# Determining trade trigger at higher frequencies

Let's say that I have an exponential moving average as a momentum indicator on a stock at the daily scale (EOD).

My trading strategy is to buy the stock on a sufficiently large close over the EMA.

I am interested in being able to preempt this indicator intra-day, if there is a sufficiently large breakout during that time.

For example, there may be a sequence of strong hourly closes, each higher than the last - this could give a reasonable estimate of the chance of triggering at the end of the day.

My question is whether there are any frameworks available to determine what constitutes a "strong" breakout at the hourly closes, say for a given probability.

Many thanks

• Thanks for your edit, I now immediately understood what you want to accomplish. – Bob Jansen Dec 6 at 7:59
• It is quite a tricky problem. Even though you may be able to say there is a high probability of the day ending above the threshold, it may be a bad decision to buy now, because the close (while still above the threshold) could be below the current price, an thus buying prematurely may cost you in the end... (It is a difficult tradeoff, when I try to do this judgementally I almost always get it wrong ;) Better to implement the trade exactly as backtested). – noob2 Dec 6 at 20:23
• If I had an estimate of the market microstructure vol and the underlying vol, what is a simple deflator I could apply to the intraday prices using these two quantities? – GlaceCelery Dec 7 at 3:46
• your question is unfortunately too general. A "framework" to determine what constitutes a "strong" signal is actually a full backtesting framework. You will presumably receive either very generic answer or not answer at all I am afraid – Ezy Dec 7 at 4:26
• Consider this as a concrete example: If we assume that the Price(oberved) = Price(real) + error (where error is a constant Gaussian with its own variance, uncorrelated to the price, it can be market structure noise)... If I want a price move in the 95th percentile in the latent price process....what is the size of the observed move I need to see given the noisy estimate... – GlaceCelery Dec 7 at 5:32