We started the discussion with a relatively textbook proposal to assess the trading profile and characteristics of a hypothetical system. Key data was provided to enable the relatively mechanistic process of churning out a number of trading related performance measures through well known trading formula.
There was a suggestion that breakeven (BE) trades and commissions might be a cause for the frequent malaise of a system’s hypothetical potential never being quite fully manifested. While it’s very true that there is significant underestimation (or even complete disregard) among newer traders to take the operational costs of trading into account, do these factors significantly impact the system’s performance values?
Calculation of drawdown and statistically consecutive losses was something that not many traders appear to do. An intuitive grasp of the likely magnitude of these data appear largely sufficient as an operational basis. Are they? And at what point does intuition become a bankable asset? If ever. And how does one recognise this event?
The myopia that often afflicts those married to their trading formula rather than the reality of trading itself was highlighted by the lack by most to spot the disparity between the stated R:R for the system and the Aw and Al given. We look at the data, examine the formula, and fail to think it through on an elemental level.
Assumptions were made by some to come up with a ‘reason’ for this shortfall and disparity between estimated performance profile and the performance attained in reality. Use of trailing stops was a suspect. However, given you can as easily use a trailing stop to override a target that’s being exceeded as reduce your risk if your position moves against before hitting the initially calculated stop, is this necessarily valid? Which case it more likely to prevail?
Random clumps of consecutive losses were recognised as more likely than the statistically well-behaved estimation of worst case scenarios. Making both drawdown and risk size somewhat less of a science than we should be comfortable with. So what can we use to calculate optimum risk size and likely drawdown?
The psychology of traders when faced with consecutive losses (and wins for that matter) was referred to. This will get a great deal more development in this thread.
We came up with another formula stating after X trades of a system with any known profile, you could have expected N consecutive losers. Big deal. Did you get this number? More? Less? What does that tell you.
For those reading between the lines my comments regarding the fact that the performance of any system changes as a function of each trade and the non-relaity issues relating to standard formulaic functional treatment of trading performance criteria, coupled with the fact that the people more likely to use these formulas are new-ish to trading and therefore, unlikely to have much of a string of trades to call real data, are likely to have already half-formed (or better) an view on where I’m likely heading with al of this.
The deviation your Aw and Al massively impact the effectiveness of ANY statisitical study of your data. You can either grin and bear it (most do) or pretend to take a sensible approach and filter out the outliers (LOL1 – Some do, honest!).
It was hinted at that there may be some non-Gaussian techniques that will yield performance calibration that more closely resembles reality and indication that we could be considering the use of higher-level market (integral analysis) data to help set a more reasonable context for our calculations.
There was (I hope) a suggestion to consider the trader’s own expectations of their system was actually a factor that (a) impacted upon the performance of that system and (b) therefore impacted their P&L for that system.
Another trader suggested that we sometimes don’t take trades we should, because they don’t ‘feel’ right. We’ll get into this ‘feel’ a great deal, but for now, is the trader who doesn’t take a trade which meets his or her criteria and better/worse than a trader who takes a trade that doesn’t meet their criteria? Does the outcome of that trade that was taken that shouldn’t have been (or the one that should have been taken bust wasn’t) make any difference to your answer to the previous question?
Yet another trader suggests a metric of about 20% discretionary (is that the same as intuition). How does that sit with other consistently profitable traders? Anyone else care to offer a metric for the split between system/discretionary? The mechanical/discretionary is a side issue for this thread as I’ll be developing the deeper issues of what constitutes successful discretionary approaches, but it would be interesting all the same to get some feedback on that one.