No registration required! (Why?)

Fading the Opening Gap

I've updated the "league tables" for the raw gap play for month on month. This will give you an idea of which months have been better months historically to fade the opening gap using the "raw gap play" technique. These can be found here:

Briefly, the "raw gap play" assumes that your entry is the opening price and if the previous settlement price is touched you are filled at that price. The "raw gap play" does not use stops and as such is not recommended as a technique but as a benchmark to compare your gap fade results against.

If you notice that a month is missing and the tables haven't been updated then post a response here and I'll get on it.
"League table" has just been updated. Click link in previous posting. Gap fade strategy made 31.00 points in the ES for June 2005.
Email query and response that I thought that everyone might be interested in:


You asked, "I haven't combined BB's with the gap study but it is certainly something that could and should be looked at. For contracts that trade continuously yet still have a gap effect from RTH hours, are you looking at the BB over the continuous data or over the RTH data?"

Good question - I'm so used to looking at commodities (and their almost daily gaps which most often reach the BB) that I think that's what the whole world looks at! I've been looking at commodity gaps on 30-min charts which make it to the to the BB - followed by reversal price action for entry.

One can certainly argue that the PA would happen regardless of the BB being there... and so they may as well be ignored. I haven't really decided one way or the other though.

The continuous indices (and their relatively rare gaps) don't seem to have any Bollinger Band edge imo.

One of the things that I want to look into (and that needs to be looked into) is the concept of why there is a gap and the period between market close and market reopen. During this period there is a virtual market in the heads (minds) of the participants that is continuous and this market runs continuously while these people think and receive information outside of market hours. The fact that it isn't recorded anywhere only becomes apparent when the market opens and gaps.

With a continuously traded contract the gaps can only show up during the settlement periods (usually about 15 minutes at the end of each day) and the weekend.

With continuously traded contracts we can synthetically create gaps by overlaying the traded time period of the continuous contract with the pit traded OR liquid trading hours - usually RTH.

In my opinion the Bollinger Bands should run across the continuous market - even if no data exists for this market (this is obviously hypothetical) as the Bollinger Bands are designed to adjust to volatility. As such they should be looking at the data in between the close to open periods to calculate their values because there was obviously movement and volatility during this period - even though it wasn't recorded.

So if we find a market that reacts well to certain Bollinger Band levels then we might be able to use the touched or piercing of a Bollinger Band as an extra confirmation with a gap.

To put this in practical terms here is an example:
Trading the ES we look at the gap between 16:15 in the evening and 09:30 the following morning (EST). The Bollinger Bands we run across the continuous contract.
The market opens gap up by 4 points but in the middle of the Bollinger Bands. Opening activity pushes the market up to (say) 6 points gap up and this pierces the Bollinger Band. Perhaps you know that a Bollinger Band touch in this market gives you a 60% chance of a 6 point correction using the same size stop. You may also know that gap that has increased in size by 50% after the market has an 80% chance of filling.
This might be a higher probability trade than the two single strategies by themselves. I don't know because I haven't tested it. These are just my mental ramblings.

I hope that somebody gets something out of this.
Very interesting that the best months for the gap strategy are the "summer doldrums" that everyone bemoans. Very interesting indeed.

I also recently saw a study which showed that if you bought SPY at the close and sold at the open from 1993 to 2005, you would have made 143 SPY points. Buying at the open and selling at the close? One would have lost 67 SPY points. There's money in them thar gaps!

Thanks to Guy for helping make them more understandable.
You're very welcome Rob.

My theory on the summer out-performance in gaps is because summer months tend to trend less and rotate more. So a gap is more likely to trade back to the previous session's close during the summer months and for the rest of the year we're more likely to see "gap and go" days.

To confirm this theory I think that we would need to define a rotational day and a breakout/trend day and then compare the numbers of those days in different months of the year in order to determine if the summer months see a higher number of rotational days etc.

One of the rotational versus trend "indicators" that I use is what I call the Trend %. Although I haven't seen this written up anywhere else, the concept is so simple that I would not be surprised if it exists under a different name.

The Trend % simply measures the absolute difference between the open and close against the day's range. This measure by itself would be an interesting one to measure and could easily be done with daily OHLC data and a spreadsheet.

If I find some spare time I'll do it...
Quote from rhogaine:

There is actually a newsletter/advisory service that offers a "Gap Fade" strategy and so far it has been extremely succesful. Check them out delta. I am curious to see hear what you think. Take care.


Thanks for the link rhogaine!

We know that the Gap Fade strategy works from all of the Gap Fade backtesting that I did in the ES. One of the major problems that I came across during the back test was that the best solution was to not use stops. No matter how I used stops, unless they were ridiculously large, they reduced profits overall.

The web site that you quoted (Crowder Investments) states win rates of 76 and 78% which are consistent with my back tests in Q1 - What is the probability of a gap fill (total) and which day of the week is a gap fill most likely?

They also state "The strategy typically trades 2 to 4 times a month..." which would be consistent with Q11 - Are the large gaps worth fading or are they just too risky? which isolates the more profitable gap trades which are the larger ones and eliminates the smaller ones which are likely to fill but have more risk exposure per profit unit if you aren't using stops.

Given this information I would guess that Crowder Investments probably have a solid strategy that should be further investigated if you trade this strategy - thanks for the heads up.
What about the time of year with the gap fade? That email that you sent out shows the summer months as being more profitable. Do you think that Crowder used the time of year in their back testing?
Originally posted by trade2win

What about the time of year with the gap fade? That email that you sent out shows the summer months as being more profitable. Do you think that Crowder used the time of year in their back testing?

No, it doesn't look like it. It looks like they ignored seasonality. Remember that this "Gap Fade over Summer" has worked better over the last few years but I haven't done extensive back testing on it so I don't know how it works on other instruments or how far back this applies to the E-mini S&P500.