Answer to Lunchtime Quiz
I think the “BigBucks” system is prototypical of the types of systems that vendors sell for a lot of money to unsuspecting traders. This is a shame, because even if the rules are disclosed (which they are usually not), it is very hard to tell the difference between a real system and a piece of garbage.
In this case the first flaw is that the addition of the rising 10-day slope rule, generates too many trades, and for a low returning trend system, this is just not a surmountable hurdle. The second is that the system just doesn’t work well on other indexes–performance is poor on the Nasdaq and the Russell. This is especially troubling because the latter two indices as a general rule trend much better than the S&P500 and should respond well in this case. The third and fatal flaw is that in testing different rising slope parameters from 1 day through 50 days, the 10-day slope is the very best, while almost none of the other parameters produce returns anywhere near the same level as the original system. Sadly, many of the other slope parameters result in negative returns. This is a sign of a poor system right of the bat—–it is not robust. A good system can work with multiple indexes or stocks and will still work as you vary the parameter settings.
In this case, removing the 10 day rising slope rule actually produces a robust system. The rising 252 day average works well (roughly 6% annualized for long only, 10% for long/short), and produces solid returns on all the indices as well as most S&P500 stocks. Furthermore, it is a fairly slow turnover system. If we tweak the settings, the performance doesn’t change too much.
To conclude, i would suggest that you dispel the notion of “systems” and think more about underlying concepts. The market is either in one of three states: 1) mean-reverting 2) trending or 3) random. First you should focus on strategies, and this means adjusting to the behaviour of the current market conditions. Mean reversion is the most profitable and consistent strategy over the last 10 years. Trend following still works and has always worked, but it requires more patience and a tolerance for drawdowns. Of course, the market can become random, and if it does, all bets are off. Whatever strategy you decide to use, keep your “system” simple and robust, and don’t go crazy trying to combine too many indicators or strategies at one time. Its sort of like good cooking–the best chefs focus on technique (concepts) and tend to use very few ingredients (strategies)–but they work hard at finding the best ingredients (strategies). They add spices (unique indicators or combinations of indicators) very sparingly to improve and differentiate their end product. In contrast, rookie chefs are always looking for the magical spice, or are looking to add 20-30 ingredients to their recipe. Much like rookie traders, they end up with a muddled mess.
Note: due to business commitments i will put off presenting the DVI until tommorow.