A New Breed of Oscillator
In the previous article we developed an oscillator that wasn’t prone to stickiness in the overbought and oversold regions and doesn’t have the lag that oscillators have which are built on moving averages. Next up, we’re going to study it’s usage and usefulness.
Tekst: Dirk Vandycke
The Monest Value Indicator (MVI) is a short term oscillator trying to capture overbought and oversold setups. As a consequence of how it’s built, by a statistical normalization procedure, it’s unlike most other well known oscillators. For one thing, it doesn’t have the stickiness that keep other range bound oscillators in overbought and oversold zones while a strong trend develops. But also the lag of moving average based oscillators isn’t an issue with the MVI. What’s more, having no parameters in its equation, it is totally objective by definition. This still leaves us with a necessity for an objective interpretation. Let’s find out if our indicator is of any real and statistical significant usefulness.
Value consensus
In the previous article we introduced the Monest Value Indicator (MVI) and how it is constructed. Due to its normalizing nature it can be used in the same way for any time series.
Any value between -4 and +4 indicates a strong short term consensus regarding current pricing and fair value. Between 4 and 8 we have slight overvaluation and anything higher than 8 indicates plain overpricing. It’s very rare to see values much higher than 8 with this indicator, even though this not a range bound oscillator. Given that value for the MVI is rarely much higher than 9, this would indicate major overvaluation with respect to short term consensus.
Likewise we define minor undervaluation in regard to short term consensus between -4 and -8 and plain undervaluation below -8. Anything lower than -9 would be considered major undervaluation.
These levels are not written in stone, but nevertheless they are deducted from the standard deviation intervals we have in the statistical normalization process.
For the purpose of finding the advantages of the MVI, we’ll consider the violations of 8 and -8 of primary interest to us. In other words, we’ll look into what happens if we focus on just those overvaluation and undervaluation signals. Anything still further from zero would give us far to less signals to obtain a meaningful sample space.
Right, right away
Do you remember a trade where at first you have this small adverse movement against you, before the trade finally takes off? I think every trader knows what I mean. In fact, when you open a position, costs are made, putting the position in the red from the very beginning. If you can picture that, than the trade where you get stopped out before it materializes isn’t hard to imagine either, isn’t it? Wouldn’t it just be cool if we could do something about this?
Look at figure 1. It’s an example of how a setup accompanied by a short term overvaluation (MVI>4) can be postponed until the MVI drops below -8. What we did is, of course after the facts, see how the trade would have turned out if it would have been entered on the highest MVI value of the week (5.53) in comparison to what would have happened if we bought at the lowest MVI value that week (-8.43).The effect is a far lesser adverse excursion of the trade after its initialization. What’s more, the postponed entry can make the difference in getting stopped out and loosing the setup out of sight, and having a very successful trade, riding the subsequent trend in the case of the postponed entry. This is just an example of course. But It’s a start to see if the idea of just taking the undervalued signals of a system has merit. At least, the idea is sound, because if you only take setups that are undervalued, you might have a far better profit/loss ratio on the trade. And we just might have better entries leaving us with less initial adverse recursion before trade takes off. Which, in turn, might even lead to less losers because less setups are taken. We only take those setups where an undervaluation is present or turns up soon enough after the setup signal. And because of the less countermovement, fewer trades might get stopped out.
So, what if we opened long positions only on moments of short term undervaluation and, likewise, short positions only when short term consensus of value is overrated. MVI to the rescue. To prove the possible added value of our indicator here, we pull out our Monte Carlo simulator and look at a whole lot of random entries where the MVI went below -8 within the next week (remember our MVI is a short term value indicator). One could point out the fact we only take the random entries where the MVI goes below -8 within the next week, as a problem. What with all the random picks where the MVI doesn’t go below -8 in the five days that follow. Well we don’t have to take those setups. At this point we don’t want to prove that we have a better system with the MVI constraint added. We’ll get into that next. For the moment, we just want to see what happens with our initial adverse recursion of trades when we wait for undervaluation, or, if that doesn’t happen, don’t take the trades at all.
