Saturday, December 15, 2012

Weekend study: 100% JT Comp strategy

Ok, this is only a few weeks late. :)

My research was to look at how JT Composite results would compare had one ignored partial signals and waited for 100% signals before placing a trade. Personally, I'm not sure I will ever follow this strategy due to the potential for lengthy stretches on the sidelines, but then again trading for trading's sake is not exactly a prudent strategy either. In order to continue trading partial 50% JT Comp signals, one must believe they still have an edge - and that edge is equal to or greater than the edge provided by the 100% strategy. That, or they may suffer from too little patience. :)

Note: For 2009-2011, I will use SPY for stats.  For 2012, I will switch over to TNA.  Don't ask why, it's just much easier for me! :-D  The conclusion is the same either way...

Original JT Comp stats borrowed from this post are in italics. The new 100% stats are bold.

Results 2009-2011:
Compound Annual Growth Rate (CAGR): 37.10%  ...39.98%

Overall win rate: 175 winners, 87 losers (66.79%) ... 87 winners, 22 losers (79.82%)

Average Trades per year: 75 ...36


Annual return breakdown per year (SPY):
2009: 28% ...38%
2010: 46% ...35%
2011: 56% ...46%

Profit Factor (per year, SPY):
2009: 1.87  ..3.42
2010: 2.78 ...4.46
2011: 2.58 ...4.18

2012 Stats (TNA) updated through 12/14/12
Annual return for 2012: 67% ...74%
Profit Factor: 1.81 ...2.35
Win Rate: 42 winners, 27 losers (61%) ...26 winners, 12 losers (68%)
Trades: 69 ...38

The conclusion that I draw is that following the 100% signals provides similar returns each year, yet reduces risk significantly.  The reduced risk is evident by the superior profit factor, fewer number of trades, and outstanding win-loss ratio.  A potential downside is you would be sitting around a lot twiddling your thumbs, carrying no position for long stretches. On the upside, the win rate is excellent. For SPY 2009-11 it was almost 80%. Hard to beat that... 

For 2012, up until a few weeks ago the annual returns were dead even.  The recent half-short positions have hurt the original JT Comp system, allowing the 100% strategy to move ahead up to 74% for the year.  Profit factor for both strategies is solid, but like previous years better for 100% signals. The one thing to keep in mind (besides the usual shortcoming that I don't have enough back-tested data) is that the 100% strategy assumes you go ALL-IN on the first signal.  This would add a certain amount of stress, and it's the reason I liked the partial positions to begin with.  Of course, you don't have to use all your capital like me following this system (I might question your sanity if you did...)   But it's a pretty good bet that when you see a 100% signal, you can probably go ahead and pull the trigger and/or get aggressive.

Thanks, and I hope this helps!
J

6 comments:

  1. J, thanks for the info and time spent preparing it. Is the SPY returns based on only 1X no leverage? Meaning TNA trades could be roughly 3X the SPY results?

    ReplyDelete
  2. Thanks a lot, J. This is really helpful.

    I must say that the all in model has some pretty amazing statistics, especially the profit factors and winning percent.

    This is certainly something to think about.

    ReplyDelete
  3. Tks for taking the time to do this analysis. I appreciate it.

    ReplyDelete
    Replies
    1. you should put this article in your 'New System information' page section so it's easy to find and refer to.

      Delete
  4. J, this is excellent information, and pretty much what I would have expected: the 100% system is a superior system in that it has a far better risk-adjusted return. And, to me, that's worth A LOT!... if you ever want to run a hedge fund, that's the number they'll want, too. Frankly, for the minor difference in annual gains, I don't mind the 'sitting around" for getting lower risk, higher reward and probability trades.

    If you could run the analysis back a decade, that'd be worth your time to do.

    ReplyDelete
  5. J, I know that it would be a lot of work to get data prior to 2009, and I also know that to ask you to do so much more work than you're already doing for this blog is terribly presumptuous, nevertheless, it would be of immense value. Many of the systems that I've been privy to show a substantial change in market dynamics around 2008. For example, may that worked before then stopped working at that point, and others started working well only around 2009.

    I have faith in your system. But that faith would be strengthened by backtesting before what appears to be a cusp in the market.

    ReplyDelete