Sunday, June 17, 2012

Introducing: JTrader Composite System

I'm excited to bring to you a new, very profitable (backtested, at least) trading system, which offers the best of both of my current systems. I'm calling it the JTrader Composite System which essentially merges my 2 systems into 1. Here is how it works... I think it's fairly intuitive.

1) If CCI is long and JTS is long, the new system is 100% long. (TNA or SPY doesn't matter, I will post stats for both...)

2) If CCI is short and JTS is short, the new system is 100% short.

3) If one system is long and the other is short, the system is 100% cash.

4) If one is long or short and the other system is cash, the system will be 50% long or short and 50% cash. So once again we are dealing with partial positions.

Ok, now let's look at some stats and profit charts. I've calculated TNA stats and SPY stats. Results are impressive, more so than either of my systems on their own. There are plenty of trades offered, which should satisfy the thirst of even the most active swing traders.

Before I get to the stats, I should mention the 2 caveats. 1) Stats only go back to Jan 2009...when TNA came into existence. 2) Stats may or may not benefit from over-optimization of the 2 underlying systems. There is no guarantee that future performance will equal past results.

SPY results first:
Compound Annual Growth Rate (CAGR): 37.10%

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

Average Trades per year: 75

Return breakdown (per year):
2009: 28%
2010: 46%
2011: 56%
2012: 5%

Profit Factor (per year):
2009: 1.87
2010: 2.78
2011: 2.58
2012: 1.40


SPY Profit Curve 2009-Present













TNA Time... As you can imagine, considering the results above are with a non-leveraged ETF, the results for a 3x leveraged fund are going to get a bit crazy.

Compound Annual Growth Rate (CAGR): 249%

Overall win rate: 180 winners, 82 losers (68.70%)

Average Trades per year: 75

Return breakdown (per year):
2009: 135%
2010: 270%
2011: 560%
2012: 34.8%

Profit Factor (per year):
2009: 1.58
2010: 2.67
2011: 2.39
2012: 1.90


TNA Profit Curve 2009-Present













As you can see, much like the CCI profit curves, leverage and compounding caused the results to get a bit out of hand and unrealistic. But it's fun to look at.

I still have to work on some of the logistics. First, I'll have to modify the webpage once again and also may need to rethink how I send updates (ie, just tweet the Composite system signal, or transmit the 2 underlying signals and the resulting Composite signal). Tomorrow (Monday 6.18.12) I will post all the dates and trades the system has taken in an excel spreadsheet for your download and perusal. Please feel free to post your thoughts and comments...thanks! And Happy Father's Day!

Update:
Average return (%) per Trade (SPY): +0.43% (Median: +0.40%)
Best Exit: +5.96%
Worst Exit: -6.29%

Average return (%) per Trade (TNA): +1.87% (Median: +1.80%)
Best Exit: +23.76%
Worst Exit: -24.72%

Update #2 (Link to Excel Spreadsheet w/ all Composite Signals...if this link is broken, email me at jtrader21 at gmail and I will send you the spreadsheet...thanks)

20 comments:

  1. First, thank you J. This is looking like some pretty impressive work.

    I have a question for you: are your results all based on backtesting and optimization, or is there any part that is the result of realtime trading?

    As you imply, only time will tell. If this system holds up, it will be a fabulous gift to your followers.

    ReplyDelete
    Replies
    1. 2011 and 2012 are as near to realtime as you'll get. 2009 and much of 2010 were used to optimize both the CCI and JTrader systems. This year and last were not used in the development of either. Good question...thanks!

      Delete
  2. JKH,

    I would like to ask you a question. Could you email me at sweenrose@yahoo.com?

    Thanks,
    Alex

    ReplyDelete
  3. JT

    Just Thanks

    (to echo "john")

    ReplyDelete
  4. This comment has been removed by the author.

    ReplyDelete
  5. "I still have to work on some of the logistics. First, I'll have to modify the webpage once again and also may need to rethink how I send updates (ie, just tweet the Composite system signal, or transmit the 2 underlying signals and the resulting Composite signal)."

    I might suggest a known shorthand (to fit in tweet/text requirements) to give the 2 underlying signals and the composite. That way those that are following one or the other or desire more information to make a decision on their positions can have the information.

    ReplyDelete
    Replies
    1. Thanks, it shouldn't be too bad until either or both of the underlying systems become conditional. If there's a lot of action at the close and the Jtrader system is flipping back and forth, I may just have to tweet "too close to call" and send a 2nd tweet after the bell with the official position. Those days should be rare, but it could happen.

      Delete
  6. Something else that would extremely informative would be a test of the system for the bear market of 2008 (using SPY, of course).

    And, again, thank you J.

    ReplyDelete
    Replies
    1. i would love to see it in the bear market using spy, please...this would be very informative and helpful

      Delete
    2. also, j, how do the stop losses with the jtrader system play into this? I get the feeling these were not computed in the backtest? I guess we go to either 75% or 25% if the need be?

