[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Multiple simultaneous positions, Sharpe ratio



PureBytes Links

Trading Reference Links

I might be missing something here, but how can one
calculate the sharpe ratio from the equity curve
without knowing the correlations of the underlyings?

Best Regards,
Trey


--- Mark Johnson <janitor@xxxxxxxxxxxx> wrote:

> I happen to like the Sharpe ratio too.  I find that
> when I
> run a bunch of historical backtest simulations and
> measure
> lots of statistics on the resulting equity curves,
> the curves
> my eyeball and my gut like best, seem to correlate
> with
> the highest Sharpe ratio.  I've tried the standard
> set of
> candidates: Kestner K-Ratio, Sortino Ratio, Schwager
> Return
> Retracement Ratio, Seykota Lake Ratio, Spratling
> R-Squared,
> Curtis Faith R-Cubed, MAR ratio, Peter Martin Ulcer
> Index,
> Wealth Lab WL Score, ad nauseum.  But the equity
> curves
> I like, and the equity curves Sharpe Ratio likes
> (biggest
> ratio), are a better match than with any other
> metric.
> 
> However I would like to remind folks that to
> calculate a
> Sharpe Ratio, you need a time series of returns.  A
> list
> of trade outcomes is insufficient; you don't know
> the
> *timing* of those outcomes.  Different trades have
> different
> durations, and because we trade *portfolios* of
> instruments,
> many trades overlap one another.  For example, here
> are
> the positions in one of my personal accounts, as of
> Friday.
> (I'll be adding another one on Sunday night, as I
> just got
> an entry signal for LIFFE White Sugar #5).  There
> are 29
> different positions, nine of which got entered on a
> split
> fill and/or pyramid.  See Bean Oil and Corn for two
> examples.
> 
> These 29 trades, soon to be 30 trades, proceed in
> parallel.
> The daily returns of these positions add together to
> form
> the daily return of the portfolio as a whole.  We
> can't
> know the portfolio return (thus we can't calculate a
> Sharpe
> Ratio) without knowing the *timing* of the
> individual trades.
> In fact, the easiest and in my opinion best way to
> calculate the Sharpe Ratio, is to begin with the
> equity
> curve itself.
> 
> But you knew that already.  This is merely a
> reminder.
>    - MJ
> 
> Sym   EntryDate  L/S  ncars  Month  EntryPrice
> AD    20071210    L    xxx    Mar     0.87265
> BO    20071224    L    xxx    Mar    47.68
> BO    20071224    L    xxx    Mar    47.66
> C     20071126    L    xxx    Mar   409.00
> C     20071226    L    xxx    Mar   452.5
> CC    20071109    L    xxx    Mar  1951.0
> CD    20071210    L    xxx    Mar     0.99645
> CD    20071210    L    xxx    Mar     0.9964
> CGB   20071123    L    xxx    Mar   115.21
> CGB   20071204    L    xxx    Mar   116.03
> CU    20071210    L    xxx    Mar     1.4669
> DA    20071126    L    xxx    Mar    16.92
> DA    20071126    L    xxx    Mar    16.90
> DX    20071206    S    xxx    Mar    76.50
> EBM   20071126    L    xxx    Mar   109.50
> ED    20071107    L    xxx    Jun    95.79
> ED    20071126    L    xxx    Jun    96.16
> FC    20071210    S    xxx    Jan   105.85
> FF    20071001    L    xxx    Mar    95.73
> FF    20071001    L    xxx    Mar    95.725
> FSS   20071022    L    xxx    Jun    94.40
> FV    20071123    L    xxx    Mar   109^23.5
> GC    20071123    L    xxx    Feb   820.5
> KW    20071123    L    xxx    Mar   845.5
> LC    20071207    S    xxx    Feb    96.175
> LCC   20071228    L    xxx    Mar  1056.0
> LH    20071128    S    xxx    Feb    62.15
> MP    20071210    L    xxx    Mar     0.091775
> O     20071220    L    xxx    Mar   302.0
> PL    20071221    L    xxx    Apr  1527.8
> S     20071224    L    xxx    Mar  1193.50
> SB    20071228    L    xxx    Mar    11.06
> SJB   20071205    L    xxx    Mar   137.28
> SJB   20071205    L    xxx    Mar   137.26
> SM    20071224    L    xxx    Mar   335.00
> SM    20071224    L    xxx    Mar   335.25
> TU    20071123    L    xxx    Mar   105^03.5
> TY    20071123    L    xxx    Mar   112^23.75
> ------------------------------------------------
>    
> 
>