An article to read today concerning the debate I’ve been having with the quantitative analysts.
First, let’s set the scene.
The quants come from a mathematical and institutional background. They don’t do technical indicators like flags. They are purely concerned with the numbers, and ensuring that the numbers conform to institutionally approved standards of statistical reliability.
This means that when they test a system, they measure its performance and also the ‘numbers behind the numbers’. They seek to analyze how reliable the performance numbers are likely to be in the future over a similar time period – in other words ‘robustness’. This is what is called an empirical approach, one that is focused purely on the numbers over literally millions of iterations.
When it comes to the OVI the numbers work. Whichever we we slice our approach, empirically the numbers are remarkably consistent from the OVIsi (S&P Index timing tracker) to the OVIcopilot stock picking funds. New studies suggest a similar level of consistency for the sectors as well.
Of course that does not mean perfection – nothing can achieve that – but for our preferred time horizons (end-of-day trading over a period of weeks) the numbers are remarkable over this past 11-year period, using institutional methods of analysis.
So, what’s the issue, you might ask.
Well there is no issue with the performance and robustness. Institutions will go for this – subject to a few extra bells and whistles concerning capacity, portfolio weightings, etc.
The issue is one concerning aesthetics.
As a retail trader you and I have taken great comfort – and derived much success – from a classic OVI-Flag setup as demonstrated by stocks such as AAPL, GS, BAC, GOOG, FB etc time and time again.
We look for persistently positive wiggling OVI together with an imminent price breakout possibility. Countless times we’ve been able to reap windfall gains from such scenarios. They’re pleasing on the eye and we can mitigate our risk easily through use of support and resistance levels defined by the flags.
The ‘problem’ is that empirical studies have unearthed many high performing stocks that don’t look very pretty, and would be unrecognisable from our desirable OVI-Flag setups.
Take this chart of TDC. TDC is empirically one of the very best performing stocks in relation to its OVI. We’re talking extraordinary hit rates over all permuatations and combinations. The ‘issue’ is that the OVI looks unreadable to the naked eye.
Admittedly, this is just a six-month snapshot of the stock, but looking back at its history TDC contains great swathes of neutral horizontal OVI behaviour which is no good to us from a naked-eye perspective.
And yet when its OVI does pop up for a few days or couple of weeks, empirically price has performed. You may not see it with your naked eye that easily, but empirically it simply is the case. And any statistical purist will vote for empirical performance over aesthetic conformity.
In a way this is a good thing. Empirical study trumps prettiness. But for the typical retail trader (and for my own accounts I am one too) prettiness and familiarity is also important.
Hence, in the OVIcopilot we are releasing (on Wednesday I’m told!) the new copilot for OVI-Flag combinations. This model works very well, with figures comparable but not quite as good as the pure OVIcopilot model.
A purist would say “Why bother with the pretty stuff if the pure OVI model is statistically better?” And they’d be right … but that would be to ignore the human element which appreciates familiarity of terrain.
The terrain here is the recognisable chart pattern and persistently positive wiggly OVI setup. The new OVI-Flag copilot model addresses the chart pattern, which means it’s ready for deployment (on Wednesday I’m told – again!). The instantly recognisable persistently positive wiggly OVI setup is not wholly addressed. The empirical purist says it doesn’t need to be – and they’re right! The numbers work.
But the human in us wants to see with our own eyes that the algorithms are picking up the entire pattern combination. This makes us feel secure. And because we’ve all been trained how to spot the opportunities and how to manually trade them (through the Guided Discretionary method) it means we feel more in control about the trade if it conforms to the instantly recognisable pattern combination.
It’s important to understand … the algorithms are finding pure OVI setups for the pure OVI model, and OVI-Flag combinations for the OVI-Flag model. Empirically we have absolute confidence in what the algos are finding. They’re not always that pretty or obvious to the naked eye, but they are there.
So, what do we do. We don’t have any issues … and yet we we want more!
I’m a hybrid between a purist numbers guy and a retail trader who likes to instantly see order in what I’m trading. This is how I’ve made my money in trading and it makes me feel warm and fuzzy to see stocks conform.
At the same time I’m so proud that what we observed and traded for many years is backed up by empirical studies consisting of millions of trade iterations. The numbers are the numbers, and those numbers fully support the original hypothesis that options traders do have an edge, and that the OVI picks up on this. The numbers also reinforce that combining this phenomenon with a price breakout is a statistically excellent strategy that substantially outperforms the markets with lower risk consistently.
The empirical numbers are beyond question, which means we can now address the aesthetics. It could not be done the other way around.
So, behind the scenes we’re working on bridging the gap between the purist empirical view and the retail trader view, provided we do not jeopardise performance or robustness. Remember, robustness is about a model’s likely resilience to changing conditions, and can be tested using institutionally accepted statistical protocols.
The initial signs are positive as we take a cautious approach, and I’ll be discussing this over the next few weeks as we examine our findings.
The outcome I’m hoping for is that balance between empirical worthiness and aesthetic prettiness, combining to produce performance and robustness.
In the meantime the market has rebounded strongly from the reversal patterns from two weeks ago. The OVI Sentiment Indicator is very close to changing colour back to green, but do remember it is calibrated as a longer term indicator – it won’t catch every swing because it’s not designed to, but it does have a remarkable record against the S&P.
All the best