Hello, M. Gozzi,
Thank You so much for your swift and straight to the point answer on my query.
That’s more or less what I was suspecting after I played around with ProRealTime for a couple of days : since the backtests are so to speak processed in silo mode, it’s not possible to test a strategy over a portfolio as a whole…
By ways of extension, this means that any strategy test with an instrument has to be treated as if it was performed over a separate trading account with no possibility to mutualize neither the gains, nor the losses, across the portfolio.
Conclusion : the many “little separate trading accounts” may well run belly up very quickly if the risk level acceptable to the whole portfolio is applied to them individually, without having the possibility to be backed up by the full portfolio equity (0.5% out of $100,000 represent 15,00% out of $3,333 : I have not heard yet of any stand-alone strategy capable of supporting such a risk burden very long !).
On the opposite, if one wants to go this way, one can apply 0.5% of risk to each “little separate trading account”.
Let’s do the math again : we then find out that 0.5% out of $3,333 represent (rounded) 0.017% out of $100,000, which is quite acceptable to the full portfolio (to say the least !).
… But with such a low level of commitment, I doubt that one will book many profits, except if the objective is to become rich after 1 century or 2 if one’s lucky, and if the R/R Ratio of the whole strategy is actually greater than 2 over the long haul (in fact, 3 or higher : a common misconception is that a 1:2 R/R Ratio is fine to net good money on the markets. Bad news is that a 1:2 R/R Ratio is most of the time barely enough to cover the trading costs only : slippage, swap, commissions, taxes and miscellaneous account maintenance fees inclusive, and to buy a coffee and a donut at the end of the day…).
Conversely, losses will feel like ant’s bites to the scale of the whole capital. Only good point, I’m afraid !
All this boils down to say that, in my humble opinion, backtesting whatever strategy in the ProRealTime environment is sheer waste of time (especially if one considers the amount of time it takes to code properly any system a bit tricky), since by no means will the outcomes be a reflection of what real-life trading might throw at the aspiring trader.
Additionally, and apart from the specific point addressed in this topic, I have noticed that with ProRealTime, it is possible to trade contracts only to the level of integer numbers, 1 being the minimum size in backtesting.
This is a major showstopper for those like me who favor volatility-based stops (ATR-linked or Standard Deviation-linked stops for instance) over fixed position sizes.
Such a method, flexible and fit to the always changing market conditions, gives position sizes to the level of decimals when some formulas like PositionSize = (Equity * MaxRisk) / (StopLossInPoints * PointValue) are applied to tune the size of the position for every trade.
Only exception is when the “round” function is used in case of trades on the crude oil and the likes, which require the positions to be integers.
Well… Enough whinnying for now !
The bottom line is that, for all those reasons and some others more (Ah ! These backtests that are running in the tested instrument’s currency only, and within which it’s not possible to set the Initial Capital parameter in USD to then ensure that the backtest of the whole portfolio is cohesive ! And what to say about the swap parameter, which is one of the variables to be taken into account absolutely in any well-planned strategy ? This one is nowhere to be found within the ProBacktest environment as I have studied it so far), I think that I won’t continue with ProRealTime much longer…
Maybe I’ll give a try to the competition… Or simply go back to my old mule (MT4, not to name it, is full of flaws and bugging nick-nacks, but I feel at home in it) !
Whatever, I wish All The Best to you, and have safe and good time in the markets,
Kind Regards,
MF
.
