Predict Something (Quantifiably)!
I don’t want to go into Christmas on a sour note, so HAPPY HOLIDAYS folks! I’m eternally indebted to you all for your thoughts and time. Having said that, random rant for the day…
I know I don’t make many friends whenever I get on this soap box, but it irks me that financial bloggers don’t make more predictions that can be easily quantified and tracked.
The quant-minded blogosphere (myself included) is very good at churning out gazillions of bits of market data or showing sexy backtests of what worked (past tense).
That’s super duper. That’s the noise we all sift through to find our bits of trading gold and I love it, but it makes it difficult to separate the wheat from the chaff.
As I’ve ranted before, the only way to judge a pundit’s usefulness is based on decisions they made in real-time using the information available at that moment.
And the only want to capture those decisions is if pundits are making frequent, easily quantifiable predictions and then tracking them.
I’m not talking about vague hedged calls that’ll be construed as a win no matter what, and I’m not talking about cherry-picking your winners.
I’m talking about making clear, concise calls that a dollar figure could easily be applied to and then reporting on all of them publicly and in an empirical manner.
In other words, I’m talking about putting your money (or foot) where your mouth is.
You can say a lot of things about me, but one thing you definitely can’t say is that I’m afraid to put my foot in my mouth.
Readers can follow our (independently-audited) proprietary strategies, or our monthly seasonality map, or our new tactical asset allocation model in real-time.
Those are frequent, easily quantifiable predictions that represent what this pundit believes to be so. Generally I’m successful. Occasionally I’m horrifically wrong. But the important part is that you know.
Let me repeat that, YOU KNOW.
Good or bad. Guru or imbecile. Love me or hate me. I’m okay with all of them. But at least your decision will be based on some semblance of fact.
How many other pundits can we say that about? That we understand clearly how they would have traded and how all of those trades would have worked out?
The point of this post isn’t to pat myself on the back. The point is to hopefully motivate all of us to expect a little more accountability from our experts (even if that sometimes necessitates more feet in more mouths).
Rant over…and HAPPY HOLIDAYS!
Happy Trading,
ms
P.S. I would be remiss not to say that some in the quant-minded blogosphere do a very good job at making quantifiable predictions and tracking them empirically. I’m not naming names (because I don’t want to shine the light on folks who aren’t doing such a good job) but you’ll find more than a few of them on my blog roll.
. . . . .
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: Random Stuff | 9 Comments
My New Years resolution, as a result of this post, is to model the S&P 500 index and start a TimerTrac record for Rydex S&P 500 2x leveraged mutual fund. The results will be on a dedicated page from inception. I already model the SP futures so it won’t be too time consuming to model the index.
Merry Christmas & Happy New Year,
James
Good on you sir. I look fwd to seeing that! michael
Hey James,
Great idea, I hope it gets your systems more publicity — I suspect it will.
Merry Christmas to you and Michael, and everyone else out there.
=Carlos=
Hello Michaël,
As you tell us about predictions and random, I thought you could be amazed by looking at this small simulator.
It shows the equity curve when tossing a coin each workable day, 10 years long (2 500 days) and winning 1 € if showing “heads” while loosing 1 € if showing “tails” ,
http://www.stockengineering.com/tirage-pile-face.php
If you understand French, you might be interested in reading one of my last weekly newsletters :
http://www.stockengineering.com/fr/news_display.html?id=370
I am there asking the question : if random looks so strong like stock charts, why wouldn’t stocks charts reflect mostly random ?
Congratulations for your blog and renewed thanks for sharing so much with us.
Pierre
Stock prices are indeed *almost* completely random. I’ve played around with an excellent ARMA toolbox from a professor at my university, applying it to SP500. The result: 9 out of 10 times the best found model is just a random walk. This is in line with the Efficient Market theory. However, a little portion of the stock price is *not* random. If you look really well, there are quite a few edges to find. One of them is ‘overshoot’ (in engineering terms) that happens at the open. Another is a reaction of the market to the trading day in the last 1/2 hour of trading, when the end-of-day traders step in. Relative prices (as used in pairs trading) have even more edges.
Predictions for 2011: Soros will eclipse Zuckerberg through Mandelbrot, Toyota gets its due and The Xinhua News Agency becomes a recognized world power:
http://bit.ly/hf4cqR
Just a note to say that I have started a TimerTrac strategy and the first trade for the year has been recorded. When I have a few, I will embed the TimeTrac graph on a web page. Wishing everyone a happy and prosperous year!
Bravo James – I look fwd to seeing that. michael