I’ve shared quite a few good trading strategies on this blog. Some of those have been a little complicated and I’ve always known that regardless of how good the strategy was, the average investor would never be able to apply them to their own trading because they just didn’t have the right tools. So…we have a new feature:
The State of the Market Report (also, see link on top navigation bar)
This report will be a snapshot of what some of the strategies we’ve discussed are predicting each day about the following day. I tried to pick a diverse group because I didn’t want to just look at the market through the market’s eyes. The report will also use the VIX, TED spread, Treasuries, SOXX, and time-based indicators.
Adding another layer of über-cool, I’ve programmed two additional concepts into the report: confidence and adaptation (checkout the report for additional information). This is a living document, so some of the future strategies we discuss on this blog will also be added to the report.
Lastly, remember, mechanical strategies are simply projecting future probabilities based on past results. Even if we knew those future probabilities with absolute certainty, we would still be wrong (often). As I wrote recently, we don’t need to be perfect to win this game – we just have to find enough quantifiable edges to be right a bit more often that we’re wrong. And that’s what this report is about.
I’m excited about this new feature…it should be fun to watch her grow. At some point this may become a subscription service for a small pittance just to justify the time it takes my person to put it together each day. But for now, it’s just another value-added service of being a MarketSci Blog reader.
I’ll leave you with a few recent reports just to give you a feel for how the report moves (click to zoom).
Prediction for Thursday, 11/13 (actual return: +6.9%)
Prediction for Friday, 11/14 (actual return: -4.2%)
Prediction for Monday, 11/17 (actual return: unknown, but based on the overnight market as of the time I’m writing this, more bearish than bullish)
Happy Trading,
ms
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: State of the Market | 7 Comments
Great concept. I think that this simplifies things a bit for the average guy reading the blog that currently is unable to develop these concepts.
A couple questions/comments….
1) I see 11 different indicators, of which some appear to be bullish & some appear to be bearish. How should an average reader interpret this? I could still see the average person not being able to draw a conclusion from the information presented.
2) Have you considered suing some sort of data mining tool with some sort of weighting algorithm on these indicators? I have heard of some great results being achieved with that technique.
Thanks,
Eric
Checking in from the airport — Very cool Michael. Keep up the excellent work.
RE to Eric: thanks for the kind words. RE: question #1 – I think this is the biggest question most folks are going to have…I’ll post some thoughts on the blog this week.
RE: question #2 – let me roll this one around in my head a while. The “confidence” concept we’re using would make a weighted total easier. The biggest problem I see is that some of the strategies (such as the two prop. OB/OS strategies) are using a very similar approach to the market (even if they use different data to get there) so I’d have to manually underweight each of those. Hmmm…more to follow.
michael
Michael – One of the things I do when weighting and/or adding additional criteria to a model is to re-evaluate the ‘model’ with the addition/changing of criteria to determine if the model is improved over the previous ‘model’ with less complexity.
I usually evaluate this by using a t-test. I am going to be using this with a system I am working on with equities, where the # of positions is a variable number and the total dollar risk is a variable percentage related to the portfolio value. In this case, I would be looking to see if adding additional positions (and potentially reducing the $ allocated to each position) would improve the results of the portfolio. At this point, it appears that there is a decreasing marginal gain from adding additional positions and the marginal gain is smaller when the positions are correlated.
Regards,
Eric
RE to Eric: very good comments. I’m hesitant though to use too much optimization in this particular report. I can already tell you that some of the more robust indicators (like the prop. OBOS) are going to dominate the portfolio and some of the fringe things like TED spread, etc. are going to get swallowed up. I’m intentionally doing the opposite here…presenting different viewpoints and all. Very good thoughts though.
Thanks,
michael
So if a cell in the Confidence column is green does this mean that particular model is predicting a gain for tomorrow? And is the percent the confidence interval for that prediction? I’m a bit confused…
Josh
RE to Josh: that’s correct, green (positive) means that model is predicting a gain, and red (negative) means predicting a loss. The percentage is not “confidence” in the mathematical sense of the term. It’s just the model’s confidence relative to a threshold that I determined…+/- 100% being something that has historically been VERY predictive, 0% being completely unpredictive, and everything else somewhere in between.
Geeky side notes: confidence has been both detrended (assumes a neutral trending market) and is normalized to account for volatility.
Hope that didn’t confuse things more…michael