Wednesday 14 April 2010

The Computers are Smokin'

It was great that there was so much interest in auto trading and Flo. It looks like there are some of you, like bakrob99 who are already going down that road. By the way, bakrob99, are you fully auto or just to enter and then manage?

A couple of you commented about slippage and not getting orders filled. I have said that using limit orders I have taken care of slippage but where we are vulnerable is when the range bars don't overlap and we miss the trade. This will have an impact on profitability, as we catch all the losers but miss some winners. Our experience so far running live in demo mode is that we make about 66% of the theoretical profits. I know this as we are running the same logic over eSignal data and live broker SIM. Any of you care to share your experience?

This project has a number of challenges.
  1. Getting historical data that is accurate. I am part of a private group that enabled me to get good data with volume attached in MultiCharts format. I can't provide data under my agreement with them.
  2. Ending up with statistics that show every month profitable over at least a 12 month period. This comes about by selecting the right range bar, stops and targets. There is no change in logic for the different markets.
  3. Achieving a reasonable maximum draw down. This is an important metric as it reveals whether there is a period when the model is losing money. When you eyeball the charts you get a handle on why the losses occurred.
Once we have Flo working in real time with real money, I'll provide a way for people to see it in action live. I'll also then start on the other markets we want to automate. So, as a portfolio, we 'll trade futures for Euro FX, Bund, ES, Dow Euro 50, mini Light Crude as a start.

12 comments:

  1. Hello Boss,

    What is the risk/reward (target) you are using at the 66%?

    Thanks,

    Rino

    Ps Spending time with Euro FX (6E) I find it even more spiky then the TF. (Russell)

    ReplyDelete
  2. Hi Rino. We're keeping the stops far away - about 5 times the profit target - its a drop dead stop. We're using the logic to exit and that's another thing we are monitoring in demo mode to see what the actual risk metrics are.

    ReplyDelete
  3. Hi
    First , thanks for MC & MD indicators & set up chart.
    Could you explain better or a link where you explain your trading rules.
    thanks

    ReplyDelete
  4. Hi Tom and Kiki: I have been reading your blog since close to the beginning but this is my first comment. I am very immersed in trying to create a trading system that is based upon all that I read here.

    I have been working to automate this system system (actually it is currently just a signalling system at present) and it is showing reasonable results in my backtesting so far. I am running Investor/RT which uses the same RTL language as Market Delta.

    However, I am having a problem which Kiki and others on this blog seem to have overcome. The problem is that when I include logic that uses the Volume Breakdown indicator within my RTL, then the Investor/RT needs to recalculate all of my RTL at each incoming tick. Because my RTL logic is fairly complicated, running it at every tick is bringing my computer to its knees as I try to use my automated signals in real time.

    Kiki, did you had similar issues and if so how did you resolve them? Did any other people on this board have similar issues that they managed to overcome.

    I would really appreciate any advice that others may have in this regard. In turn, I am happy to provide my advice to anyone that needs it.

    Thanks for an excellent blog Tom and Kiki.

    ReplyDelete
  5. Shrike, it's all in the blog. Look ast the charts.
    Tartan, if you want to email us your code I'll get Kiki to look at it to see if it's anything obvious. Send a chart definition of the chart you run it on too. MD and IRT are the same except for MD's looking inside the bar stuff. VBs are the same.

    ReplyDelete
  6. Tom,
    My strat enters and exits without my intervention. It does much better than I typically do at exits ... although I often have a better entry. My drop dead stop is 13 ticks which has been rarely hit (3x since January).

    MM rules:
    at +5 ticks move stop to -3
    at +8 ticks move stop to BE
    Exits:
    1st target min: 12 ticks.
    Trailing stop exits use a variety of methods to exit but mostly do not occur until the strat is +20 ticks.
    Max: 2 trades per day.

    Pretty standard stuff.

    Works on EURO too but I have a different and simplere strat for the EC which gives a better RR, so I don't trade the Range Breakout Strat yet on other markets.

    ReplyDelete
  7. Oh ... one other thing. My historical testing showed that missing trades by waiting for the retrace to the close of the entry bar was a losing proposition. In fact, if the strat detects that there was no retracement to the close (entry point), it is profitable to extend targets and/or add a contract at the close of the next bar.

    Another option is to intitiate the trade entry with fewer contracts and add on a pullback at --4 ticks. Leave the stop alone. This improves profitability on exits and reduces average loss size. Bigger breakout trades that don't pullback can be added as described above.

    ReplyDelete
  8. Tartan, I have been thinking about the issue you bring up relating to VB recalcs and I am pondering some options to alleviate the issue. I should have something done on that front soon.

    ReplyDelete
  9. LS_Chad, In case others on this blog don't know - it is worth saying that you work for Investor/RT. Thanks for looking into options to minimize this issue.

    By the way, Kiki recommended that I nest my code with if/then/else statements so that all other conditions are processed before VB conditions are assessed.

    ReplyDelete
  10. Bakrob, I'm working on an EL type strategy for the Euro at the European open, preferably automated (eventually).

    You mentioned you had a simpler strat for the Euro - any hints on that?

    ReplyDelete
  11. Kevin - MY Euro Strat does use range bars 9Tick and does enter using a Buy Stop order 1 Tick Outside of the close of a bar. The difference is my entry criteria is quite a bit simpler and is based on a MACD pullback when the MACD histogram goes below zero while the signal lines are still above (for longs). Take the next MACD crossover backup if it iccurs within a few bars.

    Basiclaly it is a pullback in a trending situation. The key for this strat is Money Management, My target is 38 ticks firm. My stop averages 12 ticks and I do not move my stop until 50% of the profit has been achieved.

    Avg Winner is 2x average loser and win rate is around 70%... so it is doing very well. Key is - I only tale 1 trade per day and only if it occurs within an 80 minute window which happens about 3 days of the week.

    If you want to work on the project or share ideas email me.

    ReplyDelete
  12. Important new information regarding Volume Breakdown recalculation efficiency in the most recent beta versions of Investor/RT and Market Delta: http://www.screencast.com/t/MzdkMzY3

    ReplyDelete