      Delete
    3. I agree, that would be very informative but the original Jtrader system is heavily reliant on how TNA is performing. In hindsight I should've built the system based on the RUT and who knows someday I may attempt to fix it, but for now I'm just done. It is what it is, and what it is is pretty darn good. :) I'm confident it will do just fine in a bear. In fact I welcome the volatility. Fortunately, the CCI is a different story. It is not nearly as complex, and as a system it simply requires SPY data (daily high, low, & close) I have that back to early 2008, but I know I can get that farther back and it's on my to-do list, but not for at least a few months. I need a solid break from model development and crunching stats...

      Delete
  7. Please don't make this another squeeker...

    Please :D

    ReplyDelete
  8. J, just keep it simple... call it the "composite system", and just say what percent it is long/short. And, if there's a stop, state what it is (and if it's all or half).... That's all that matters.

    ReplyDelete
    Replies
    1. I don't agree, I think the choices (since they ARE separate systems) give us more options in deciding what we do with our money.

      Just my 2 cents...

      Delete
    2. Thanks, I'll try to be as informative and creative as possible in my tweets. Just be warned there may be days when a more definitive signal may come just after 4p ET.

      Delete
  9. Thanks, J... as for the new "combined system" stats, here are a few tidbits for ya:
    1.) If you are compounding returns, then you must use Logarithmic (not linear) vertical scaling on your profit curves. Straight lines on a log plot mean constant growth rate, which is what you want. That will solve your "hook" problem with TNA and you won't get the "hook" problem with anything when you go to 10+ years look-back.
    2.) Only do single-leverage analyses. Save yourself the trouble of 2x, 3, etc. Anyone can leverage up from there as they want by putting on more futures or higher leverage products. For comparison purposes, always use 1x leverage.
    3.) You really (really) need to go back 12 years or so. 2009-now have basically been only a bull mkt -- and one driven by unusual financial forces. Your analyses skips any bear market, and you should include at least two bull and two bear mkts - which puts you back to 2000. If you stick to just SPY (or SPX, for that matter), and you automate the system (e.g., in eSignal), you can back-test 20+years in a single click. And you need that, so people can appreciate how long the system can go through stagnation/draw-downs, and prepare themselves, accordingly.
    4.) Be careful calling 2011 "close to real time". We have LOTS of time to make "real time" trades for this (new) system. Real time (for this system) will start the day you call the first "combined system" trade.

    ReplyDelete
    Replies
    1. One more thing... the prior 3 years have averaged 44% return per year. That's doubling your money every ~1.6 to 2 years. Start with just $100k, and you will have a million in 7 years, and $11M in 13 years, and $102M in 19 years. Somewhere along the way, you will have serious liquidity problems getting in/out of SPY, that's for sure.
      The avg return for the 3x system runs into the same problems only much sooner. 322%/year means you take $100k and turn it into $100M in about 5 years. Again, serious liquidity problems in TNA at that point. You will then be 1/5 of the total $-flow of daily TNA.
      Seriously, you need to run this system back 12 years. Please do that on the 1x product. I think you will find some serious "burps" when you go out-of-sample w/this system. I suspect, unintentionally, it's been fit to the bull mkt profile of the past 3 years.

      Delete
    2. Believe me, I would love to see these systems more extensively backtested than anyone. See my response above on why that is impossible in its current form.

      I started doing this stuff for fun several years back using Excel (needless to say, I am now an excel expert!) Obviously excel has its limits - I've found most of them.

      If the lack of backtesting concerns anyone, that is understandable. If it prompts them to dismiss it altogether, that is their choice. If they like what they see here and want to incorporate it into their decision making, that is great! If they choose to follow me using whatever amount of capital with which they are comfortable, that is very exciting to me. If the lack of robust backtesting is that concerning, for the followers I suggest lowering risk. Whether that means trading 1x funds or allocating lower amounts of capital, they are free to do so. I am certainly not forcing anyone to follow me.

      Lord willing I will be on this planet for a long time. Since this is my hobby and I still find it quite enjoyable I will continue to provide these signals for a long time. As u alluded to, over time I will accumulate a large amount of data and signals. Eventually we'll be able to look back at the data and see how it performed during bulls and bears. Yes it will take time, but I have time on my side (i hope).

      I have Jtrader signals live back to 2010. The CCI system was optimized using 2008-2010 data which is why I stated that 2011 and 2012 were as close to realtime as I can provide. You are correct that realtime composite signals begin now, but technically it began when CCI signals went live back on May ~7 as the composite is a simple consolidation of the 2.

      I really only wanted to provide SPY stats/charts with this post b/c I agree the leveraged TNA stats and profit charts are downright silly, but you know people were going to ask... ;)

      I can only dream of having liquidity problems! LOL. All trades and stats are obviously hypothetical. I hope no one here believes they will become millionaires in 2 years trading $10,000 worth of TNA. :)

      Thank you for your comments, JKH. I appreciate your helpful and thoughtful participation.

      Delete