@ Vonasi – Let me put forward my 1 pence (:-)) worth on the issue of using % based stops/limits. I actually prefer using them to straight points based ones for the following reasons :
- In reality every investment is ultimately assessed on a % return basis – ie if I told someone I made £100 profit last month and same again this month, they would assume that the trading performance was the same each month. But then if I told them I used £100 capital the first month but £10,000 the next to generate those profits, then the return performance is viewed much differently – 100% vs 1%
- Using absolute points levels can create issues in backtesting over the longer term especially when the underlying instruments’ value has fluctuated dramatically. For eg if you used a 60pt target level in the DAX, 10 yrs ago it would have been difficult to hit this level within a day as that would be a 1-2% range, whereas today 60-100 pt (0.5 – 0.8%) intraday moves in that index are very common. But if you actually look at the % moves in the index over the 10yrs rather than the daily points one (eg using a log chart) then it paints a very different volatility picture to a pure price chart. Therefore trying to optimise using points rather than % limits would create very different results that are less realistic because you are trying to “fit” one fixed number across a longer timeframe.
- Markets generally tend to move in % based increments because thats how investment managers, funds, traders, HFT algos etc view and therefore trade their returns (even margin requirements are stated in % terms). For eg. in equity indices the index move is determined by the underlying constituents’ % mkt cap moves (ie value) and not their absolute price moves (the Dow is an exception) so therefore it makes more sense to use this format when testing out strategies. In some markets it’s less relevant but even in parts of FX trading, correlation and carry trade strategies are always evaluated on a % basis.
So broadly speaking, if you are backtesting a strategy on a shorter term then it may be ok to use points rather than % as in that short a term, the instrument value and inherent volatility may be fairly similar over that period. But If you are looking at a longer period I would prefer to code %’s so as to perform a more robust backtest that could be more likely to accurately reflect future movements and performance.
This is just my humble opinion and a few thoughts for others who may find it useful, I’m by no means suggesting it’s the only way to go as it depends upon the strategy and your personal preference, I myself use some points based systems where relevant. In fact, in this case where we are limited by the short time range data set available, I would guess that optimising either on a % or points basis would probably make little difference to the overall results.
.
