Sunday, June 28, 2009

Black swan identified


At long last, traders may now identify a black swan event using charting analysis.

Wednesday, June 10, 2009

Yummy soybean meal


Daily chart of soybean meal going back to 1969 with some exponential lags laid over top. This is a backwards-adjusted continuous contract and notice how the price goes negative. This is creating some ugly glitches in my system that I can't seem to figure out yet. Thus, I take a break to post this.

This is the first time I've ever actually looked at the chart for a price series I have tested. I need to program some chart functionality into my system for sure.

On another note, thinking about the drawdowns incurred from trading this crossover system on soybean meal evokes many feelings, including intense fear and elation.

Yikes!

Also, here's a screenshot of all the markets I have purchased. Ready to do some testing, but acquiring data has raised some other issues first, including accounting for foreign exchange when calculating PnL in foreign futures markets. Still this is a big step.


Back to work. Cya!

Monday, June 8, 2009

D-D-Data!

Finally bought some data today! 70+ markets going back as far as 1969. $99. It was WAY too cheap it seemed. I also wrote a data scanner and an initial screening of the data highlighted only one anomaly. So far so good. The guys at the Trading Blox forum vouched for the provider, Pinnacle, as a low-cost, less frills option, which certainly works for me now. Norgate and CSI seem to be other healthy options that I will need to dip into to test data reliability.

Now to prepare the feast....!

Sunday, June 7, 2009

Drawdown curve



Coming along now in various facets in designing the backtesting software. This past week completed adding equity and margin limits to both systems as well as wrote a piece to run market correlation studies. I've included many more performance statistics, including robust statistics mentioned in Curtis Faith's Way of the Turtle, which I recommend for anyone designing systems. And, finally, I can now generate drawdown curves which is a nice complement to the equity curves I have been spitting out. Please enjoy as I do!

Moving ahead, I need to account for contract rollovers in my simulation as well as formally verify the Support and Resistance system output in Excel. I also am ready to buy some data -- finally! I will buy data from two vendors for about 70 markets, domestic and foreign. It runs between $100 for $250 from what I have seen, although I do not know how reliable these vendors' collections are. I'll have to write a data scanner to test the data for inconsistencies before I can employ them.

Also, here is what my system output is looking like at the moment:

Exponential Crossover System Backtest
Side = Both
Slow Lag = 325
Fast Lag = 85
ATR Lag = 20
ATR Multi = 5
WarmUp = 20

Starting Equity = 1000000
Heat = 0.1
Start date = 12/31/1974
End date = 9/27/2005
Ending Equity = 110550027.500
ICAGR = 0.1531
RAR = 0.1328
Max. Drawdown = 0.6037
Longest Drawdown = 91 mos.
Seykota Bliss = 0.2535
R-cubed = 0.1105
-----------------------------------
# of Trades = 29
Avg. Trade Length = 22.5 mos.
Max. Trade Length = 49 mos.
Win Percentage = 0.5517
Max. Losing Streak = 4
Avg. Trade = 3777893
Avg. Profit = 7170253
Avg. Loss = 397319
Avg. Profit / Avg. Loss = 18.05
Expectancy = 3777892.84
T-test = 11.87
Optimal F = 1305581
-----------------------------------
Max % of Account Employed = 1.0005
# Margin Calls = 14
# Forced-liquidations = 2
# Pared-backed Positions = 8
# Insufficent Funds for Entry = 0
Total commissions = 49365.00
************************************

Be well!!

Monday, June 1, 2009

Whiling away time or not?

Since I don't really want to get a "real job", I have been forgoing the job search in lieu of programming, which, truthfully, could be a lot of work for nuthin'. There are really robust software packages like Trading Blox, that, if I had a job, I could purchase and do a great deal of the testing I would need to do. But, instead, I am spending 6 1/2 hours a day adding functionality to two buttons because I am treading in new programming territory. I don't know if it's worth it.

So I have feeling a little lost. But all I want to do is program, build a system, get my capital, and trade the system. I really don't want to do anything else if I can help it. A Wharton MBA degree could help provide some cachet and a safety net if I blow out and I wouldn't mind having that. But I find it hard to believe that I will get in there, especially if I do not take a semi-prestigious job now.

Maybe if I just do what I want to do--what I REALLY want to do in pursuit of my goal--then everything will fall in place. Maybe the Trading Tribe posse are right when they say:

INTENTION = RESULTS

Sunday, May 17, 2009

Creating a monster


Just wanted to share an equity curve with y'all.

  • This is from an Exponential Crossover system, trading S&P 500 futures and Gold futures both long and short between 1975 and 2005.
  • Starting capital = $1,000,000
  • Ending equity = $124,036,093
  • Annual growth rate = 15.7%
  • Largest drawdown = 60.4%
  • Bliss (Growth rate/Lgst. drawdown) = .26
Looking forward to more testing, especially on more markets at once. Because, as of now, these system characteristics aren't too sexy, cf. the Bliss ratio. Looking for a Bliss ratio of 1.0 or better.

The end is the beginning

Last month and a half has found me forgoing short-term trading altogether and instead focusing on finding a salaried position and working on my long-term trend-following backtesting system, a project with code-name "Sunyata." Yes, I name my babies after myself.

I have finally acknowledged that short-term trading 1) does not accentuate my strengths and 2) does not appeal to my imagination: my strengths being careful intellectual analysis and creative plasticity and my true-love the "Long Pull" (or the "Short Push", whichever). This is fine with me. Self-discovery is fine with me.

Consequently, I have been putting some hours into Sunyata and have completed verification for the Support and Resistance code for multi-market testing as well as--and this I consider quite a personal triumph--coding a steamroll function that incorporates multi-market testing. So now I can push a single button and and the system will spit out every combination that I specify in a range, so that I can easily locate optimal parameters. Eureka!

Will keep you updated.

Peace!