Backtesting exit points for discretionary trades

Discussion in 'Trading Software' started by Spearhead, Dec 24, 2010.

  1. Not sure if I'm describing it properly, but I am mainly a discretionary trader, but do look for certain technical setups as part of my decision-making process for entering a stock. I'm less decisive when it comes to exiting my positions. Is there software out there that will let me compute the optimal exit strategy (%change in price, time-based, etc.) if I manually enter all my trades in the past couple of years?

  2. Yeah i'm also looking for something like this. I figured I'd need a full customizable trade entry/exit software that runs through IB's api.
  3. backtesting exits on your discretionary trades is just one of many features. trades are overlayed on charts, and there are a myriad of helpful stats - time of day, by volume, MAE/MFE, etc. $50/mo. Manual uploding is not necessary as IB and many others supported for uploading trades.

    (no affiliation, just a client)
  4. ronblack


    The optimal exit strategy for any arbitrary setup in the case of loss is based on the percentage of account at stake.

    The optimal startegy for exit with profit depends on particular market conditions, like s&r and fibo levels, specific patterns, etc. and I don't see at all how backtesting can help you with that.
  5. I think my rationale for entering the trade is good. However, I exit the trades in a discretionary manner right now. I'd like to compute the optimal hold-time or target percentage that would yield the most expected profit (assuming of course that the strategy I use is profitable). I have a couple hundred trades over the past year that fit this category and I'd simply like to see if there's a simple adjustment I can make to increase profits. Since the trades are news-based, I wouldn't hold anything for more than the amount of time it takes for the news to play itself out, so I'd only want to backtest say, from an hour to maybe a few days after my entry.

    Thanks for the suggestion daf, I will try out stocktickr when I get the chance.

  6. tim888


    The first thing you should do IMO is to analyze the trades statistically to see if they are random or not. If they are random there is nothing you can do, do not waste your time. If they are not random you may have an edge and you can try to imrpove it. You can either do the analysis yourself or post the trade returns here, either absolute values or as a percentage of your bankroll and maybe somone will analyze them and tell you the story. Your sample is very small for many people. For me it is maybe fine I will have to see.

    Do you know your win rate, profit factor, etc.?

    Happy holidays to everyone.
  7. I've done my own analysis using my entry price and the closing prices on subsequent days and believe I have a profitable discretionary strategy. As for sample size.. I do realize it's on the small side, but using my actual performance as well as "dumb" analysis using closing prices, there seems to be enough of a profit % that it would take a pretty big statistical anomaly to make it unprofitable. However, I'd like to use intra-day data to figure out the minimum hold time and/or target profit %, and perhaps use a little more technical analysis to figure out a more optimal exit strategy. Since I'm trading the news, minimal hold times seemed better for reducing risk to other events.. so I've generally taken profit pretty fast. However, I'd be willing to hold my positions longer if the analysis backed it up. This could also lead to more automation, at least for exits.
  8. OP can you quantify your exits? What numbers/metrics/levles do you use to exit?

    Most simulations and optimizations are not real/true optimization. Consider 3000 data points (3000 trades) where you have a modest risk-reward and a 50/50 hit rate (meaning 1500 winners and 1500 losers out of 3000 trades).

    Most optimization engines simply remove the losing trades through statistics by identifying criteria that exist only during losing trades. You are not actually improving your trades, simply removing losers which increases the hit rate on your overall strategy.

    Actual optimization of a strategy (mechanically or via a system) is very difficult and borders on AI. What you are asking is to improve your actual trades versus the traditional "remove bad trades" from your strategy to improve the strategy.

    You will need to quantify the criteria for your entries and exits and then run infinate variable scenarios based on each set of criteria. The more criteria data you can identify the better... Be careful of scams from vendors selling "optimization"... Make sure that you don't share your strategy and make sure it isn't just going to simply remove losing trades.
  9. tim888


    It would be silly IMO to pay $2k when you can get Amibroker for only $200 and do all that plus a lot more.
    #10     Dec 26, 2010