like now, many commercial timing models are not programmed to handle fast moving
changes with such elevated daily changes. A good model would determine this and spend more time in cash.
Trading in this market environment is comparable to testing a new race car on a track that was built 20 years ago.
Let’s say that until now, there had been no “crashes’ from race cars making 180 degree turns on this track. Of course,
the embankments on these turns were designed to be quite a bit higher on the outside to prevent race cars from simply
flying off the track. Even if this track had been used successfully for the past 20 years by race cars going at 200 miles/hour;
it would not be surprising if a race car going 300 miles/hour went flying off the embankment. The reason, of course, is that
the track was not designed to handle cars going that fast. This is similar to the market we are in now. Numerous models
in use today were NOT designed for this extreme volatility. Additionally, many models are not prepared to internally make
that determination and go to CASH until after the market settles down, but MIPS is programmed to act long before then.
With respect to all of the above, MIPS has internal algorithms that address each issue. This helps investors attain good returns during
the trading day, and still sleep well at night.
Paul Distefano, PhD
Founder / Developer
MIPS Timing Systems