sharpe ratio calculation

Discussion in 'Professional Trading' started by orangelam, Apr 17, 2016.

  1. I can't see your formulas. Maybe use google sheets rather than dropbox.

    GAT
     
  2. Maths is correct. Not that a Sharpe ratio from 2 months trading is very meaningful...

    GAT
     
  3. thanks a lot! yes, i think at least 1-2 years. do you know normally what sharpe ratio will be regarded as good? but i am not HF trading and it is discretionary with slow move and low margin so i don't know whether using SR is appropriate or not.
     
  4. tdazio

    tdazio

    Are you sure? Averaging daily log-return to get yearly return introduce error: it seems me need trasform numerator of SR formula to a linear geometric return..
     
  5. The result will be similar when the day return is small, for example account value change from 100 to 105,

    Log return= ln(105/100)=0.048

    Geo return = 0.05

    Based on my knowledge sharpe ratio def is from pricing, assuming asset price following log normal distribution so log return is preferred.
     
  6. tdazio

    tdazio

    Interesting question. But if price raises from 100 to 150 (typical elite-syle backtest profit :) log-ret error became intollerable
    To avoid error in presence of big returns (positive or negative), I agree with benjiamin m gross recipe here: http://quant.stackexchange.com/questions/7290/computing-the-sharpe-ratio.

    #1 calculate log return
    #2 calculate vola on log return
    #3 apply time scaling to vola (vola_annum = vola_day*sqrt[252] for daily)
    #4 convert log ret to annual: r_day*252
    #5 convert log to geometric ret: r-geo = exp(r-log)-1

    i.e. this is way PerformanceAnalytics R library compute annual return.