Intraday V-Bottoms and the PSE Index

This is another of Gary Smith’s trading insights: the “V-Bottom Reversal” which is a pattern said to be predictive of short-term trend change. As quoted from Smith’s book:

V-bottom reversals occur intraday where the Dow has been in decidedly negative territory the entire trading session, down at least .75 percent, and then makes a furious comeback, closing either near the unchanged level, or preferably, up for the day. The later in the day the reversal occurs, the more significance I attach to it. I will instinctively trade V-bottom reversals if the previous day closed down or if the Dow has been in a recent downtrend.

Fairly straightforward. This is a study to test the applicability of the V-bottom concept on trading the PSEi. To test the idea, we have to reduce the pattern into objective criteria:

1. Previous day closed down.
2. The low of the day is at least .75% lower than yesterday’s close.
3. The day closes unchanged, defined as closing within .2% (up or down) of yesterday’s close.

These rules now allow us to find V-bottoms that have happened in market history:

On June 8, 2007, the market had traded lower from yesterday’s close of 3,528, with an intraday low of 3,468, but the market closed unchanged at 3,526. The next trading day (after the weekend break), the market 40 points higher at 3,571.

On December 7, 2006, the market traded lower from yesterday’s close of 2,797, with an intraday low of 2,774, but managed to close unchanged at 2,796. The next day, the market closed 30 points higher at 2,831.

On October 18, 2006, the market traded lower from yesterday’s close of 2,543, with an intraday low of 2,523, but managed to close unchanged at 2,546. The next day the market gain almost 50 points closing at 2,597.

Outside of well chosen examples, we can run the rules defining our pattern throughout our data and check the market results the day after the pattern:

From a total of 2,723 trading days, the V-bottom reversal as defined by our rules occurs rather rarely: for a total of 21 trading days. Cumulatively these 21 days have resulted in a gain of 107 points the next trading day, compared to a cumulative loss for all other days combined. The higher winning percentage contributes to the positive expectancy of the pattern compared to the breakeven expectancy of all other days, which confirms the effectiveness of the pattern in predicting positive market action.

Despite its positive performance, the rarity of the pattern might be disturbing to traders–since the waiting time in between trades may be too long to be practical. To address this, I tested the results by waiving the first rule–that the previous day must be a down close. I just checked all days that qualified as a V-bottom, regardless of the previous day’s close. Here’s the summary:

By ignoring the first rule, we have more trading days in our sample: a total of 60 days, which cumulatively earned 111 points compared to a cumulative loss of 403 points for all other days combined. However, if we compare the winning percentage and expectancy of these results to the original definition, we find that waiving the first rule results in inferior winning percentages and expectancy. This means that while you may be trading more often, you are actually going for less favorable odds by waiving the first rule.

Insight And Application

This is another pattern that reveals the micro structure of the market. The idea behind a V-bottom is a change in market sentiment from decidedly negative (at the low) to an unchanged close, which is neutral or even positive. The news or events that trigger these reversals may vary, but the key insight is that the market participants express the change in sentiment collectively through their actions by buying up a market that had been declining previously.

Whenever such dramatic change in sentiment occurs, is usually a signal of a short-term change in trend which can be exploited by the astute trader.