36. The law of relativity and performance of a system per time frame

Traders are used to the idea of the law of relativity (see post 9) as applied at the front office, in the sense that the claim that the market is going up or down is entirely depending on the time frame of observation. Different time frames give different results.

What they are not so much aware is that the law of relativity can apply also to the backoffice. This means that the same system has different performance, and risk metrics at different time-frames, even if it is run at a single front-office time frame. E.g. a trading system run in quarters (M15), when the results are granulated and presented in say quarters again M15, or hours H1, or days D1 have completely different results. And two different systems S1, S2 run e.g. in M15 may have almost identical performance if the presentation is granulated at quarters again m15, but completly different if granulated at D1. Therefore any specification or requirement to find a system with such and such a performance should state it explicitly in what back-office time-frame is the measurement. In fact a really detailed specification of a trading system performance (for design purposes of the system) should state the performance in multi-time frame way.