Greeting folks, any good ideas out there on how to best model the likely duration of a live trade. Thanks.
If I understand this correctly, you are trying to model how long a trade lasts? That would depend on you and your risk tolerance? Even a trending stock could suddenly, stop trending or go the other way. "A better approach is trade with the trend as long as it lasts." When the trend ends, that is when you exit. Modeling is useless as each stock moves differently, some more volatile than others. So, how are you going to measure that and use it profitably?
Duration vary on your trading style and market condition. Swing trading would have longer duration then say scalping so there is no definite answer for it.
You don't chose the 'trade duration'. You chose when to exit an opened position. And that's 'THE QUESTION'. When to exit is the most important question we should ask because we make or lose money only when we exit a position.
Hi thanks for the post. The ultimate aim is for risk management. So I have multiple positions in different assets classes, sometimes the same asset class and I need to check the correlation between the PnLs in-order to positions size new entries so as to keep within a risk limit. The best way to do that is simulate the future correlations of PnL and take the worst case values in the simulated range as the correlation value on which to assess overall 'portfolio' risk. That needs a rough idea of the time horizon for which the trades will be open simultaneously.
The time horizon you choose can be commensurate with the period of correlation you are measuring. That corr within a time interval is what you are trying to capture/avoid. So, it is a signal. However, it would be good to have an idea if your main signal for taking a trade has a time component. Let's say that your main signal is a breakout. How long do you think the signal is good? Indefinitely? So, there may be multiple factors that will affect your time in trade.
It is trivial with a large enough back test using historical data. Assuming your rules can be tested. If you are mostly discretionary, i'm not sure how you can model that. The are old stories about George Soros who might of exited a position if he felt his back was hurting.
Hi, thanks for the post. Yes you're right. Most of my signals are shorter term in the H1 timeframe. I think the signal decays in predictive quality the further away the target SL and TPs are from the current price, unless the signals are in a higher timeframe on which you then enter in a lower one. So far the, the very crude measure of trade duration I have done shows a uniform distribution of trade open times. But then there must be factors that affect the duration like volatility etc...and also the 1(tp):2(sl) ratio would be a factor. I was thinking some kind of monte-carlo simulation based on the price now, sl and tp and then using that to find the higher ranges of trade open times maybe. I'm not sure and it's something that I will start working on in a few weeks so i'm canvassing for ideas/know-how to save me time in the future.
Consider target minus entry / ( ATR/2) as a possible way if you don't chase for momentum in your positions. Go from ATR/2 as basic, ATR/2,2 as "slow", ATR/1,8 as fast and so on.