Add new comment

I have been working on finding strategies that work for real-time trading through an ATS (Automated Trading System). I did some initial tests using MACD, before moving on to implementing one of the systems in "The Original Turtle Trading Rules".

While the pdf describing the system used to be available in the public domain and easily downloadable from the net, it seems people are now charging for it. You might find it freely download other places it if you Google for turtlerules.pdf . If can probably buy it (from here) if you can not find it.

Anyway, the original turtles system (AFAIK) used daily charts. I wanted to implement and test the system using shorter timeframes, and eventually I implemented and tested it for various periods from 3 to 30 minutes. 

For backtesting strategies I obtained real time data from anfutures.com (see Datasources) and off I went. The data consists of one minute candle data, showing the contract date, timestamp, open, high, low, close and volume for each contract. There is little data prior to 2003-03, so I do not consider any results and/or data before this valid, even though the historical data from anfutures extend back to 2000. The Euro FX (EC) contract expires four times a year, so each contract is trader for three months.

The system I implemented is basically the "system 2" as described in turtlerules.pdf (a "Channel Breakout system taught by Richard Donchian" using a 55-day breakout, where I used a 55-candle breakout instead with varying candle lengths). Since I have no plans on doing the backtesting or trading manually, I needed some additional rules to make things sane:

• Only consider entering the market if the EMA(5) of the volume is 20 contracts per minute or more
• Always exit at the end of the trading day (16:00 hours)
• Exit if liquidity (measured by the EMA(5) of volume) falls below 10 contracts per minute

I backtested the system with an imaginary $100 000 account which is reset for each contract period (so there is no accumulation/losses across expiring contracts - all gains/losses are within each contract period).

One first observation; some of the gains are staggering, but so is the volatility. High drawdowns occur from time to time. The shorter timeframes (3 and 5 minutes, the red and green lines) really hit the jackpot on a couple of occations. One needs to keep in mind that the turtles system takes into account the overall account size when trading, so if it gets more money each bet will become bigger.

Volatility (not surprisingly) increases with shorter timeframe, and the 3-minute (red) interval shows this clearly. It hits quite a few jackpots with multple 10+ baggers during the tested timeframe, but scaringly enough it also has a significant drawdown the last two periods tested. Although it does not swing as far down as up, keep in mind each period starts with 100 000, so if it gets close to the -100 000 area it means EVERYTHING is lost (the maximum drawdown for the 3-minute interval is just below 67%).  Looking at the backtester output for the 3-minute interval might be interesting: 

0003 profit      +0.00   +0.00% trades   0 ppt    +0.00 w/l 0.00
0006 profit      +0.00   +0.00% trades   0 ppt    +0.00 w/l 0.00
0009 profit      +0.00   +0.00% trades   0 ppt    +0.00 w/l 0.00
0012 profit      +0.00   +0.00% trades   0 ppt    +0.00 w/l 0.00
0103 profit   -1987.50   -1.99% trades  14 ppt  -141.96 w/l 0.50
0106 profit   -4825.00   -4.82% trades  11 ppt  -438.64 w/l 0.20
0109 profit  +81000.00  +81.00% trades  60 ppt +1350.00 w/l 0.69
0112 profit  +19962.50  +19.96% trades 214 ppt   +93.28 w/l 0.47
0203 profit   -8375.00   -8.37% trades  47 ppt  -178.19 w/l 0.18
0206 profit   -3600.00   -3.60% trades 126 ppt   -28.57 w/l 0.32
0209 profit   -7962.50   -7.96% trades 370 ppt   -21.52 w/l 0.32
0212 profit  +14525.00  +14.53% trades 179 ppt   +81.15 w/l 0.48
0303 profit  -51575.00  -51.58% trades 313 ppt  -164.78 w/l 0.31
0306 profit +135187.50 +135.19% trades 605 ppt  +223.45 w/l 0.32
0309 profit +247487.50 +247.49% trades 687 ppt  +360.24 w/l 0.52
0312 profit +1634575.00 +1634.58% trades 736 ppt +2220.89 w/l 0.40
0403 profit +1028587.50 +1028.59% trades 815 ppt +1262.07 w/l 0.41
0406 profit  -52425.00  -52.43% trades 905 ppt   -57.93 w/l 0.35
0409 profit +1476850.00 +1476.85% trades 844 ppt +1749.82 w/l 0.48
0412 profit  +99775.00  +99.78% trades 969 ppt  +102.97 w/l 0.42
0503 profit +618987.50 +618.99% trades 1045 ppt  +592.33 w/l 0.51
0506 profit +1179175.00 +1179.17% trades 1127 ppt +1046.30 w/l 0.47
0509 profit +119400.00 +119.40% trades 1127 ppt  +105.94 w/l 0.44
0512 profit  -65712.50  -65.71% trades 1109 ppt   -59.25 w/l 0.42
0603 profit  -59875.00  -59.88% trades 955 ppt   -62.70 w/l 0.43

