Gap Fade Stats for January

Feb 1, 2008: 6:24 AM CST

As per reader request, I conducted a quick summary of all gaps in the DIA (Dow Jones ETF) for the month of January to see what would have happened if a trader faded all gaps.

The question is the following:

How many days in January had overnight gaps in the DIA (Dow Jones index), and second how many (and what percent) of those gaps filled.

I also added a very quick, very unscientific calculation of returns had a trader attempted to fade a gap.

Let’s start with the hard facts:

13 of the 20 trading days in January experienced some sort of gap greater than 10 Dow Points ($0.10 DIA points); thus 65% of trading days in January experienced an overnight market gap.

Of these 13 overnight gaps, 9 of the 13 experienced a complete and total gap fill, representing 70% of all gaps filled.

Let’s view this in a quick table:


The following days resulted in some sort of overnight gap greater than 10 Dow Points in January, 2008:


Legend:

BLACK: Not a Trading Day
GREEN: Successful Gap Fade
RED: Failed Gap Fade
WHITE: No Gap Fade

Now, onto the subjective material:

I took each gap fade day and tried to assess visually the potential profit or loss for each day. For successful gaps, I took the intraday low and counted price to yesterday’s close only.

For failed gap fade trades, I took the maximum stop-loss size of $0.50 (50 Dow Points) for an arbitrary stop-loss, assuming that entry took place at or near the gap open. The losing gaps were all at or near 100 Dow points (or $1.00 DIA).

Here are the non-scientific statistics based on my analysis only: (no representation that these numbers are perfect or near perfect, but only represent possibilities designed to give you an idea of a gap-fade strategy that does not include commissions):

January

gaptable.png
Total Wins: $14.40

Total Losses: $2.00

Total Outcome: $12.20

Recall that $1.00 for the DIA (Diamonds ETF) is equivalent to 100 points on the Dow Jones Index.

SUMMARY:

Of the 20 trading days in January, 13 (65%) of those days produced gaps in the DIA, and of those 9 (70%) of those gaps filled for a profit using the classic “fade the gap” strategy.

Based on crude and unscientific statistics, traders using this strategy had the potentiall to gain $12.20 from all trades (13) taken!

Thoughts:

How would volume confirm? What about other indicators? What about morning news report? What about size of the initial gap? How would you place stops?

All of these questions could help us filter out gap trades, but would complicate the simplicity of the classic “fade the gap” strategy.

While nothing is ever 100% in trading, the classic “fade the gap” strategy produced an edge both in percentage and potential dollar profit in the month of January, 2008.

7 Comments

7 Responses to “Gap Fade Stats for January”

  1. JP Says:

    great analysis ! thanks.
    I love your blog.

  2. Wolverine Says:

    Really great info!
    One question why do the loss days include the full range of the gap? If you’re entering at the open, your loss would be limited to where you set your stop, half the gap range, or .25 cents, as you wrote. Making an even better profit for the month. Not that I would complain about even 9.40 in profits over 13 trades…

  3. Corey Rosenbloom Says:

    Wolverine,

    Excellent question! I will re-do the numbers and re-annotate the post to reflect this change. That was a mistake on my part. I have posted the correction. Note that an arbitrary 50 cent stop-loss per failed gap would have only removed $2.00 from the overall profit, leaving the net balance at $12.20. For 100 shares, that’s $1,220 or for a standard day-trading “plunge” position, that’s $12,200!

  4. jkw Says:

    Don’t forget that your stops will be hit on some days when the gap is filled. You need to look at the intraday chart to determine how close your stop could have been without having it trigger. It is a somewhat complicated optimization problem to determine where to place stops and limits. I wrote a script to do simulated trading of gap fades over my historical data (about 18 months) on ES. It calculates the optimal values for stops, limits, and minimum gap size over a trailing time period and then uses those on the next day. I haven’t put in any allotment for slippage or commissions yet. The results are around 3-7% a month depending on how long the lookback optimization period is. Interestingly, the results get bad (even negative) once you optimize over a longer time period than the previous 3 months. I’m not quite sure how good the results will be once I put in slippage and commissions losses. I also don’t know if some indices are better for gap fades than others.

  5. Javier Says:

    Hey guys .. Do you know of any website that provide services on stock with Gaps??

    Thanks

  6. Gap Fade Stats for September « David's Blog Says:

    […] January Gap Fade Statistics February Gap Fade Statistics March Gap Fade Statistics April Gap Fade Statistics May Gap Fade Statistics June Gap Fade Statistics July Gap Fade Statistics August Gap Fade Statistics […]

  7. http://centroenea.it Says:

    Hi, always i used to check webpage posts here in the early hours in the break of
    day, for the reason that i like to learn more and more.