Exploring the DVI Indicator: Extreme Readings

29Jul10

In this series we’re analyzing CSS Analytics’ DVI Indicator (read part one).

The DVI is an intermediate-term indicator that is contrarian (i.e. buying weakness) and usually best traded long-only. Click to calculate.

In our first post I looked at how the market has responded when the indicator stood below its midpoint (50%). In this post I’ll look at other more extreme readings.

The graphs above show S&P 500 index results since 1970 the day following DVI readings of 0 to 10%, 10 to 20%, etc. (close-to-close).

I’ve shown results in terms of (1) annualized return, (2) annualized volatility-adjusted return (ann. return / ann. standard deviation), and (3) win %. The grey dotted line represents the average result for all days.

Geek note: see end of post for assumptions about dividends and trade frictions.

What does the data say about the DVI?

Let’s first talk about (bullish) readings below 50%.

The first graph (annualized return) appears to show that the deeper DVI travels towards zero, the more bullish it’s become, but that’s a little bit misleading.

As the DVI has traveled closer to zero, returns increased but they’ve also become much more volatile, which is why vol-adjusted returns actually begin to decrease.

This makes sense. Readings of 0-10% only come up on the most oversold of days when the market is prone to all sorts of BIG things (usually good, but often bad).

In short, DVI readings below the midpoint (50%) have been pretty evenly bullish regardless of where they’ve fallen. In this respect, DVI is very unlike short-term OB/OS indicators such as RSI(2) or DV(2) which are clearly more bullish/bearish the deeper they travel towards 0/100%.

What about (bearish) readings above 50%?

Above 50%, returns drop off quickly. From 50% to 90% next-day returns are all negative. Not enough (absent other confirming indicators) to justify shorting I think, just stepping out of the market.

Above (roughly) 90%, something interesting happens; returns again turn positive.

These are the times when the market gets legs and just keeps on pushing upwards through overbought resistance. This observation has been volatile but reasonably consistent since 1970.

Putting It All Together

• Since 1970, the DVI has been similarly bullish at any level below 50%.
• The DVI has been bearish from 50% to roughly 90%, but not enough to justify taking a short position.
• At very high readings above 90% the market has again turned bullish, but not to the same degree as readings below 50%.

It should go without saying that there is a bullish tilt to these results because of the bullish tilt to the market over the test period, so in a follow up post we’ll also look at long/short results when the market is up/downtrending.

[Edit: click for a summary of all posts in this series on the DVI indicator]

Happy Trading,
ms

Test assumptions: (a) I’ve calculated the DVI using the S&P 500 cash index (which traders generally use), but calculated returns based on daily dividend-adjusted data (dividends interpolated from quarterly data), and (b) results are frictionless (i.e. do not account for transaction costs/slippage) but could be closely reproduced in today’s market using certain mutual funds less part of an annual expense ratio.

. . . . .

To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.

Filed under: Trading Strategies   |  3 Comments