It starts off rough, loosing half your stack. Then it gets a lot better, until 0406 when it looses half again, then some more multi-baggers until it looses more than half both for 0512 and 0603.  Fortunately, moving onto larger timescales improves things at least volatility-wise.

Here is the same output from the 10-minute trading interval:

0003 profit      +0.00   +0.00% trades   0 ppt    +0.00 w/l 0.00
0006 profit      +0.00   +0.00% trades   0 ppt    +0.00 w/l 0.00
0009 profit      +0.00   +0.00% trades   0 ppt    +0.00 w/l 0.00
0012 profit      +0.00   +0.00% trades   0 ppt    +0.00 w/l 0.00
0103 profit      +0.00   +0.00% trades   0 ppt    +0.00 w/l 0.00
0106 profit    -500.00   -0.50% trades   1 ppt  -500.00 w/l 0.00
0109 profit  +22900.00  +22.90% trades  28 ppt  +817.86 w/l 0.75
0112 profit  +17775.00  +17.77% trades  57 ppt  +311.84 w/l 0.75
0203 profit   +3887.50   +3.89% trades  33 ppt  +117.80 w/l 0.54
0206 profit  -26537.50  -26.54% trades  64 ppt  -414.65 w/l 0.48
0209 profit  -10462.50  -10.46% trades 124 ppt   -84.37 w/l 0.46
0212 profit   -4162.50   -4.16% trades  89 ppt   -46.77 w/l 0.52
0303 profit  +36862.50  +36.86% trades 116 ppt  +317.78 w/l 0.39
0306 profit  +64300.00  +64.30% trades 179 ppt  +359.22 w/l 0.50
0309 profit +212012.50 +212.01% trades 183 ppt +1158.54 w/l 0.78
0312 profit +136100.00 +136.10% trades 245 ppt  +555.51 w/l 0.47
0403 profit +265625.00 +265.62% trades 299 ppt  +888.38 w/l 0.41
0406 profit  -58675.00  -58.68% trades 306 ppt  -191.75 w/l 0.35
0409 profit +146500.00 +146.50% trades 317 ppt  +462.15 w/l 0.28
0412 profit  +69650.00  +69.65% trades 341 ppt  +204.25 w/l 0.44
0503 profit  +77300.00  +77.30% trades 320 ppt  +241.56 w/l 0.30
0506 profit   -1700.00   -1.70% trades 335 ppt    -5.07 w/l 0.39
0509 profit +341325.00 +341.33% trades 328 ppt +1040.63 w/l 0.44
0512 profit  +60087.50  +60.09% trades 321 ppt  +187.19 w/l 0.38
0603 profit   -5075.00   -5.08% trades 326 ppt   -15.57 w/l 0.34

Unless you have literally no risk-aversion, this probably looks better, even though you would still have lost more than half your chips on 0406.

The turtle trading system has an addon rule which says you should add on units (up to four) to your position when it moves in the right direction, and this probably contributes significantly to the volatility, like the graph for the 10-day interval graph with and without addons below illustrates:

Is this system tradeable? As always, that depends on your risk tolerance. If you like to gamble, there seems to be potential profits, but with significant risk. Do not bet your whole stake on one contract period. Without the addons, the volatility becomes less, but the X-baggers become fewer as well.  YMMV.