When the markets gap up or down, what you really want to know is whether or not to fade the move, whether the gap will close anytime soon. In recent papers titled "Fine Structure of Vol Feedback..." and "Endogenous Dynamics of Markets" Bouchaud and his co-authors show that a significant portion (the "lion's share," sometimes nearly all) of overnight market movement is endogenous; i.e. "determined by trading activity itself and not due to rational processing of exogenous news." He also posits that the endogenous movement can result in a short-term "decoupling between prices and fundamental values." In short, when endogenous effects dominate exogenous factors, you want to fade the gap. To determine if the endogenous effects dominate on a given overnight gap, use the methods in the paper "Nested Factor Model..." to find the common vol and directional factors among past overnight returns (go back a year at least) and any contemporaneous return series you think might be relevant; e.g. Asian stock indices and short/long rates, European indices and rates (cut off at 9:30 am NY time), FX rates (7:00 pm to 9:30 am), Asian/European session metal and energy prices (same cutoff), etc. Find the factors and use them to do a contemporaneous estimate (nowcast) of the overnight return. The residuals or error series are your estimate of the endogenous component. On mornings when the market has gapped and the residual is large, FADE THE GAP! You may have noticed that this method is conceptually similar to Bright Trading's "Opening Orders" method, only using nested factor decomposition instead of simple OLS regression and a wider and more indirect range of predictors. So if the complex multi-step factor decomposition method outlined in Bouchaud's paper proves too difficult to implement, just regress the overnight log returns against the predictor variables listed above, and use the residuals from that contemporaneous regression in the same manner.
Up or down, does not matter, as long as it moves, its easy to trade. The trick is not direction, the trick is picking stuff that will actually have good volatility for that particular session, whether is a gapper or something else, you just dont know. Key is knowing if something will move well, not where. Trend day up, trend day down, or V day, rest is much harder to handle and profit from. Show me something that will move for any particular session and chances I end with a nice green P/L are very high, but how do you pick them ? how do you know who will have a nice volatile day ? Thats the real secret of daytrading.
Kevin, Thats a very nice exposition linking research to practical trading applications. Most of the time, this linkage can be complicated to understand even when I read and understand a paper. Thank you very much. Much Appreciated!
It's based on a fair bit of time and effort trying to find a way to make money from trading gaps. Of course I may have missed some edge there with typical minor gap opens, but if so it was not for lack of researching. I don't risk any significant amounts of money based on personal opinion.
That first and last sentences are problematic. It is not enough for a pattern to have been somewhat predictive in the past in order for it to be tradable for profit - if only it were that easy. You must also have strong reasons to believe that the pattern will continue into the future, rather than getting competed away, or disappearing (or, worse, reversing) due to some structural shift in the markets, or never having been more than noise in the first place. You must also be able to use a risk control method to avoid suffering a blowup along the way - typically, risk control (hedging; or dynamically hedging via stop-loss orders or equivalent) costs money. And of course you must also be able to overcome transactions costs, your cost of capital, taxes etc. The other problem with fading most typical gaps is that you are pursuing moderate reward, and taking quite high risk if there is a continuation move (either that or setting close stops and being vulnerable to extensive whipsawing). I prefer to focus on bigger gaps. It is very rare for a large gap to just ranging and chop around all day - usually it is the result of dramatic news, emotions are high, many traders are operating for risk-control, fear, or greed, rather than pure profit-seeking. A move usually occurs fairly quickly and is often highly predictable and big. This is a much fatter, more reliable, and identifiable edge, which allows greater size, higher win rate and thus lower drawdowns. Gaps that occur after substantial moves, or when emotions and crowd sentiment are at extreme levels, offer further opportunities to enhance trade odds.
I disagree with what you wrote previously, but I would be interested to hear how moves after 'bigger gaps' are 'highly predictable'.
I did some analysis on gaps in ES. They did not work as well as predicted in 2013 ie small gaps close, big gaps run. In previous years it was more predictive. I abondoned this strategy.