Figure 2 displays an average trade (of thousands of random entries) in a few months of a strong bull market where an undervaluation (MVI<-8) followed within the next week. In the same figure one can see the difference if we just wait for the MVI to go below -8 before taking an entry. The difference is clear. The random entry has an initial drawdown of up to 10%. On the position, that is. Remember that, through position sizing, one might invest 20% of one’s portfolio while taking only 1% risk on that portfolio. Which, in turn, might stand for a 10% loss on the position (but only 1% on the portfolio). Nevertheless, there’s hardly any drawdown left when the random entries are postponed until the MVI goes below -8 (or the setup gets canceled if it doesn’t go below -8 in the next week). This simulation didn’t take costs into account. But the point for using our MVI is clearly made. I won’t go into a similar simulation for a bear market, using short positions. Results were indicative of a similar effect taking place.
Of course this is based on what we see for an average trade. It doesn’t say anything about what could happen with any specific trade taken.
Let’s dig deeper.
With the promising results from our simple preliminary test, I wanted to dig deeper into the possible benefits of the Monest Value Indicator (MVI). So I came up with the following test setup.
Let’s see what our average trade would look like under different entry strategies. One of which will be our undervalued filter based on an MVI value below -7. In contrast, as the opposite entry extreme, we have an overvalued entry strategy, buying only when the MVI is greater than 7. Our third entry strategy is a dollar cost averaging one, buying every first day of each month. Of course for the purpose of comparison, we still have our average random entry trade and also our random entry trade only made when the MVI is below -7. Hence we compare 5 strategies to a mere random entry strategy. Again the test is run in a pure bull market.
All six strategies were compared on a 50 day basis after each entry. The result of which can be seen in figure 3.
The entry systems had, on average, the following MVI value on entry:
Random Entry: -0.01
Undervalued: -7.70
Overvalued: 7.53
Dollar Cost Averaging: -0.84
Undervalued Random Entry -8.09
The period tested being a bull market, still doesn’t seem to help our overvalued entry system, which is barely able to make money with several large periods in the red zone. The dollar cost averaging entry system performs a lot better, even better than the random entries system. Exactly what could be expected, because dollar cost averaging benefits from buying more shares at lower prices and less shares at higher prices. The best performing system is, as we hoped, the entry system buying only on undervaluation. It performs best over almost the entire period.
Combining undervaluation with random entry (taking only the random entries when the MVI is below -7), does seem to improve the random entry system. However the equity curve becomes more volatile. Which could be the reason it losses from merely random entry by a big performance drop at the end.
The overall conclusion is that buying on undervaluation, as indicated by the Monest Value Indicator, results in a better performance and far less time of the trade being spent in the red.
Conclusion
If time is money, I would add that ‘timing is money’. There’s evidence supporting the idea that our Monest Value Indicator (MVI for short) has true added value not only in measuring short term valuation consensus, but also might be a valuable system add-on to lower the frequency of losers by lowering the frequency of trades, dumping the ones with a poor profit/loss ratio. It seems as if the initial drawdown on trades can be kept smaller, meaning fewer positions are getting stopped out because of initial adverse recursion. In our final article on this new indicator we’ll go into its effect as a possible trade system enhancer (taking only the undervalued setups).
As a reminder. This indicator is available at www.chartmill.com, both on the charts as well as in its screener. So it’s possible to add screening constraints on the MVI value (greater of smaller than a certain level), giving, for instance, only undervalued equities and ETF’s in your custom screen based on other criteria.
Figure 1: An overvalued (or at least not undervalued) versus an undervalued buy.
What happens when we buy a setup not on the signal but on the first undervaluation (MVI<-8) after that. In this chart two entries are compared, coinciding with the highest and lowest MVI value for that week.
Figure 2: Profit chart of an average random buy versus a postponed random buy.
Comparison of an average random buy (during a bull market) and what happens if the buy is postponed until a MVI undervaluation (<-8) is recorded or, if that doesn’t happen within a week, the setup is aborted.
Figure 3: Different entry strategies compared to an undervalued MVI entry system.
Buying while short term undervalued (MVI<-7) seems to pay off against other entry strategies. All systems are compared to a random entry system
This article was originally published in the December 2011 issue of Traders' Magazine. You can download the pdf here